I am a professional classical violinist, but I also have a business doing live concert recordings of classical ensembles. On occasion, I end up recording a concert on which I also perform. When that happens, I can't just sit out in the venue during the dress rehearsal to work on levels. The best I can hope for is a few moments at the beginning of rehearsal where the conductor will have the orchestra play one of the louder sections of the concert while I frantically set levels. After rehearsal, I would then go home, listen back and make notes about what I needed to do differently for the concert the following night. Still, I always had that "did I leave the oven on" type of feeling on stage wondering if I got the gain staging right. After all, the dress rehearsal doesn't account for the audience clapping or the performers getting extra amped up for the concert. This is especially true of vocalists, who rarely sing in full voice during the dress rehearsal. The obvious solution, of course, would be to not record concerts on which I also perform, but then I would make half as much! Then, 32-bit float entered the picture. I replaced my Zoom F8 with the F8n Pro and have not had that feeling since. Turn it on, hit record and the levels are perfect every time. This also comes very much in handy when I am not performing. I no longer have any fear when accepting recording gigs on short notice. I can walk into a venue, spend all of my pre-concert time assessing the room and deciding on a stereo mic technique and not waste a single second worrying about gain staging. As for the sound quality, I have recorded, mixed and mastered nearly 100 live classical performances and I cannot hear even the slightest decrease in fidelity when using 32-bit float. If you hear a noise floor, it's your mic, simple as that. And Beethoven's Symphony no 9 with 100 voices is not loud enough to clip the inputs when mic'd properly. Just stop with all of the caveats. If you're not recording Falcon Heavy launches, then you are going to be fine. If you are skeptical about taking the leap, think of it this way. When I first started doing concert recordings I used a 8U rack with a 24-channel mixer bolted to the top and recorded everything on a Roland VSR-880 hard disk recorder. All of that functionality, and immensely more, is now packed into my little F8n Pro. Now, do you think I have any nostalgia for the good old days of pushing 100 pounds of equipment all around town? Hell no! And I won't miss worrying about gain staging when I only have one chance to get the recording right. I am an enthusiastic convert to 32-bit float and I will never go back.
Rob, Thank you so much for sharing this real-world experience. Your story is a fantastic example of how 32-bit float can transform a workflow. The "did I leave the oven on" analogy is spot-on - and it's wonderful that you've found a way to eliminate that worry from your recordings. It's always inspiring to hear how new technologies can solve real problems for professionals. Your success with 32-bit float is a great testament to its potential, especially for those who might be hesitant to try it.
I had no idea anyone would get so worked up over 32-bit FP. I record almost everything at 32-bit FP because I work in film, specifically as a sound designer, and the dynamic range is awesome. Same with dialogue. When working in 24-bit I would do something similar to what Sound Devices does with their 32-bit FP; meaning I would record to two channels, with one channel properly gain staged and another about -10dB below that. Using 32-bit FP saves me setup time, and editing time. My deliverables are still almost always in 24-bit 48kHz because that's the current industry standard. But my preference is recording in 96kHz 32-bit FP then downscaling. I've never heard a client complain, and frankly most of them don't even know enough about sound to complain. I will add that I've clipped in 32-bit FP a few times when making sound effects, and had to adjust how I'm setting up mics.
Cool! Yeah, You can still clip the analog input side. I ran into that with the F6, but I haven’t been able to do that with the MixPre yet, even when close-miking drums.
Great explanation! It reminds me of the early days of digital stills photography (late 90s). As a co-chair of the Digital & Advances Imaging committee for the Professional Photographers of America we were introducing and teaching thousands of our members and non members about the new tech. Almost without exception, we heard “digital will never replace film”. As you said it was fear of the unknown new tech and workflow. Try to find a place to develop film now. As I transitioned into cinematography, the film and television industry went through the same “digital will never replace film” denials 10-12 years later. Some of it is protectionism and resistance to change as you said. Ten years from now only the die hard OGs will be clinging to their antiquated 24 bit recorders and looking for people to repair them.
I totally agree, Rob! I've used that exact analogy myself. It's a great way to illustrate how new technologies can overcome initial resistance and eventually become the standard. It sounds like you have firsthand experience with this kind of transition in both photography and cinematography. Your insight is incredibly valuable for anyone who might be hesitant to embrace 32-bit float recording. Thanks for sharing your perspective!
I don't use 32-bit float because people are now selling perfectly good 24-bit equipment for bargain prices. 16/24-bit was fine for decades for professional workflows. It's certainly good enough for my silly hobby projects for the next few years.
Absolutely. 32-bit float isn’t about better audio quality. It’s more about the changes it enables in workflow, reducing the likelihood for error and recovering sounds that might otherwise get lost. What’s amazing is that the equipment isn’t necessarily more expensive either.
Right. If you’re working at home and not on the clock, as I am these days, you can spare the time to get your gains tickety boo. But if I ever go back into a studio, do location work or record a band I can definitely see the benefit to workflow of deciding on a baseline level and using that for everything. No more having to decide exactly how much harder a drummer will hit when the red light comes on as compared to when you’re asking him or her to ‘hit as hard as you will in the song’ to get the best level! One of the great variables of life.
Hats off, man. Grest subject to cover. I agree, the experience I had with the F6 was a real gamechanger, albeit for field recording and location sound but now I want to try it for music purposes thanks to you! Subscribed.
I also work with composers n sound designers that are not audio geeks or industry boffins, to them a zoom 32 bit recorder is just an easy to use device with no gain knob. Jus saying not everyone who makes music is an audio nerd (like me!)
The argument I have made for years is my consoles in the studio have about 6 DBU (+26dbu) more headroom than the best 24/32 integer converters (which run at +20dbu) on the market so if I want to slam the mixbus or push a mic preamp I am going to overload the converter, the only option was a 32 bit floating point converter like we see today. The dual gain stage of the converter gives us the headroom needed. Sadly there are not any large 24 or 32 i/o interfaces for console users like myself. But I have been using a Mixpre for about a year now as my final mixdown rig. So When we do our final mixes the output of the console is captured to a separate rig running a Macbook Pro and a mixpre. But we really need a Thunderbolt interface with DB25s and around 32 i/o to really use in commercial studios!!
That's pretty interesting... I don't know if it would suit your needs., but the Sound Devices Scorpio has 32 channels. It doesn't have DB25s, but it does support 32 channels of I/O over Dante. I guess it doesn't support 32-bit float in the firmware yet, but when I talked to Zach at Sound Devices, he said the hardware supports it. They haven't implemented it in the firmware yet, largely because there is no demand from [professional] customers. I think this is due to the prevailing mentality that 32-bit float is for amateurs or is a fad.
@@simonpeck I looked at the Scorpio but it won’t work for our needs we need all line level inputs to use the direct outs on our 2 consoles..Secondly USB devices aren’t good enough for tracking simply due to the larger latency they have. We prefer thunderbolt which is now part of the open source USB4 standard so hopefully Sound devices will go that route with the next generation and make it a USB 4 device and it can work with whatever you have from USB 2.0 to a thunderbolt port. But I have said for a decade now that 32 bit floating point is the answer to finally catch up digital with analog recording devices like tape machines since like analog the loudness doesn’t matter. 32bit float is also better since is doesn’t have any distortion artifacts as well. But you will find in the audio world most people want to work like it is 1977 and only embrace technology when they are forced too. Cheers!
@@zagatoalfa What I mean by that is that there is no hard ceilings that produce undesirable errors in analog. So like I said previously. The best integer converters can only handle about +20 dbu and that is the very highest end of converters chips and not interfaces like Focusrite or Presonus which max out at +14 dbu. So a basic analog console can run at +26 dbu before distortion which is a bit higher, so this is why it has always been a challenge to work with most audio interfaces in a 24 bit environment. There is no room for mistakes and this can kill creativity when you are trying to achieve certain things. Lets take the example of pushing a Neve 1073 into saturation. This will overload a converter that is 24 bit or even 32 fix integer but a 32 bit floating point converter could actually handle the output at line level without issue and capture the output without errors or digital degradation. So loudness isn't something you think about in the analog domain as much as you would in digital because it analog more forgiving but once you add a fixed bit converter to the chain now you have a hard and fast ceiling of -0.1 dbfs before clipping and intersample error distortions. 32 bit floating point solves this issue of actual headroom above 0 dbfs during tracking. Hope that explains why I think 32 bit floating point converters are an important step forward. Cheers!
Thx for the explanation. I mainly use the SD Mixpre6ii for voice. The levels are set in pre with the limiters activated in 24/48kHz. Saved a lot of processing time in post and a lot of HD-space. 32-bits came in handy when the sound of thunder and rain was needed and when I was filming on a construction site. Those situations were unique and it was handy to be better save than sorry. At that moment a time saver for shure.
Great video, I shoot flying events, or just hang out at a local airport where anything from large Boeings to a Piper Cub can be out and about. The 32 bot float on a Mixpre-6 Mk.2 is fantastic!
The mixpre 32bit recorder has saved my butt more than once, I record live concerts and some times I don’t have time for a soundcheck, usually I can guess with good confidence, but sometimes it’s out of bounds! That 32 bit comes in reallllll handy in those occasions
I didn't realize there were only two practical levels these recorded at. The floating point format has 8 bits of exponent, so there is a LOT of unused range there. I guess they just wanted to use a format that was widely supported and well understood.
It's interesting that half a year after you posted this video, ZOOM released their essential series 32-bit recorders with NO gain controls. I jumped into 32-bit float (Tascam X8) because I'm a solo musician and record all my live sessions. I have enough to do without having to run a soundcheck and try to get optimum levels. 32-bit allows me to just hit record and not worry about if the recording is too loud or too soft. I was fine with 24-bit for years, but sometimes the levels weren't set well (I had to guess) and the recordings were way in the red, making them unusable. 32-bit gives me a usable recording as long as I don't overload the front end.
You can hear the distortion on the back end of transients for 32-bit float. Specifically zero in on the snare and hat sounds. They aren’t as crisp. This is a known issue with the dual ADC that was only recently patched by Sound Devices. I have not seen any sort of fix from Zoom yet.
This leads to a reason why many engineers are hesitant with 32-bit float. Even this video describes a signal chain with some sort of “magic” imposed on the signal. As it turns out, that magic was creating noticeable distortion outside the engineer’s control. Its great that SD has patched that issues but they’re still passing the signal into some mystical process the engineer can’t control.
Your killing me with that groove. You have demonstrated what everyone always says in these videos; what really matters is the performance' lol Great video man thanks!
I use four (4) Zoom F6 recorders time-synched together to do live rock band recordings in 32 bit floating point. It generally works perfectly. I have, however, identified the need to carry with me a few XLR in-line attenuators to deploy when connecting certain sources - most notably any signal coming from a direct output on a guitar amp. I have had to apply as much as 20dB attenuation in some cases. Easy enough to catch before the performance starts but devastating if you miss it.
I am just an unpaid hobbyist. I was curious about 32bit float, so picked up a Zoom F2, then a Tascam X8, then a Zoom F8n Pro, a Rode Wireless Pro set. I love that I dont have to be worried about levels while recording, which is important when you are a one-man show, being the cameraman, the director, the actor, the sound guy, the producer and the PA. The only problem still is that not all software support 32bit float yet and even some that do will not handle it 100% properly. For example audio files from the Zoom F2 usually come out at too low a level and Davinci Resolve cannot push the volume high enough. I either have to normalise it in some other software before loading into DR or do some other tricks to sort this situation. But it is less headache than when you have 24bit audio recorded at a wrong level.
@neomatrix888 I have not checked how DR19 does it, in previous versions there was a limit on how much you can increase the volume level, and this limit was way too low for files from my F2. Files from the X8 were usually fine as in txt reecorder, as well as the F8nPro you do set volume levele into the ballpark before recording. But with the F2, I sometimes either had to set the volume level in some external software, or use a trick: amplify the track volume to max, then copy it onto a second track and amplify that again. Maybe they have fixed it since.
I’m resistant because I don’t want to have to buy new converters! But seriously, when I need to get new ones I’ll think about it and I follow the conversation with interest. I don’t feel I have anything missing because I record 24 bit and I’ve only just switched up to 48kHz sample rate for the slightly less latency and, nowadays, greater compatibility. But the idea of never (or almost never) having to concern myself with levels going in is certainly appealing. Saying that, I can’t remember a time when I’ve maxed the converters when recording since the days of the Mitsubishi X-880s we used to have, which were 16 bit, I believe. But we were learning and still knee-jerk pushing the needles to keep away from tape noise, even though that was no longer a problem. Actually, now that I think of it, I was doing location sound as a favour for someone and I think we went over once or twice at the start. My inexperience in that field showing and I can certainly see how 32 float would be ideal for location sound or live music.
The use case I can imagine for 32bit is for when you need to record something that you don't have a chance to try out first to set your levels. If you're recording a jet flying by or a building blowing up, or a space shuttle launching, that might not be repeatable and you won't have a chance to check your levels.
But why not for everything? It’s the first time we have had available to us a bit depth that captures the entire dynamic range of audio. We went from 16 bit to 24 bit for some reason. I guess the industry just decided that that was enough.
@@simonpeck oh, for sure it would be great to have it just be the "final resolution". But if you're going to buy one piece of gear that has it, then field recorders is a good place to spend the money.
I love it when musicians talk about computer science. As usual the marketing department chose to misuse a technical term because it sounded cool not because it was accurate at all for what it was describing and thus "32 bit float" somehow became an "audiofile" buzzword...even though this is all about sampling and compression. The real issue is everything is optimized for MP3 and Mp4 and all the 32bit float formats are proprietary.
How do you feel when Computer Engineers talk about music? 😉 While I agree that the marketing around "32-bit float" is a bit misleading, it's important to note that 32-bit float is simply a way to store a single numerical value within a computer. It's a standard format, not a proprietary one, and it has nothing to do with sample rate or compression. Sample rate determines how often we capture a snapshot of the sound wave, while compression involves discarding data to reduce file size. These factors are independent of how the individual values are stored (whether as 32-bit float, 24-bit integer, etc.).
@@simonpeck As soon as you convert an analog wave to a digital format you've compressed it. You wouldn't be the first engineer I've had to explain how software and math works to. The very first step in compression is the hardware that samples the acoustic wave and converts it into a digital format. THIS is what has been optimized for the current software suite usually destined for MP3 compression. For example: I cant split sampling to take the top and bottom ranges in a sample then take the mid ranges in the next sample and stack them in into split frequency sampling on the same ADC. This could easily double the range of 24 bit to 48 bit and completely alleviate clipping but most hardware is incapable of sampling like that even though they can easily support the 4x sample rate that would obligate...because the hardware is optimized for antiquated file formats. Additionally: 32 bit float is tied to how binary systems store decimal places... that was impressive in the 90s for audio but even my shitty computer now has 32gigs of RAM; runs at 64 bit, and has high enough bandwidth to deal with 4k video. ..and .WAV files have been around forever and absolutely are proprietary to Microsoft. The reason we're still using 24 bit is because, again, everything is optimized for MP3--and MP3 doesn't store decimal places.
@@TurboLoveTrain You're absolutely right that the first step in any digital recording is converting the analog signal to a digital one. However, this initial conversion doesn't have to involve lossy compression. If the sample rate is high enough (above the Nyquist frequency), the original waveform can be perfectly reconstructed. That's why we use sample rates like 48 kHz for standard audio and even higher rates for specialized applications. While I appreciate your technical insights, I'm a bit confused by the alternative sampling technique you're proposing. Could you elaborate on what advantages this approach would offer over traditional sampling methods or 32-bit float recording? It's worth noting that most modern DAWs use 32-bit integer or even 64-bit float internally to process audio, providing ample headroom and dynamic range. As you mentioned, modern computers are certainly capable of handling high sample rates and large amounts of data. This makes 32-bit float recording even more appealing, as it eliminates the need to set gain during recording and ensures that no audio data is lost due to clipping. This can be a huge benefit for many recording scenarios, especially those involving unpredictable sound levels. Regarding the continued use of 24-bit, while MP3s are indeed widely used, it's not the sole reason the industry continues to rely on 24-bit. Many professionals work with lossless formats that can fully utilize the dynamic range of 24-bit recordings. Additionally, factors like established workflows, compatibility with existing equipment, and (surprisingly) concerns about storage space also contribute to the continued use of 24-bit. Interestingly, even high-end recorders like the Sound Devices 8-Series and Scorpio are technically capable of 32-bit float recording, but the feature hasn't been enabled in firmware yet due to some pushback from the professional audio community. This suggests that the preference for 24-bit is often more about tradition and established workflows than technical limitations.
@@simonpeck Lossless is also a marketing term. The act of sampling an acoustic wave into a digital format is one of the definitions of compression. Lossless refers to the fidelity of the digital file during type conversion--it does NOT refer to ADC input sampling of an analog wave. The file must already be in a digital format before "lossless" or "lossy" even comes in to play. As for the sampling method I was alluding to it is actually based an old version of how to trick a computer display into mimicking higher resolutions while minimizing system load... I used it as one example of the significant limitations of the current, traditional, audio input pipeline. Take an input stream and break it into two sampling steps. One sample step records only the upper and lower 1/4 of your amplitude range. The second sampling step records the middle 1/4 of the amplitude sample range. You then stack (multiplex) these into one output file with the equivalent of 2x your ADC's normal amplitude range (bit depth). That would be the advantage--you could almost double the sensitivity of your ADC while keeping file sizes relatively small. 1/2 the file can be disregarded as padding. Because of "sampling theory" this means you would have to have a sample rate of 4x the highest frequency in the time domain (not 2x as is normal) to maintain fidelity... however input devices don't record like that, they usually use a continuous wave input method which means you can not clip and stripe the sampling that this would obligate. The hardware capturing device itself forces you into a limited system of wave capture methods which railroads you into limited file types. So to put the pieces together: 32bit float was a method for increasing the bit depth of input from the ADC. This is A method for increasing bit depth and another way to do it would be to multiplex input sampling--there are other was to do it as will but it's all tied to how the input sampling is done (and that's still stuck in the 90s). ...also: please name a file format that supports 32bit float that isn't proprietary. Just because I'm not aware of any doesn't mean they don't exist. Audio isn't my focus but I do work a lot on compression and multiplexing. For perspective: youtube, the second most trafficked site on the internet uses MP4 and most media released in the past two decades has been in MPx format... I'm not exaggerating when I assert everything is optimized for MP3/mp4--it's been like that for a very long time.
@@TurboLoveTrain I'm a retired software engineer (got my start in 1965), trained as an electrical engineer. You are either misusing the word 'compression' here (the word is used in different ways in different contexts), or else you just don't understand what you're talking about.
I think a lot of people don't understand it. recording without manual gain is a win, setting the desired level in post is trivial. my old Zoom H6 recorder has an option to record the signal twice, once with a reduced input gain. it recently saved my ass when the event was louder than expected and there was no time to check the level. 32f recording makes this a non-issue, but is more flexible while requiring the same amount of resources (2 gain stages and two ADC channels). I look forward to all recording being gainless.
As a 16 year location Sound Dept. veteran (and hack home studio enthusiast), I can tell you that it is definitely the under paid post department that is to blame for the lack of utilizing 32-bit float. I get blow back for recommending the post department download the Sound Devices "Wave Agent" software, so that they can edit/recall meta-data or edit the mono/poly wave data...let alone, asking them to get into the weeds with editing 32-bit float. Another issue with 32-bit float is "what else is in the signal path?". Your lavaliere's element (or diaphragm) can distort (unless you use the new DPA's) or your transmitter can over-modulate (maybe not if you have the latest A20s by Sound Devices). You'd have to REALLY plan out your signal chain to fully utilize 32-bit float....and at the end of the day, you're not getting paid any more money to rent this kit out...nor is there any guarantee that post will use it. It's not like the camera world, where RAW recording and LUTS were accepted with open arms. Sound Department is literally One-Half of the product, but is viewed as a minor annoyance. It is going to be a LONG time before 32-bit float catches on. In the studio world, there still are no multi-channel audio interfaces with 32-bit float. Steinberg has "32-bit converters" on their interfaces, but they are not full on 32-bit float. It's like "we have a cure for cancer!" but the world responds with "meh....". Frustrating for sure.
Thank you for sharing this perspective. I’ve been very curious about that. It seems like manufacturers are waiting for demand at the professional studio level. Zach at Sound Devices said their pro level hardware (Scorpio and 8 series recorders) support 32-bit float in the hardware but they haven’t released firmware for it yet because of lack of demand. Meanwhile, the MixPre II recorders support it and the 10 can even be used as a 32FP audio interface.
Hi! I am just learning about these mobile recorders and wonder how you would attempt recording a band, say drums first, then bass, etc.. Do you hear playback when overdubbing? What about different takes? ... Probably trial newbie questions. Your videos are 👍
i think 32bit float is great. the only concern i have is what some people pointed out; pushing work to post can disrupt existing workflows. i work in radio broadcast as an audio engineer, and the recent introduction of izotope RX and supertone clear made for some pretty grueling experiences where reporters asked me to fix their shitty reverbant (and sometimes clipped) field recordings, because they know we can do that, to an extend. if we had 32bit float some issues we have now wouldn't be there, but i feel like reporters would care even less about standards that make sense beyond the old tape noise floor logic. having the tools to fix things in post is nice but doing it right from the start is often very easy, especially if the workflows for it are long established. if i end up freelancing as a videographer again, i would probably consider using 32bit float, because i'm on my own and i save time by fixing it in post. at my job, there is no time to fix it in post because it's going on air in 5 minutes. different workflow :D
Thanks for sharing your professional perspective! It's definitely true that 32-bit float could disrupt established workflows since it requires normalization in post. However, it could also save substantial time by eliminating the need to use RX to fix the situation you mentioned with clipped recordings (unless the mic itself is overloaded). With 32-bit float, your effective gain setting happens in post, which might lead to higher quality recordings overall. While normalization adds a step, it might replace much more time-consuming repairs for clipped recordings in your fast-paced environment. I understand your time pressures in broadcast. Do you think there are potential workarounds or workflow adjustments that could make 32-bit float viable in a professional environment? The human element is always the hardest thing to change, especially where it crosses roles and responsibilities. I'd love to hear your thoughts.
If you export your 24 bit 32 bit float recording example to Soundcloud in 24 bit wav format i think we probably will hear a difference. But You tube audio quality MP4 is often about 192 kbps MP3. What do you upload?
Great video. I'm going to say the problem with 32-bit is that we lack a device that can handle say - 12 channels. Presonus new interface has 8 channels at 32, which is great - but then I have a bottle neck with all my outboard ADAT gear. I felt if Presonus made two interfaces that you could mate for a total of 16 inputs at 32 bit float.
Thanks! The Sound Devices MixPre 10 has 8 XLR Channels. I have two synced together with a BNC cable for timecocde, providing 16 XLR channels and four additional unbalanced channels. They physically connect together so you can stack as many as you want. The Scorpion has 32 channels.
Yeah, I'm a huge fan. It's all really solid well-built equipment. I'd like to make a video about linking up multiple recorders and recording a live session.
Quantization noise doesn’t exist in fixed point bit depth formats. It exists only in float point formats. Quantization is just the process of division by two. This is what every resistor does in the comparator chain of an ADC. The remains are rounded and there's nothing to worry about, as the range they are responsible for is below the dynamic range of the bit depth. So 24 bits fixed doesn't introduce quantization noise. The noise we have at - 144dB is just phase noise. But in floating point formats we deal with exponential numbers, so they are much less accurate, than fixed point. Here's where we'll find quantization noise right above 0 dBFS and far below -138 dBFS. Again there's nothing to worry about, as all these inaccuracies would always be far beyond human's hearing range. And even if we amplify them we will hear subtle changes in volume of every phase of the signal, which are so tiny that they can never be a problem in comparison with the tolerances of the analog chain components. Shockwave and 194 dB are not connected with each other. Shockwave is the product of Doppler effect and the speed of sound. 194 dB peak (or 191 SPL) cannot be exceeded due to atmospheric pressure, which is equal to 1 bar or 100000 Pa. If we rise the pressure level, we could exceed the maximum amplitude of a sound wave. Dither is not used in 32 bit converters. Two parallel signals are analized, then divided into portions which are picked in turn and merged together when they fit the parameters.
Simon: Good piece! I learned a couple of things. At 13:21, you said “Obviously the 1528 decibel dynamic range claimed of these recorders is excessive, since anything over 194 decibels is considered a shock wave.” You’re confusing two entirely different things with this analogy. Dynamic range is the difference between the highest level you can record without clipping and the level of the noise floor. That range compares one level to another and is expressed in dB. The 194 dB figure you quote is a sound level, not a range, and it is more properly thought of as 194 dB above the Threshold of Hearing. As such, it is expressed as 194 dB-SPL. The SPL stands for “Sound Pressure Level,” and it is a specific standard level, 0.0002 dynes/cm2, that is the minimum level detectable by humans. When using decibels compared to a reference level, the reference is always indicated, hence the “dB-SPL.” In this case, we don’t know the signal level of the highest audio signal we can record on a 32-bit floating-point recorder, but what we do know is that the lowest signal possible, at the noise floor of the low-level ADC, is 1528 dB lower (i.e., there’s no reference level). It would be perfectly valid, for example, to set the clipping point of the high-level ADC as 0dB and the noise floor at -1528 dB. That said, there is a fallacy that’s being overlooked here. While the recorder has a theoretical dynamic range of 1528 dB, the program (i.e., the signal that’s being recorded) will always have a signal-to-error ratio of approximately 141 dB. The reason for that is that it’s limited by the 24-bit mantissa. (24 x 6dB/bit, less 3dB due to quantization error of the least significant bit, which occurs 50% of the time.) Another way of stating this is that while the recorder will automatically adjust the gain up or down (using the eight additional floating-point bits), the noise floor can never be lower than -141 db below wherever the peaks are because whatever you record has only 24 bits to describe changes in the waveform. Essentially, what I am saying is that a 32-bit floating-point signal IS a 24-bit signal that can be freely rescaled without clipping or added noise. This is why the 32-bit float and the properly-recorded 24-bit signal sound identical. There is no difference in sound quality. The question I am not hearing addressed by proponents of 32-bit floating-point recording is why one would increase their data storage needs and the time needed for transmission of that data by 33% for no perceptual increase in audio quality? A 3-minute stereo audio file at 48.0 kHz and 24 bits is about 50 MB. A 3-minute 32-bit floating-point recording at 48.0 kHz is 66 MB. They sound identical, so what am I gaining by going with the larger file size? The one advantage I am getting by recording with 32-bit float is that I don’t have to set levels, and even when the signal is recorded too hot or too low, I can change the level in post-production. Cool, but guess what? We can only hear about 60 dB at any one time. The Signal-to-Error ratio of my Audio CD is about 93 dB, which means I can never hear a CD’s hiss. No engineer will tell you this, but you could set your maximum peaks on a 24-bit recording at -24 dB. Your noise floor will be around -141 dB because it’s 24-bit. Then you go ahead and give the recording +20 dB of gain (try it!). Now, your peaks are at -4 dB below clipping, and your noise floor is raised to -121 dB. That’s well below the noise on a CD and still completely inaudible. In fact, because 24-bit has eight more bits than 16-bit recording, it has 48 dB more dynamic range (8 bits x 6 dB/bit); you could bring any 24-bit recording up as much as +48 dB before you would match the inaudible noise floor of a 16-bit CD. Of course, in practice, you wouldn’t do that because you’d also be bringing up any noise generated by your mic preamps up by the same amount of gain. There’s nothing wrong at all with 32-bit floating-point recording; it simply strikes me as unnecessary. You can accomplish the same thing by recording in 24-bit with peaks between -12 and -18 dB (or less), giving you an additional 12-18 dB of headroom as a cushion for unexpected peak levels, and later applying 12 dB of gain (or more) across the board. You would still have to properly gain-stage in the analog domain prior to conversion, but there’s no reason why your 24-bit recording level has to be hitting -3dB. We’re not recording on analog tape. Having said all that, your conclusion is spot on. I 100% agree with your comments about the ease of usability of a 32-bit recorder like the F6. The fact that you can record drums without a computer and never worry about recording levels makes it a great tool for your application. In a studio setting, where an engineer has greater control over levels and chaos, and is recording directly to a computer, the advantages may not outweigh the additional data overhead.
I’ve always worked in 24 bit and I’ve never had a problem when I’m in the studio alone. BUT I often record people that have no to little experience speaking or singing into a microphone and I’ve had to do many takes over because someone thought when they suddenly scream they have to get closer to the mic to accentuate that -- and whoops ruined… I assume I should try working in 32 bit (considering to buy the SSL 12) to enable a 32 bit workflow…
You're absolutely right that 24-bit recording can be more than enough in a controlled studio environment. But the challenges you describe with inexperienced performers are very real. While the SSL 12's 32-bit fixed-point format offers a huge dynamic range, it doesn't eliminate the need to set gain before recording. If you're looking for a way to completely bypass the need to set gain during recording, eliminate potential clipping errors, and have a versatile audio interface for your studio, the Sound Devices MixPre series might be worth exploring. These recorders offer 32-bit float recording and can function as high-quality audio interfaces.
Normalizing is the process of adjusting a waveform's loudness level after recording to move the peaks a specified distance from 0dB. The way "peak" is defined depends on the algorithm used, and there are many ways of calculating it, to name a few: • the root-mean-square method (RMS) • loudness units full scale (LUFS) over a short duration, longer duration or the full track. • the actual momentary highest peak of the signal For mixing songs, individual tracks are often recorded so that the body of the waveform is around -18dBFS with the peaks a little higher. There's a judgement call here, because when setting levels, intuition is often used to gauge what "body" means. And you don't always know exactly how loud a performance will be compared to a sound check. After recording, you can normalize a track if you want to correct any mistakes made during the recording process, though this can be destructive (unless you record in 32-bit float). Alternatively, you can adjust the faders. Then, in mastering an overall, integrated loudness level for the full track can be targeted. For TH-cam, I target -14 LUFS-I for the master per their specifications.
I asked them that and they said the hardware supports it. Lack of customer demand has prevented them from including it in the firmware. As I understand it, on the post production side of audio it has been negatively stigmatized and probably misunderstood. And I think there’s concern that recording professionals will be more cavalier about setting levels, passing that work onto the post house. But I suspect we’ll see it eventually. The industry is slow to change.
They’re about to when they release the new 10.0 firmware update. Just as I suspected it wasn’t a hardware thing for the 8 series given their earlier released MixPre series were already able to do 32 bit float. It was some perceived unprofessionalism or bureaucracy in the types of people using their flagship products. The Kashmir preamps have been able to do it all along. I will be happily upgrading to the 10.0 firmware in both my 888 and Scorpio when Sound Devices release it. In my humble opinion, the big obstacle is post production being able to effectively/efficiently use the file type or the post sound supervisor potentially inferring through some passive aggressive communication to the powers that be, that they hired a lazy production sound mixer.
With the zoom, which I believe does not have a gain control, how do you deal with the noise floor of the preamps themselves? I would think that on a much louder source, turning down the input gain would net you less noise induced by the analog part of the preamps themselves into the 32bit digital recording. Even with two separate preamp circuits, that is two different gain levels, and a compromise on the analog side, I would think? Especially if you are using a microphone like an RE-27 that requires a lot of gain to get a usable level out of it, about 60db of gain on an 1173 with a singer right on the grill, but less if you are using it on a floor tom with a heavy handed drummer, or a kick drum, etc.
I'd have to say that I don't think that noise is a problem with it. Although, I haven't conducted an extensive analysis, I've used the F6 with low-level spoken word and cranked the gain up with headphones attached with various microphones. The only electronics noise I've noticed is with a tube mic. The F6 is really geared for dialogue, so I think they designed it to perform well at lower SPLs. I think their target market is filmmakers, so when I record drums with it, I can clip the analog circuitry of the thing. (I recently acquired a MixPre 10 II and have not been able to clip the inputs, even with close-miking on the kit.) I recorded and wanted to include a noise example in this video, but I don't have access to an anechoic chamber and the quietest space I could find had room noise much louder than the internal noise of the F6, so I had to leave it out of the video. While there are likely two different noise levels between the preamps, it's relative to the signal. So the higher noise level of the high gain side would still be way below the signal that's being captured from that side.
lower noise floor and no clipping, but you pay for that with audiofile sizes. prolly make sense for live music, especially something with very quite parts. for in the box edm stuff 44.1/16 is plenty enough
Can you explain the normalization advantage? In traditional DAWs you just click the normalize button and that’s it so I’m wondering how it gets easier. I understand how it gives you more flexibility, but not how it makes it easier to normalize or faster.
What I meant was that in my experience it’s easier and less time consuming to normalize in post than to set the gain perfectly when recording, especially in a multiple microphone scenario.
I have Behringer XR-18 in the details I read it says 40-Bit floating-point DSP features “unlimited” dynamic range with no internal overload and near-zero overall latency , is that correct if anyone knows! Thanks
The key in there is that the internal processing is 40-bit. Meaning within the internal processes (EQ, comp, FX, routing...) it should be impossible to clip etc. Not meaning that the input side and A/D conversion is floating point or infinite DR. I am sure you will add noise or clipping with improper gain setting on XR-18. It would be misleading to exactly call these devices 40-bit floating point devices.
Why won’t it catch on? Does it sound worse? No Does it sound better? No Will any of us ever record anything with more than a 90db range? Probably not. Save time by not setting gain? Combine your process of normalizing with 40 tracks, many with multiple takes, over a 12 song record, and I bet getting it right at the front end is quicker. Still, how am I supposed to send a decent headphone mix and monitor what I’m recording without proper gain ? You almost get one of the most important parts about the actual sound at the end. Using an API (or any analog pre) will require you to adjust the gain, and how you adjust that gain has a significant impact on what you get out of that pre. Studios aren’t going to give all their preferred pre amps even if the above mentioned problems weren’t a thing. This set up is nice for throw and go, but it’s kinda solving a non existent problem.
Well... that's kind of my point. It sounds the same, yet it can offer recovery from mistakes and save time in some cases. For me, normalizing 40 tracks takes the same amount of time as 12; select all, normalize. I suppose there could be some clips that would be sliced up, depending on what you're recording, but... let me flip it around. There's nothing preventing you from setting levels when recording in 32-bit float (at least with the MixPre and F6). It still sets the peaks relative to 0dB, even though it doesn't change the recorded waveform. You could work exactly the same as when recording in 24-bit, with the benefit that if you didn't happen to nail the level perfectly for a track at record-time, it could be recovered it in post. Another way to think about it: if you had started out recording in 32FP would you be actively seeking to switch to 24-bit? For headphone/monitor mix, you could still set the knobs as you have in the past, or if you're using Sound Devices, you could use Auto Mix: www.sounddevices.com/automatic-mixing-101/ Regarding an external pre, I didn't go into it in detail, but that's exactly why I want to be able to use one. I have used an AEA TRP2 into an F6 and was able to clip the analog portion pretty easily. I haven't tried that with the MixPre yet, but I'm certain it will be able to withstand a more preamp gain.
@@simonpeck I can’t even imagine the clumsiness of tracking a band and a trying to get good headphone mixes to 4 or 5 people with an unknown gain structure, then trying to do a punch in with an unknown gain structure tweaking volumes to give suitable working headroom after the fact, trying to integrate outboard gear, doing punch ins, etc. seems like a pretty messy way to work, nor do I ever want anything at at 0bbfs. Why wouldn’t I bother using 32bit float? As stated, integration with the analog world. I do have the capability of recording at 192khz but never do. Taxing the CPU load with no benefit isn’t in my nature. It’s so, so easy to set a good gain structure, replacing 32 channels of AD and DA converters would be a pointless expense. Again, seems a fine way to do what you’re doing, but I wouldn’t expect it to become a standard in the pro world anytime soon.
@@zagatoalfa No, this is way overthinking it. It isn't that complicated. It's just a data format. So somebody could walk into your studio and change your recording format to 32FP and you wouldn't even know it. You would do everything exactly the same. The only difference is that if you happened to make a mistake or didn't set the levels perfectly for a channel, you can correct it after the fact with no negative impact on sound quality. That's it. This assumes, of course, that your interface supports 32FP. The scenario described -- where the position of the gain control knobs is totally ignored is not a requirement when using 32FP.
@@simonpeck exactly. I could run my sessions at 32bit and not notice and setting gain isn’t that hard, so I won’t bother running 32b float. I also don’t think considering implementing the technique of not bothering to set gain in a real world situation is over thinking it. A quick search shows studio based 32bit float A/D or D/A conversion. Only Sound Devices and Zoom field recorders, despite the tech being around for some time.
When I record my acoustic guitar with the Zoom H4 essential (about 12" from the neck/body joint) and bring the Zoom file into Reaper, the signal level is so low I have to boost that channel to its max (+24db) to get peaks anywhere near 0db. Is this normal, Am I missing something?
What format were you recording in? If 32 bit float, you can normalize it in Reaper. If not, you need to make sure your levels are set appropriately when recording. Are you using the built-in mic?
@@simonpeck Yes, 32bit with the built in mics. Near the end of your video I saw how to normalize the levels in Reaper, and set for 0db peaks i get 30db+ gain bringing it up to a usable level.
A good comparison would be a camera's focus (or f-stop/ISO/shutter for that matter). Imagine shooting a 30 minute interview with the camera out of focus. Imagine being able to pull that focus in post. That's what 32 bit float can theoretically offer. Most professionals manually pull focus in the same way audio engineers gain stage, There's a specific person that solely operates and pulls focus and they get paid LOTS. Most resistance to utilizing auto focus was purely due to it's poor and inaccurate implementation, usually resulting in undesired, adverse, effects. People need to experience 32 bit float in order to trust it, in the same way you would need to experience and play around with good quality, modern, auto focus. Auto focus works well for people who don't have the in-depth knowledge or patience to dial in exactly what they want, 32 bit float satisfies that exact notion. Someone who's been working in audio for ages would have zero need or use for 32 bit float because they know exactly how to capture what they want, the same exact way a photographer uses manual mode because they've learned exactly how and why to configure the settings to accommodate the desired outcome. Technology like 32 bit float audio recording is such a massive step, that it's hard to even justify the amount of freedom it could restore to an artist. We've grown so accustomed to only being able to capture a hollow representation of audio and that it's been proven time and time again that any step further gives us diminishing returns. So despite the heavy resistance in standard use cases, 32 bit float can and should thrive within it's own, separate, classification of audio capture. The same way we have stereoscopy, or 360 action cam videos. At the end of the day it's just simply a different way of representing captured information, and therein lies the root of the problem; since it's a different way of representing audio, the thought process revolving around it should also be reconsidered.
Woody, I'm thrilled to see this post. This is an excellent analogy. I used to work in the film industry and have witnessed professional camera assistants pulling focus first-hand! Interestingly, when I interviewed Zach, he mentioned the comparison to light field photography where you can focus an image after the fact. I cut that part out of the video, but perhaps I could post the uncut interview if there is interest.
The other problem with autofocus is that it's hard to use when you want to pull focus in a certain way or keep some things slightly out of focus. Also most AF systems can't be programmed to keep two points within the DOF or go slowly between A and B. The closest I can think of doing something like this are the broadcast servo lenses where you can control the focus servo speed. Decades ago Canon had a two point DOF AF in some of their film cameras but IIRC this feature was never carried over to the digital models and it was not a very good implementation anyway.
We don't need that Kashmiris Sarcasm to be floating around to be sure of what we're doing 24/7. I don't know, I never had those extra hours without taking flights. Namastè.
I am confused, just casually saying to normalize the audio like its nothing. Isn't normalization similar to compression? Its bringing up the quiet parts and bringing down peaks? Thats going to change the sound? If it didnt change the sound then it would just be gain and not called normalization.
Normalization measures the difference between the loudest peak and the desired level of the loudest peak, and adjusts the whole recording by this difference to make the peak match the desired level. Either the whole recording is turned up or down, there is no difference for quiet or loud parts, so normalization is not similar to compression.
@@simonpeck AHA! Cool man.. I didn't notice if you mentioned the spelling in the video, but I have the attention span of a gnat when I am listening at work lol.. This was very informative and answered quite a few questions I had about 32 bit recorders, but was too lazy to research! :)
The brave thing would be to make recording devices with no gain control whatsoever. The gain control doesn't set the "zero point" but the "one point": you're deciding how much dynamic range is between the number zero, and some other number that is not zero. Also, your introduction of the term "dithering" here is inappropriate, it has nothing to do with the range selection/combination process. Also, having more than one ADC is not strictly necessary, and for all intents and purposes 24-bit integer PCM is also fine for gainless line level ADCs... and in practice probably also fine for gainless mic preamp-ADCs, but there are extreme outlier microphones for which this is not the case.
These devices effectively don’t have gain control when recording in 32FP. No matter how you set the gain, the exact same waveform is recorded. It only defines how the waveform appears when you initially import it into the DAW. I imagine it’s some metadata stored in the wav file. Whether it’s called a “zero point” or “one point” is a matter of convention, as I understand it. I could be wrong about whether or not dithering is used in combining the data from multiple ADCs. That was speculation based on research I did on possible implementations. Sound Devices doesn’t reveal their implementation details. You’re right that they could use a single 32-bit ADC, but they don’t. From what I read on these converters, there are technical challenges in doing so accurately. I can dig up the source if you’re interested. At any rate, it was their design decision to use multiple 32-bit converters so I trust they did that for a reason. The most popular microphone in the world is the Shure SM57/58, which has a max SPL of 180 dB, so it wouldn’t need to be an outlier microphone to produce a wide dynamic range.
@@simonpeck they do have a gain control though, the knob is there, that's what I mean. As for calling it the zero point... That makes no sense, setting the zero point is setting the DC offset, not the scale... Also, it's not necessarily a "32-bit ADC" that you would use, since ADCs in production do not produce floats regardless (though theoretically there is a way to design a self-ranging ADC that produces floats). If it's a 32-bit ADC today, it's still only producing integers. As for their decision to use several ADCs, that will depend on their particular architecture, but it is not necessary in principle. I know the limitations on single ADCs in these applications.
@@microcolonel Ahh, yes, these do have knobs, but there are 32FP recorders that do not. Setting the zero point refers to setting how the peaks will appear relative to 0dB, not DC-offset. When I mentioned 32-bit ADCs, I was referring to Sound Devices' implementation, not a theoretical one. It's what they use. How they convert the output of those ADCs into 32FP is proprietary.
24 bit integer precision makes zero sense for listening to music. 24 bits are only needed for writing music because you can record your signals low and then boost them without boosting the noise floor caused by imprecision errors. You need to record your source at a very low volume, like barely hearable, much lower than whispering, for 24 bits to start having issues - but why would you even do that? 24 bit precision is 256 times more precise than the default 16 bits and a finished track will be indistinguishable between two formats. 32 bit float precision is also quite questionable because floating point numbers can be very imprecise and result in errors. To me the format always seemed like an internal way for very old DAWs to work around the hard limitation of very old audio interfaces that could only do 16 bits at a cost of extra CPU power.
It depends on what your goal is. It makes total sense if you want to save time and reduce the possibility of human error while recording. Floating-point rounding errors are inaudible, because we’re talking about extremely small differences in amplitude.
Auto Align 1 is great on drum tracks. Auto Align 2 is even better. th-cam.com/video/K3DOCTx_qS8/w-d-xo.html Of course wav 16, 24, 32 float or 64 float bit ...it is an upgrade but these things like Auto Align give a lot of depth and sound to the drums that is far beyond the improvement going from 24 to 32 bit float. Phase and alignment is important/essential for drums and for a mix.
Well? That wiring scenario. Will cause your speakers. To mimic, the motion of the microphone, diaphragms. And which way are those microphone diaphragms moving? When a sound pressure wave, blasts into the microphone diaphragms? Which way is that microphone diaphragm moving in? That's right. It's moving in. It's being compressed into the body of the microphone. As the sound pressure wave, smacks into it. And you absolutely positively, don't want your speakers. Moving like your microphone diaphragms. Now would you? Isn't that kind of backwards? Shouldn't the speakers be moving in the other direction out toward you? Because they are not. They are sucking inward, away from you. The speakers are sucking away from you in front of you. They are not protruding out, toward you. To provide you with a Positive Pressure Wave Immersive Experience. No. It's sucking sound away from you. And it sounds all wonky and crappy. Because it's in Negative Absolute Polarity. Unless you turn both wires around on both speakers. As connecting the positive terminal on the amplifier to the positive terminal on the speakers and the negative terminals together. Isn't the right way to connect it. No. Positive does not get connected to positive. Positive gets connected to negative. So current can flow. And negative gets connected to positive. So current can flow. Because current cannot flow. When positive is connected to positive and negative is connected to negative. Current cannot flow in a DC circuit that way. So the electrical engineers know, DC Theory 101. Positive gets connected to negative. You are not jumping a dead automotive car battery. When you connect the positive to the positive. And the negative to the negative. In that parallel wiring. That's to jump a car battery. That is not how to connect speakers. Where positive gets connected to negative. Because the electrical engineers that have designed amplifier circuitry. And the speaker itself. Had to follow DC theory 101. With positive getting connected to negative. So current can flow. And then when they get to the output of the audio power amplifier. They go against that wiring? And they connect the positives up? Why that inconsistency? Since that is not DC theory 101. What's the matter with them? Why would they change at the output? And cast their DC theory 101 out the window? But they in fact, have. A gross inconsistency. In technical engineering theory. One of those what the FUCK, moments. That makes no sense. No engineering sense. The reason for this anomaly, inconsistency and downright incorrect way of wiring your speakers. It is due to, grossly ignorant fools. Who have demanded. From the manufacturers. THEY DON'T WANT ANY OUT OF PHASE ANYTHING IN THEIR SIGNAL CHAIN! Yes but what about where the polarity has to be inverted to obtain Absolute Positive Speaker Polarity? At the output of the audio power amplifier to the input directly to the speaker. Which is the only place. Where Absolute Positive Polarity can be technically Established. Polarity cannot be inverted. At the output. By simply inverting the phase of the waveform, at the input of the audio power amplifier. That does not change the absolute polarity at the output. Which will remain the same. And nobody understands this. Or what is occurring? Because they think they have their wiring right. They refused to listen to it the other way. Because they don't want to appear, stupid. So they shun. Even listening to it. Where you will know, instantly. You have been always listening to it the wrong way. For many years. From the get go. Nobody ever teaches anybody this. And nobody can think this through, for themselves. Though JBL did. Back around 1948 I reckon. It was certainly by 1968. When I first came into that scene. Now I have fixed and corrected a lot of big expensive, well-designed, recording studios. That just never sounded right. Because of miswired speakers they thought was right. Because of ignorance and stupidity. And the ability to exercise one's, gray matter, synapses and neurons. So if you flip the two wires around. On the back of both of your speakers. You will then force your speakers into Absolute Positive forward Polarity. And you will hear a whole new listening experience. For those of you. With those fine, Active, self powered, Speakers. That you also have to plug into an electrical wall outlet. For each speaker. You are going to require a, screwdriver and soldering iron. You will remove both the woofers and the tweeters. You will use your soldering iron. To reverse the connection. Both on your tweeter. Then on your woofer. Then you will reinstall the speakers into the box. Do not over torque down those screws. You can warp the speaker frame if you do. Just make it snug. No Hulk action death grip tightening. No. Just make it snug. This will correct your speakers into Positive Absolute Polarity. You will now hear and experience a vast difference. Like you've never heard before. It will sound the way you always thought it should. But never quite, did. Now it will. This is all part of my Control Room Design Concepts. For a Positive Pressure Wave Control Room Design. I make certain of that. I'm known, for that. And then your lousy acoustics in your room. And we both know they are lousy acoustics. Will seem a lot less, lousy. They may mostly disappear. They may sound different. They may not be as wonky and resonant. It changes the whole feel of your room. It's not like you've heard it before. For those of you fortunate enough to have chosen, JBL Speakers, for yourselves. You are good to go. Do not change anything. Do not rearrange your speaker wiring. They were designed to be, Correctly, Miswired. That's right. JBL actually did that. So you would be hearing their speakers correctly. As they marked the positive and negative connections, backwards. But they aren't backwards. They are correct. To ensure. The speaker will protrude out toward you. To provide you with a Positive Pressure Wave. Surrounding you. Enveloping you. Sending you two new listening heights.. You will hear height. You will physically hear the lead singer. Standing at your height. Just feet in front of your face. Where you can hear their voice. Coming out at you. Right there in midair in front of you. It can seem a little eerie. A little strange. Kind of weird. There is a voice hanging in mid air in front of you. You can almost reach out and touch. You've not heard it this way before. You will hear stereo like you've never heard stereo. It will have a very wide stereo soundscape. Wider than you have ever heard it, before. It won't matter where you sit or stand. You will hear glorious stereo. The other strange psycho acoustic phenomenon. The timing of everything arriving at your ears. Will be almost precisely. The way the producer, engineer and artist. Wanted you to hear their creation. Their work of art. That way. My way. The JBL wiring way. That all other speaker manufacturers in the world got backwards. Beyond stupid, clueless and totally ignorant. You haven't lived. Until you've heard Positive Pressure Wave wired monitor speakers. You have not lived. RemyRAD
You’re missing the point. It has nothing to do with a dynamic range increase. Rather, it’s about optimizing the bit stream so that the 16 or so bits actually needed can’t possibly go unused. Two preamps with very different sensitivities permit encoding of so broad a digital range that the full required range cannot fail to be encoded and be decodable.
@@artysanmobile I'm not missing the point, the claim is BS, 32 Bit is capable of recording 194.4 dB, that cover everything from a pin drop to an explosion. A jet fighter on full afterburner at 10 meters is 150 dB, that's more than enough for any situation without making a ridiculously false claim.
Releasing 32bit was a smart marketing ploy to generate sales, and is still primarily being marketed towards videographers (and anyone) who ever distorts in 24bit, (and even 16bit under normal real life recording, mixing, playback), which indicates they don't know how to gain stage yet. If someone says they need 32bit because they can't be there to monitor levels does not understand proper gain staging. Even a rock concert can be recorded on stage perfectly fine in 24bit and even 16 bit...because it's not about max level, it's all about DYNAMIC RANGE. -Not allowing the sound engineer to adjust gain is like putting a race car driver in a self-driving car. -Bit depth does now equal sound quality, PERIOD -Bit depth = theoretical dynamic range PERIOD (some manufacturers quote specs that are not possible in even the coldest temperatures on earth. Also specs are quoted that are component based (in theory in space) and not based on full circuit design. -Human pain threshold (frequency dependent) starts around 120dB -An electric drill at 3-5 feet is approx 96dB -Loud motorcycle at 10-15 is approx 96dB -Rock concerts measured front of stage 110 -120dB -An extremely loud scream at 1 foot away is approx 120dB -Full stack guitar cabinet (ie: 4 12-inch speakers in a 4x12 cabinet) at 2 inches away is approx120-130 dB Max dynamic range available on BlueRay and all 24 capable streaming services is 144dB -32 bit is for those recording levels that exceed +/- 144dB -Professional mic can handle spl often up to 140-160dB spl without distortion (for use touching the guitar cabinet grill) Can and should, can and do are all different things. -A Boeing 747 at 6' away would be somewhere in the 150-160 range - instant deafness occurs around 160-180 (very close explosions/gun shots) Earth's atmosphere pressure (at sea level) limits us to around +/-194dB SPL , This means it's impossible to produce sound levels louder than that, because you can't have less than a vacuum in terms of pressure. Pressure beyond +/- 194dB is no longer sound, it's shock wave. (ie:explosions) Death would occur instantly at 240-250dB, because again, it wouldn't be sound beyond 194dB, it would be you standing in an explosion an d dying from a shockwave. PUNCH LINE: 32Bit floating point recording has a theoretical dynamic range limit of 1500dB. I challenge ANY manufacturer to demonstrate how 32bit is "better for the music and recording industry", vs a money grab, for all outside of scientists and sound designers and OSHA etc. It's not adding quality, it's simply adding more seats to the bus for potential loud volume. Now instead of taking a family trip in a 144 seat but, the industry is forcing a 1500 seat bus that messes with the audio levels I'm getting from TV and film video editors. What a mess! Instead of doing the right thing and educating the market how to actually use their devices, they reboot their entire line with designed to record more dynamic range than anything anyone will ever want to subject themselves to. If ANY manufacturer disagrees, I will very gladly produce a nice 32bit recording taking advantage of the technology and provide a listening room for the executives to gather in for a full dynamic range listening experience to see what they've so generously provided the industry. Marketing departments are advertising 32bit as an "upgrade" in sound and the best the industry has to offer etc etc etc. Misleading and false, PERIOD Happy to discuss with anyone who genuinely wants to get past the marketing smokescreens and the "tech support" and company representatives who have "pro sound" experience who ever say you need 32bit capability because it will improve your sound. When talk of preamps and/or any other part of the device is discussed along with bit depth you can immediately know the person actually doesn't understand the function of bit depth. Bit depth are the seats of the bus and have nothing to do with the quality or capabilities of the engine of the bus, or the tires, or the radio, or the windows or the seat belts, or , or , or....., beyond how they are given the info from the audio path PERIOD. ----- 28 years in the professional audio industry. TV, Film post, sound design, film composing for 3D4D motion ride , immersive audio (incl Dolby Atmos) for streaming and cinema. Music and audio systems consultant and designer/integrator. More than 850 hours of Direct face-to-face education and hands on experience and side by sides coming from a broad majority of the worlds top audio equipment manufacturers producing AD/DA and sound recording devices, mixers, monitors, amplifiers and processors etc.
Not needed at all, and to my surprise it even defeats its purpose, taking into account what Zach said. LoL. Besides, those ZOOM preamps are really a piece of crap, they re really bad man
Well? I worked in Pro Audio Manufacturing. Starting over 40 years ago. And I'm what you call a Studio Fixer. As some of us have, well seasoned, extremely acute, superior, hearing apparatus. In other words. I hear things nobody else hears. And I'm not an audiophile. I don't believe in, audiophiles. They haven't a clue what they're hearing or what they are listening to. Neither, do you. Yeah… In a double-blind listening test. With over 30 other well seasoned top engineers. Apparently. I'm the only person. Who could tell the difference. Between an original source recording. And that of, 24-bit, 192 kHz, high-end, PCM, digital converters. Yes, it's true. Everybody else got it wrong. What did they get wrong? Everybody else thought. All of what I heard as artifacting. Was the original source material. But it was all artifacting. This made me realize. In all of the other high resolution digital audio demonstrations. None of it sounded like original source. None! It was all loaded with artifacting. At any bit depth. At any sample rate. It continues to sound all, fizzy and grainy as PCM. Because it is, PCM. A flawed digital format. So recording a 32-bit float. Makes no difference. It's still PCM. It might be PCM that has a theoretically large, dynamic range. Spanning more than 150 DB. Well Lah Dee Dah. None of the input or output electronics can deliver, anything like that. Nothing like that. It's a phallus, see? This is such idiotic mumbo-jumbo. That it hinges upon, wacky techno-audiophile gobbledygook. That actually has no bearing in, Audio Fidelity. It's more gobbledygook garbage because it's still PCM. And it will sound like PCM. Because it is. It does not sound like original source.. Because it's, PCM. You don't even know. That you're in phase, left and right channel monitor speakers. Are both in, Negative Absolute Polarity, together. Because you think it's the other way. You are convinced. You have a test. You can prove you are right. You are not right. You are absolutely wrong. As this is one of the biggest, acoustic, Technical FUCK up's in Pro Audio History. Because everybody is working on in phase but Negative Absolute Polarity Monitor Speakers. Because they think it's the other way. They can prove, it's the other way. No you cannot. You cannot prove anything. I can. I have. I do. And I can prove how stupid you all are. And you can hear for yourself. How stupid you have been. When you follow my advice and suggestion. And what is that? What is the fix for what I'm talking about? Of the way your monitor speakers currently are. In phase but in Negative Absolute Polarity, together, in phase. Nobody understands that Polarity and Phase are actually 2 different things. Though based on similar concepts. They are vastly different from one another. In the land of Pro Audio. Unfortunately. Most other audio guys are also reaming idiots. And don't know what they are hearing. Don't know what they are listening to. Don't know the way things are supposed to sound. But they have college degrees. That says they are supposed to know. Unfortunately, they've never figured it out. All of the electrical engineers and all of the acoustic engineers. Have been getting this issue wrong now. Going back well over 70 years. It's always been wrong. Though it changed. For some. Back in 1948. From a single speaker company manufacturer. JBL of Los Angeles California. This is not an advertisement. This is not an endorsement. This is a Technical Fact. They figured this out. No other speaker manufacturer. Has ever figured this out. Never. None. Because they think they are all, right. And they have a flawed test. To prove it. With a 9 volt, battery. The 9 V battery. When connected to a speaker. Will cause the speaker cone to either move out or to suck in. You wanted to go out. And that indicates. The speakers Absolute Polarity, for connection. And that would be plus polarity. Now when you connect that speaker up to your amplifier. The instruction manual for your amplifier and your speaker. Indicates you should connect the positive terminal of the output of the audio power amplifier. To the positive terminal on the speaker. And the negative terminals get connected together. And that is absolutely wrong. And here's why. When you do it that way. To get everything In Phase. It will be in phase in Negative Absolute capillary. Meaning that. The speaker is not protruding out toward you, the listener. No.. The speaker is sucking inward. Away from you. It's sucking into the box. It is not protruding out toward you. It is sucking inward into the speaker box. Because it is in phase. With the microphone input. And that's just plain wrong. Why is it wrong? (More Happy Cluelessness in following post)
![BOWKYLION - วิงวอน (ex-change) [Official MV]](http://i.ytimg.com/vi/cLdnZWDaO-w/mqdefault.jpg)
I am a professional classical violinist, but I also have a business doing live concert recordings of classical ensembles. On occasion, I end up recording a concert on which I also perform. When that happens, I can't just sit out in the venue during the dress rehearsal to work on levels. The best I can hope for is a few moments at the beginning of rehearsal where the conductor will have the orchestra play one of the louder sections of the concert while I frantically set levels. After rehearsal, I would then go home, listen back and make notes about what I needed to do differently for the concert the following night. Still, I always had that "did I leave the oven on" type of feeling on stage wondering if I got the gain staging right. After all, the dress rehearsal doesn't account for the audience clapping or the performers getting extra amped up for the concert. This is especially true of vocalists, who rarely sing in full voice during the dress rehearsal. The obvious solution, of course, would be to not record concerts on which I also perform, but then I would make half as much! Then, 32-bit float entered the picture. I replaced my Zoom F8 with the F8n Pro and have not had that feeling since. Turn it on, hit record and the levels are perfect every time. This also comes very much in handy when I am not performing. I no longer have any fear when accepting recording gigs on short notice. I can walk into a venue, spend all of my pre-concert time assessing the room and deciding on a stereo mic technique and not waste a single second worrying about gain staging.
As for the sound quality, I have recorded, mixed and mastered nearly 100 live classical performances and I cannot hear even the slightest decrease in fidelity when using 32-bit float. If you hear a noise floor, it's your mic, simple as that. And Beethoven's Symphony no 9 with 100 voices is not loud enough to clip the inputs when mic'd properly. Just stop with all of the caveats. If you're not recording Falcon Heavy launches, then you are going to be fine.
If you are skeptical about taking the leap, think of it this way. When I first started doing concert recordings I used a 8U rack with a 24-channel mixer bolted to the top and recorded everything on a Roland VSR-880 hard disk recorder. All of that functionality, and immensely more, is now packed into my little F8n Pro. Now, do you think I have any nostalgia for the good old days of pushing 100 pounds of equipment all around town? Hell no! And I won't miss worrying about gain staging when I only have one chance to get the recording right. I am an enthusiastic convert to 32-bit float and I will never go back.
Rob, Thank you so much for sharing this real-world experience. Your story is a fantastic example of how 32-bit float can transform a workflow. The "did I leave the oven on" analogy is spot-on - and it's wonderful that you've found a way to eliminate that worry from your recordings.
It's always inspiring to hear how new technologies can solve real problems for professionals. Your success with 32-bit float is a great testament to its potential, especially for those who might be hesitant to try it.
I had no idea anyone would get so worked up over 32-bit FP. I record almost everything at 32-bit FP because I work in film, specifically as a sound designer, and the dynamic range is awesome. Same with dialogue. When working in 24-bit I would do something similar to what Sound Devices does with their 32-bit FP; meaning I would record to two channels, with one channel properly gain staged and another about -10dB below that. Using 32-bit FP saves me setup time, and editing time. My deliverables are still almost always in 24-bit 48kHz because that's the current industry standard. But my preference is recording in 96kHz 32-bit FP then downscaling. I've never heard a client complain, and frankly most of them don't even know enough about sound to complain. I will add that I've clipped in 32-bit FP a few times when making sound effects, and had to adjust how I'm setting up mics.
Cool! Yeah, You can still clip the analog input side. I ran into that with the F6, but I haven’t been able to do that with the MixPre yet, even when close-miking drums.
such a great video, thank you so much for sharing this with us!
Thanks for the kind words!
Great explanation! It reminds me of the early days of digital stills photography (late 90s). As a co-chair of the Digital & Advances Imaging committee for the Professional Photographers of America we were introducing and teaching thousands of our members and non members about the new tech. Almost without exception, we heard “digital will never replace film”. As you said it was fear of the unknown new tech and workflow. Try to find a place to develop film now.
As I transitioned into cinematography, the film and television industry went through the same “digital will never replace film” denials 10-12 years later. Some of it is protectionism and resistance to change as you said. Ten years from now only the die hard OGs will be clinging to their antiquated 24 bit recorders and looking for people to repair them.
I totally agree, Rob! I've used that exact analogy myself. It's a great way to illustrate how new technologies can overcome initial resistance and eventually become the standard. It sounds like you have firsthand experience with this kind of transition in both photography and cinematography. Your insight is incredibly valuable for anyone who might be hesitant to embrace 32-bit float recording. Thanks for sharing your perspective!
I don't use 32-bit float because people are now selling perfectly good 24-bit equipment for bargain prices. 16/24-bit was fine for decades for professional workflows. It's certainly good enough for my silly hobby projects for the next few years.
Absolutely. 32-bit float isn’t about better audio quality. It’s more about the changes it enables in workflow, reducing the likelihood for error and recovering sounds that might otherwise get lost. What’s amazing is that the equipment isn’t necessarily more expensive either.
Right. If you’re working at home and not on the clock, as I am these days, you can spare the time to get your gains tickety boo.
But if I ever go back into a studio, do location work or record a band I can definitely see the benefit to workflow of deciding on a baseline level and using that for everything. No more having to decide exactly how much harder a drummer will hit when the red light comes on as compared to when you’re asking him or her to ‘hit as hard as you will in the song’ to get the best level! One of the great variables of life.
Hats off, man. Grest subject to cover. I agree, the experience I had with the F6 was a real gamechanger, albeit for field recording and location sound but now I want to try it for music purposes thanks to you! Subscribed.
These vids are quite good, I’m keen to get into 32 bit mode.
Thank you!
I also work with composers n sound designers that are not audio geeks or industry boffins, to them a zoom 32 bit recorder is just an easy to use device with no gain knob. Jus saying not everyone who makes music is an audio nerd (like me!)
Totally agree. That’s one of the reasons I love this: you can focus on your craft and not worry as much about making a mistake getting the sound.
The argument I have made for years is my consoles in the studio have about 6 DBU (+26dbu) more headroom than the best 24/32 integer converters (which run at +20dbu) on the market so if I want to slam the mixbus or push a mic preamp I am going to overload the converter, the only option was a 32 bit floating point converter like we see today. The dual gain stage of the converter gives us the headroom needed. Sadly there are not any large 24 or 32 i/o interfaces for console users like myself. But I have been using a Mixpre for about a year now as my final mixdown rig. So When we do our final mixes the output of the console is captured to a separate rig running a Macbook Pro and a mixpre. But we really need a Thunderbolt interface with DB25s and around 32 i/o to really use in commercial studios!!
That's pretty interesting... I don't know if it would suit your needs., but the Sound Devices Scorpio has 32 channels. It doesn't have DB25s, but it does support 32 channels of I/O over Dante. I guess it doesn't support 32-bit float in the firmware yet, but when I talked to Zach at Sound Devices, he said the hardware supports it. They haven't implemented it in the firmware yet, largely because there is no demand from [professional] customers. I think this is due to the prevailing mentality that 32-bit float is for amateurs or is a fad.
@@simonpeck I looked at the Scorpio but it won’t work for our needs we need all line level inputs to use the direct outs on our 2 consoles..Secondly USB devices aren’t good enough for tracking simply due to the larger latency they have. We prefer thunderbolt which is now part of the open source USB4 standard so hopefully Sound devices will go that route with the next generation and make it a USB 4 device and it can work with whatever you have from USB 2.0 to a thunderbolt port. But I have said for a decade now that 32 bit floating point is the answer to finally catch up digital with analog recording devices like tape machines since like analog the loudness doesn’t matter. 32bit float is also better since is doesn’t have any distortion artifacts as well. But you will find in the audio world most people want to work like it is 1977 and only embrace technology when they are forced too. Cheers!
@@joesalyers 😂 (Great comparison to tape, btw. I hadn't thought of it that way, but makes total sense.)
@@joesalyers where in analog does loudness not matter?!
@@zagatoalfa What I mean by that is that there is no hard ceilings that produce undesirable errors in analog. So like I said previously. The best integer converters can only handle about +20 dbu and that is the very highest end of converters chips and not interfaces like Focusrite or Presonus which max out at +14 dbu. So a basic analog console can run at +26 dbu before distortion which is a bit higher, so this is why it has always been a challenge to work with most audio interfaces in a 24 bit environment. There is no room for mistakes and this can kill creativity when you are trying to achieve certain things. Lets take the example of pushing a Neve 1073 into saturation. This will overload a converter that is 24 bit or even 32 fix integer but a 32 bit floating point converter could actually handle the output at line level without issue and capture the output without errors or digital degradation. So loudness isn't something you think about in the analog domain as much as you would in digital because it analog more forgiving but once you add a fixed bit converter to the chain now you have a hard and fast ceiling of -0.1 dbfs before clipping and intersample error distortions. 32 bit floating point solves this issue of actual headroom above 0 dbfs during tracking. Hope that explains why I think 32 bit floating point converters are an important step forward. Cheers!
Thx for the explanation.
I mainly use the SD Mixpre6ii for voice. The levels are set in pre with the limiters activated in 24/48kHz. Saved a lot of processing time in post and a lot of HD-space.
32-bits came in handy when the sound of thunder and rain was needed and when I was filming on a construction site. Those situations were unique and it was handy to be better save than sorry. At that moment a time saver for shure.
Great video,
I shoot flying events, or just hang out at a local airport where anything from large Boeings to a Piper Cub can be out and about.
The 32 bot float on a Mixpre-6 Mk.2 is fantastic!
That is a perfect use for it!
The mixpre 32bit recorder has saved my butt more than once, I record live concerts and some times I don’t have time for a soundcheck, usually I can guess with good confidence, but sometimes it’s out of bounds! That 32 bit comes in reallllll handy in those occasions
Great vid. Just bought a Zoom H6essential, and this one of the few pro-32bit Float essays out there. I embrace the future.
Great video.
Once again, as with all processes in recording audio...it's "Horses for Courses".
I didn't realize there were only two practical levels these recorded at. The floating point format has 8 bits of exponent, so there is a LOT of unused range there. I guess they just wanted to use a format that was widely supported and well understood.
Very cool and helpful video, thanks!
It's interesting that half a year after you posted this video, ZOOM released their essential series 32-bit recorders with NO gain controls.
I jumped into 32-bit float (Tascam X8) because I'm a solo musician and record all my live sessions. I have enough to do without having to run a soundcheck and try to get optimum levels. 32-bit allows me to just hit record and not worry about if the recording is too loud or too soft. I was fine with 24-bit for years, but sometimes the levels weren't set well (I had to guess) and the recordings were way in the red, making them unusable. 32-bit gives me a usable recording as long as I don't overload the front end.
You can hear the distortion on the back end of transients for 32-bit float. Specifically zero in on the snare and hat sounds. They aren’t as crisp. This is a known issue with the dual ADC that was only recently patched by Sound Devices. I have not seen any sort of fix from Zoom yet.
This leads to a reason why many engineers are hesitant with 32-bit float. Even this video describes a signal chain with some sort of “magic” imposed on the signal. As it turns out, that magic was creating noticeable distortion outside the engineer’s control. Its great that SD has patched that issues but they’re still passing the signal into some mystical process the engineer can’t control.
How did SD patched that issue? Firmware update or?
Your killing me with that groove. You have demonstrated what everyone always says in these videos; what really matters is the performance' lol Great video man thanks!
Thanks so much! 😂
I use four (4) Zoom F6 recorders time-synched together to do live rock band recordings in 32 bit floating point. It generally works perfectly. I have, however, identified the need to carry with me a few XLR in-line attenuators to deploy when connecting certain sources - most notably any signal coming from a direct output on a guitar amp. I have had to apply as much as 20dB attenuation in some cases. Easy enough to catch before the performance starts but devastating if you miss it.
I am just an unpaid hobbyist. I was curious about 32bit float, so picked up a Zoom F2, then a Tascam X8, then a Zoom F8n Pro, a Rode Wireless Pro set. I love that I dont have to be worried about levels while recording, which is important when you are a one-man show, being the cameraman, the director, the actor, the sound guy, the producer and the PA. The only problem still is that not all software support 32bit float yet and even some that do will not handle it 100% properly. For example audio files from the Zoom F2 usually come out at too low a level and Davinci Resolve cannot push the volume high enough. I either have to normalise it in some other software before loading into DR or do some other tricks to sort this situation. But it is less headache than when you have 24bit audio recorded at a wrong level.
My DR 19 works perfectly with 32bit audio also as adobe premiere.
@neomatrix888 I have not checked how DR19 does it, in previous versions there was a limit on how much you can increase the volume level, and this limit was way too low for files from my F2. Files from the X8 were usually fine as in txt reecorder, as well as the F8nPro you do set volume levele into the ballpark before recording. But with the F2, I sometimes either had to set the volume level in some external software, or use a trick: amplify the track volume to max, then copy it onto a second track and amplify that again. Maybe they have fixed it since.
I’m resistant because I don’t want to have to buy new converters! But seriously, when I need to get new ones I’ll think about it and I follow the conversation with interest.
I don’t feel I have anything missing because I record 24 bit and I’ve only just switched up to 48kHz sample rate for the slightly less latency and, nowadays, greater compatibility.
But the idea of never (or almost never) having to concern myself with levels going in is certainly appealing. Saying that, I can’t remember a time when I’ve maxed the converters when recording since the days of the Mitsubishi X-880s we used to have, which were 16 bit, I believe. But we were learning and still knee-jerk pushing the needles to keep away from tape noise, even though that was no longer a problem.
Actually, now that I think of it, I was doing location sound as a favour for someone and I think we went over once or twice at the start. My inexperience in that field showing and I can certainly see how 32 float would be ideal for location sound or live music.
I hope they start releasing more interfaces with 32 bit float. Zoom has one but it's only 2 inputs
The use case I can imagine for 32bit is for when you need to record something that you don't have a chance to try out first to set your levels. If you're recording a jet flying by or a building blowing up, or a space shuttle launching, that might not be repeatable and you won't have a chance to check your levels.
But why not for everything? It’s the first time we have had available to us a bit depth that captures the entire dynamic range of audio. We went from 16 bit to 24 bit for some reason. I guess the industry just decided that that was enough.
@@simonpeck oh, for sure it would be great to have it just be the "final resolution". But if you're going to buy one piece of gear that has it, then field recorders is a good place to spend the money.
I'd like to know more about your dual F6 12 channel recorder setup and what you think about the results.
I love it when musicians talk about computer science.
As usual the marketing department chose to misuse a technical term because it sounded cool not because it was accurate at all for what it was describing and thus "32 bit float" somehow became an "audiofile" buzzword...even though this is all about sampling and compression. The real issue is everything is optimized for MP3 and Mp4 and all the 32bit float formats are proprietary.
How do you feel when Computer Engineers talk about music? 😉
While I agree that the marketing around "32-bit float" is a bit misleading, it's important to note that 32-bit float is simply a way to store a single numerical value within a computer. It's a standard format, not a proprietary one, and it has nothing to do with sample rate or compression.
Sample rate determines how often we capture a snapshot of the sound wave, while compression involves discarding data to reduce file size. These factors are independent of how the individual values are stored (whether as 32-bit float, 24-bit integer, etc.).
@@simonpeck As soon as you convert an analog wave to a digital format you've compressed it. You wouldn't be the first engineer I've had to explain how software and math works to.
The very first step in compression is the hardware that samples the acoustic wave and converts it into a digital format. THIS is what has been optimized for the current software suite usually destined for MP3 compression.
For example:
I cant split sampling to take the top and bottom ranges in a sample then take the mid ranges in the next sample and stack them in into split frequency sampling on the same ADC. This could easily double the range of 24 bit to 48 bit and completely alleviate clipping but most hardware is incapable of sampling like that even though they can easily support the 4x sample rate that would obligate...because the hardware is optimized for antiquated file formats.
Additionally: 32 bit float is tied to how binary systems store decimal places... that was impressive in the 90s for audio but even my shitty computer now has 32gigs of RAM; runs at 64 bit, and has high enough bandwidth to deal with 4k video. ..and .WAV files have been around forever and absolutely are proprietary to Microsoft.
The reason we're still using 24 bit is because, again, everything is optimized for MP3--and MP3 doesn't store decimal places.
@@TurboLoveTrain You're absolutely right that the first step in any digital recording is converting the analog signal to a digital one. However, this initial conversion doesn't have to involve lossy compression. If the sample rate is high enough (above the Nyquist frequency), the original waveform can be perfectly reconstructed. That's why we use sample rates like 48 kHz for standard audio and even higher rates for specialized applications.
While I appreciate your technical insights, I'm a bit confused by the alternative sampling technique you're proposing. Could you elaborate on what advantages this approach would offer over traditional sampling methods or 32-bit float recording? It's worth noting that most modern DAWs use 32-bit integer or even 64-bit float internally to process audio, providing ample headroom and dynamic range.
As you mentioned, modern computers are certainly capable of handling high sample rates and large amounts of data. This makes 32-bit float recording even more appealing, as it eliminates the need to set gain during recording and ensures that no audio data is lost due to clipping. This can be a huge benefit for many recording scenarios, especially those involving unpredictable sound levels.
Regarding the continued use of 24-bit, while MP3s are indeed widely used, it's not the sole reason the industry continues to rely on 24-bit. Many professionals work with lossless formats that can fully utilize the dynamic range of 24-bit recordings. Additionally, factors like established workflows, compatibility with existing equipment, and (surprisingly) concerns about storage space also contribute to the continued use of 24-bit. Interestingly, even high-end recorders like the Sound Devices 8-Series and Scorpio are technically capable of 32-bit float recording, but the feature hasn't been enabled in firmware yet due to some pushback from the professional audio community. This suggests that the preference for 24-bit is often more about tradition and established workflows than technical limitations.
@@simonpeck Lossless is also a marketing term. The act of sampling an acoustic wave into a digital format is one of the definitions of compression. Lossless refers to the fidelity of the digital file during type conversion--it does NOT refer to ADC input sampling of an analog wave. The file must already be in a digital format before "lossless" or "lossy" even comes in to play.
As for the sampling method I was alluding to it is actually based an old version of how to trick a computer display into mimicking higher resolutions while minimizing system load... I used it as one example of the significant limitations of the current, traditional, audio input pipeline.
Take an input stream and break it into two sampling steps. One sample step records only the upper and lower 1/4 of your amplitude range. The second sampling step records the middle 1/4 of the amplitude sample range. You then stack (multiplex) these into one output file with the equivalent of 2x your ADC's normal amplitude range (bit depth). That would be the advantage--you could almost double the sensitivity of your ADC while keeping file sizes relatively small. 1/2 the file can be disregarded as padding.
Because of "sampling theory" this means you would have to have a sample rate of 4x the highest frequency in the time domain (not 2x as is normal) to maintain fidelity... however input devices don't record like that, they usually use a continuous wave input method which means you can not clip and stripe the sampling that this would obligate. The hardware capturing device itself forces you into a limited system of wave capture methods which railroads you into limited file types.
So to put the pieces together: 32bit float was a method for increasing the bit depth of input from the ADC. This is A method for increasing bit depth and another way to do it would be to multiplex input sampling--there are other was to do it as will but it's all tied to how the input sampling is done (and that's still stuck in the 90s).
...also: please name a file format that supports 32bit float that isn't proprietary. Just because I'm not aware of any doesn't mean they don't exist. Audio isn't my focus but I do work a lot on compression and multiplexing.
For perspective: youtube, the second most trafficked site on the internet uses MP4 and most media released in the past two decades has been in MPx format... I'm not exaggerating when I assert everything is optimized for MP3/mp4--it's been like that for a very long time.
@@TurboLoveTrain I'm a retired software engineer (got my start in 1965), trained as an electrical engineer. You are either misusing the word 'compression' here (the word is used in different ways in different contexts), or else you just don't understand what you're talking about.
Great video!
I think a lot of people don't understand it. recording without manual gain is a win, setting the desired level in post is trivial.
my old Zoom H6 recorder has an option to record the signal twice, once with a reduced input gain. it recently saved my ass when the event was louder than expected and there was no time to check the level. 32f recording makes this a non-issue, but is more flexible while requiring the same amount of resources (2 gain stages and two ADC channels).
I look forward to all recording being gainless.
As a 16 year location Sound Dept. veteran (and hack home studio enthusiast), I can tell you that it is definitely the under paid post department that is to blame for the lack of utilizing 32-bit float. I get blow back for recommending the post department download the Sound Devices "Wave Agent" software, so that they can edit/recall meta-data or edit the mono/poly wave data...let alone, asking them to get into the weeds with editing 32-bit float.
Another issue with 32-bit float is "what else is in the signal path?". Your lavaliere's element (or diaphragm) can distort (unless you use the new DPA's) or your transmitter can over-modulate (maybe not if you have the latest A20s by Sound Devices). You'd have to REALLY plan out your signal chain to fully utilize 32-bit float....and at the end of the day, you're not getting paid any more money to rent this kit out...nor is there any guarantee that post will use it. It's not like the camera world, where RAW recording and LUTS were accepted with open arms. Sound Department is literally One-Half of the product, but is viewed as a minor annoyance.
It is going to be a LONG time before 32-bit float catches on. In the studio world, there still are no multi-channel audio interfaces with 32-bit float. Steinberg has "32-bit converters" on their interfaces, but they are not full on 32-bit float. It's like "we have a cure for cancer!" but the world responds with "meh....". Frustrating for sure.
Thank you for sharing this perspective. I’ve been very curious about that. It seems like manufacturers are waiting for demand at the professional studio level. Zach at Sound Devices said their pro level hardware (Scorpio and 8 series recorders) support 32-bit float in the hardware but they haven’t released firmware for it yet because of lack of demand. Meanwhile, the MixPre II recorders support it and the 10 can even be used as a 32FP audio interface.
Hi!
I am just learning about these mobile recorders and wonder how you would attempt recording a band, say drums first, then bass, etc..
Do you hear playback when overdubbing? What about different takes?
... Probably trial newbie questions. Your videos are 👍
brilliant thanks!
i think 32bit float is great. the only concern i have is what some people pointed out; pushing work to post can disrupt existing workflows. i work in radio broadcast as an audio engineer, and the recent introduction of izotope RX and supertone clear made for some pretty grueling experiences where reporters asked me to fix their shitty reverbant (and sometimes clipped) field recordings, because they know we can do that, to an extend. if we had 32bit float some issues we have now wouldn't be there, but i feel like reporters would care even less about standards that make sense beyond the old tape noise floor logic. having the tools to fix things in post is nice but doing it right from the start is often very easy, especially if the workflows for it are long established. if i end up freelancing as a videographer again, i would probably consider using 32bit float, because i'm on my own and i save time by fixing it in post. at my job, there is no time to fix it in post because it's going on air in 5 minutes. different workflow :D
Thanks for sharing your professional perspective! It's definitely true that 32-bit float could disrupt established workflows since it requires normalization in post. However, it could also save substantial time by eliminating the need to use RX to fix the situation you mentioned with clipped recordings (unless the mic itself is overloaded). With 32-bit float, your effective gain setting happens in post, which might lead to higher quality recordings overall.
While normalization adds a step, it might replace much more time-consuming repairs for clipped recordings in your fast-paced environment. I understand your time pressures in broadcast. Do you think there are potential workarounds or workflow adjustments that could make 32-bit float viable in a professional environment? The human element is always the hardest thing to change, especially where it crosses roles and responsibilities. I'd love to hear your thoughts.
If you export your 24 bit 32 bit float recording example to Soundcloud in 24 bit wav format i think we probably will hear a difference. But You tube audio quality MP4 is often about 192 kbps MP3. What do you upload?
Yes, TH-cam sadly is not the best place for high resolution audio comparisons. I could post the files elsewhere for comparison...
@@simonpeck keep me informed!
Great video. I'm going to say the problem with 32-bit is that we lack a device that can handle say - 12 channels. Presonus new interface has 8 channels at 32, which is great - but then I have a bottle neck with all my outboard ADAT gear. I felt if Presonus made two interfaces that you could mate for a total of 16 inputs at 32 bit float.
Thanks! The Sound Devices MixPre 10 has 8 XLR Channels. I have two synced together with a BNC cable for timecocde, providing 16 XLR channels and four additional unbalanced channels. They physically connect together so you can stack as many as you want. The Scorpion has 32 channels.
@@simonpeck Hey this is good information. You need to help these guys sell this stuff!
Yeah, I'm a huge fan. It's all really solid well-built equipment. I'd like to make a video about linking up multiple recorders and recording a live session.
Boss/Roland have been using 32bit float in the their effects pedals for at least 10 years
Quantization noise doesn’t exist in fixed point bit depth formats. It exists only in float point formats.
Quantization is just the process of division by two. This is what every resistor does in the comparator chain of an ADC. The remains are rounded and there's nothing to worry about, as the range they are responsible for is below the dynamic range of the bit depth.
So 24 bits fixed doesn't introduce quantization noise. The noise we have at - 144dB is just phase noise.
But in floating point formats we deal with exponential numbers, so they are much less accurate, than fixed point. Here's where we'll find quantization noise right above 0 dBFS and far below -138 dBFS. Again there's nothing to worry about, as all these inaccuracies would always be far beyond human's hearing range. And even if we amplify them we will hear subtle changes in volume of every phase of the signal, which are so tiny that they can never be a problem in comparison with the tolerances of the analog chain components.
Shockwave and 194 dB are not connected with each other. Shockwave is the product of Doppler effect and the speed of sound. 194 dB peak (or 191 SPL) cannot be exceeded due to atmospheric pressure, which is equal to 1 bar or 100000 Pa. If we rise the pressure level, we could exceed the maximum amplitude of a sound wave.
Dither is not used in 32 bit converters. Two parallel signals are analized, then divided into portions which are picked in turn and merged together when they fit the parameters.
Its crazy that there is any resistance to 32 bit. The benefits are so obvious
Simon:
Good piece! I learned a couple of things.
At 13:21, you said “Obviously the 1528 decibel dynamic range claimed of these recorders is excessive, since anything over 194 decibels is considered a shock wave.” You’re confusing two entirely different things with this analogy. Dynamic range is the difference between the highest level you can record without clipping and the level of the noise floor. That range compares one level to another and is expressed in dB. The 194 dB figure you quote is a sound level, not a range, and it is more properly thought of as 194 dB above the Threshold of Hearing. As such, it is expressed as 194 dB-SPL. The SPL stands for “Sound Pressure Level,” and it is a specific standard level, 0.0002 dynes/cm2, that is the minimum level detectable by humans. When using decibels compared to a reference level, the reference is always indicated, hence the “dB-SPL.”
In this case, we don’t know the signal level of the highest audio signal we can record on a 32-bit floating-point recorder, but what we do know is that the lowest signal possible, at the noise floor of the low-level ADC, is 1528 dB lower (i.e., there’s no reference level). It would be perfectly valid, for example, to set the clipping point of the high-level ADC as 0dB and the noise floor at -1528 dB.
That said, there is a fallacy that’s being overlooked here. While the recorder has a theoretical dynamic range of 1528 dB, the program (i.e., the signal that’s being recorded) will always have a signal-to-error ratio of approximately 141 dB. The reason for that is that it’s limited by the 24-bit mantissa. (24 x 6dB/bit, less 3dB due to quantization error of the least significant bit, which occurs 50% of the time.) Another way of stating this is that while the recorder will automatically adjust the gain up or down (using the eight additional floating-point bits), the noise floor can never be lower than -141 db below wherever the peaks are because whatever you record has only 24 bits to describe changes in the waveform.
Essentially, what I am saying is that a 32-bit floating-point signal IS a 24-bit signal that can be freely rescaled without clipping or added noise. This is why the 32-bit float and the properly-recorded 24-bit signal sound identical. There is no difference in sound quality.
The question I am not hearing addressed by proponents of 32-bit floating-point recording is why one would increase their data storage needs and the time needed for transmission of that data by 33% for no perceptual increase in audio quality? A 3-minute stereo audio file at 48.0 kHz and 24 bits is about 50 MB. A 3-minute 32-bit floating-point recording at 48.0 kHz is 66 MB. They sound identical, so what am I gaining by going with the larger file size?
The one advantage I am getting by recording with 32-bit float is that I don’t have to set levels, and even when the signal is recorded too hot or too low, I can change the level in post-production.
Cool, but guess what? We can only hear about 60 dB at any one time. The Signal-to-Error ratio of my Audio CD is about 93 dB, which means I can never hear a CD’s hiss. No engineer will tell you this, but you could set your maximum peaks on a 24-bit recording at -24 dB. Your noise floor will be around -141 dB because it’s 24-bit. Then you go ahead and give the recording +20 dB of gain (try it!). Now, your peaks are at -4 dB below clipping, and your noise floor is raised to -121 dB. That’s well below the noise on a CD and still completely inaudible. In fact, because 24-bit has eight more bits than 16-bit recording, it has 48 dB more dynamic range (8 bits x 6 dB/bit); you could bring any 24-bit recording up as much as +48 dB before you would match the inaudible noise floor of a 16-bit CD. Of course, in practice, you wouldn’t do that because you’d also be bringing up any noise generated by your mic preamps up by the same amount of gain.
There’s nothing wrong at all with 32-bit floating-point recording; it simply strikes me as unnecessary. You can accomplish the same thing by recording in 24-bit with peaks between -12 and -18 dB (or less), giving you an additional 12-18 dB of headroom as a cushion for unexpected peak levels, and later applying 12 dB of gain (or more) across the board. You would still have to properly gain-stage in the analog domain prior to conversion, but there’s no reason why your 24-bit recording level has to be hitting -3dB. We’re not recording on analog tape.
Having said all that, your conclusion is spot on. I 100% agree with your comments about the ease of usability of a 32-bit recorder like the F6. The fact that you can record drums without a computer and never worry about recording levels makes it a great tool for your application. In a studio setting, where an engineer has greater control over levels and chaos, and is recording directly to a computer, the advantages may not outweigh the additional data overhead.
I’ve always worked in 24 bit and I’ve never had a problem when I’m in the studio alone.
BUT
I often record people that have no to little experience speaking or singing into a microphone and I’ve had to do many takes over because someone thought when they suddenly scream they have to get closer to the mic to accentuate that -- and whoops ruined… I assume I should try working in 32 bit (considering to buy the SSL 12) to enable a 32 bit workflow…
You're absolutely right that 24-bit recording can be more than enough in a controlled studio environment. But the challenges you describe with inexperienced performers are very real.
While the SSL 12's 32-bit fixed-point format offers a huge dynamic range, it doesn't eliminate the need to set gain before recording. If you're looking for a way to completely bypass the need to set gain during recording, eliminate potential clipping errors, and have a versatile audio interface for your studio, the Sound Devices MixPre series might be worth exploring. These recorders offer 32-bit float recording and can function as high-quality audio interfaces.
when will get to 64-bit double precision?
what do you mean by audio normalizing? Thx
Normalizing is the process of adjusting a waveform's loudness level after recording to move the peaks a specified distance from 0dB. The way "peak" is defined depends on the algorithm used, and there are many ways of calculating it, to name a few:
• the root-mean-square method (RMS)
• loudness units full scale (LUFS) over a short duration, longer duration or the full track.
• the actual momentary highest peak of the signal
For mixing songs, individual tracks are often recorded so that the body of the waveform is around -18dBFS with the peaks a little higher. There's a judgement call here, because when setting levels, intuition is often used to gauge what "body" means. And you don't always know exactly how loud a performance will be compared to a sound check. After recording, you can normalize a track if you want to correct any mistakes made during the recording process, though this can be destructive (unless you record in 32-bit float). Alternatively, you can adjust the faders. Then, in mastering an overall, integrated loudness level for the full track can be targeted. For TH-cam, I target -14 LUFS-I for the master per their specifications.
Why doesn't SoundDevices more expensive 8 series recorders offer 32 bit float as an option?
I asked them that and they said the hardware supports it. Lack of customer demand has prevented them from including it in the firmware. As I understand it, on the post production side of audio it has been negatively stigmatized and probably misunderstood. And I think there’s concern that recording professionals will be more cavalier about setting levels, passing that work onto the post house. But I suspect we’ll see it eventually. The industry is slow to change.
They’re about to when they release the new 10.0 firmware update. Just as I suspected it wasn’t a hardware thing for the 8 series given their earlier released MixPre series were already able to do 32 bit float. It was some perceived unprofessionalism or bureaucracy in the types of people using their flagship products. The Kashmir preamps have been able to do it all along. I will be happily upgrading to the 10.0 firmware in both my 888 and Scorpio when Sound Devices release it. In my humble opinion, the big obstacle is post production being able to effectively/efficiently use the file type or the post sound supervisor potentially inferring through some passive aggressive communication to the powers that be, that they hired a lazy production sound mixer.
With the zoom, which I believe does not have a gain control, how do you deal with the noise floor of the preamps themselves? I would think that on a much louder source, turning down the input gain would net you less noise induced by the analog part of the preamps themselves into the 32bit digital recording. Even with two separate preamp circuits, that is two different gain levels, and a compromise on the analog side, I would think? Especially if you are using a microphone like an RE-27 that requires a lot of gain to get a usable level out of it, about 60db of gain on an 1173 with a singer right on the grill, but less if you are using it on a floor tom with a heavy handed drummer, or a kick drum, etc.
I'd have to say that I don't think that noise is a problem with it. Although, I haven't conducted an extensive analysis, I've used the F6 with low-level spoken word and cranked the gain up with headphones attached with various microphones. The only electronics noise I've noticed is with a tube mic. The F6 is really geared for dialogue, so I think they designed it to perform well at lower SPLs. I think their target market is filmmakers, so when I record drums with it, I can clip the analog circuitry of the thing. (I recently acquired a MixPre 10 II and have not been able to clip the inputs, even with close-miking on the kit.) I recorded and wanted to include a noise example in this video, but I don't have access to an anechoic chamber and the quietest space I could find had room noise much louder than the internal noise of the F6, so I had to leave it out of the video.
While there are likely two different noise levels between the preamps, it's relative to the signal. So the higher noise level of the high gain side would still be way below the signal that's being captured from that side.
@@simonpeck Cool, Thanks for the reply my good man!
lower noise floor and no clipping, but you pay for that with audiofile sizes. prolly make sense for live music, especially something with very quite parts. for in the box edm stuff 44.1/16 is plenty enough
Can you explain the normalization advantage? In traditional DAWs you just click the normalize button and that’s it so I’m wondering how it gets easier. I understand how it gives you more flexibility, but not how it makes it easier to normalize or faster.
What I meant was that in my experience it’s easier and less time consuming to normalize in post than to set the gain perfectly when recording, especially in a multiple microphone scenario.
32 does sound a bit more dynamic (open)
A 32-bit recording would still clip in a live, monitoring setting? It’s just that you can recover the recording in post later.
I have Behringer XR-18 in the details I read it says 40-Bit floating-point DSP features “unlimited” dynamic range with no internal overload and near-zero overall latency , is that correct if anyone knows!
Thanks
The key in there is that the internal processing is 40-bit. Meaning within the internal processes (EQ, comp, FX, routing...) it should be impossible to clip etc. Not meaning that the input side and A/D conversion is floating point or infinite DR. I am sure you will add noise or clipping with improper gain setting on XR-18. It would be misleading to exactly call these devices 40-bit floating point devices.
Why won’t it catch on?
Does it sound worse? No
Does it sound better? No
Will any of us ever record anything with more than a 90db range? Probably not.
Save time by not setting gain? Combine your process of normalizing with 40 tracks, many with multiple takes, over a 12 song record, and I bet getting it right at the front end is quicker.
Still, how am I supposed to send a decent headphone mix and monitor what I’m recording without proper gain ?
You almost get one of the most important parts about the actual sound at the end. Using an API (or any analog pre) will require you to adjust the gain, and how you adjust that gain has a significant impact on what you get out of that pre. Studios aren’t going to give all their preferred pre amps even if the above mentioned problems weren’t a thing.
This set up is nice for throw and go, but it’s kinda solving a non existent problem.
Well... that's kind of my point. It sounds the same, yet it can offer recovery from mistakes and save time in some cases. For me, normalizing 40 tracks takes the same amount of time as 12; select all, normalize. I suppose there could be some clips that would be sliced up, depending on what you're recording, but... let me flip it around. There's nothing preventing you from setting levels when recording in 32-bit float (at least with the MixPre and F6). It still sets the peaks relative to 0dB, even though it doesn't change the recorded waveform. You could work exactly the same as when recording in 24-bit, with the benefit that if you didn't happen to nail the level perfectly for a track at record-time, it could be recovered it in post.
Another way to think about it: if you had started out recording in 32FP would you be actively seeking to switch to 24-bit?
For headphone/monitor mix, you could still set the knobs as you have in the past, or if you're using Sound Devices, you could use Auto Mix: www.sounddevices.com/automatic-mixing-101/
Regarding an external pre, I didn't go into it in detail, but that's exactly why I want to be able to use one. I have used an AEA TRP2 into an F6 and was able to clip the analog portion pretty easily. I haven't tried that with the MixPre yet, but I'm certain it will be able to withstand a more preamp gain.
@@simonpeck I can’t even imagine the clumsiness of tracking a band and a trying to get good headphone mixes to 4 or 5 people with an unknown gain structure, then trying to do a punch in with an unknown gain structure tweaking volumes to give suitable working headroom after the fact, trying to integrate outboard gear, doing punch ins, etc. seems like a pretty messy way to work, nor do I ever want anything at at 0bbfs.
Why wouldn’t I bother using 32bit float? As stated, integration with the analog world. I do have the capability of recording at 192khz but never do. Taxing the CPU load with no benefit isn’t in my nature. It’s so, so easy to set a good gain structure, replacing 32 channels of AD and DA converters would be a pointless expense.
Again, seems a fine way to do what you’re doing, but I wouldn’t expect it to become a standard in the pro world anytime soon.
@@zagatoalfa No, this is way overthinking it. It isn't that complicated. It's just a data format. So somebody could walk into your studio and change your recording format to 32FP and you wouldn't even know it. You would do everything exactly the same. The only difference is that if you happened to make a mistake or didn't set the levels perfectly for a channel, you can correct it after the fact with no negative impact on sound quality. That's it.
This assumes, of course, that your interface supports 32FP.
The scenario described -- where the position of the gain control knobs is totally ignored is not a requirement when using 32FP.
@@simonpeck exactly. I could run my sessions at 32bit and not notice and setting gain isn’t that hard, so I won’t bother running 32b float. I also don’t think considering implementing the technique of not bothering to set gain in a real world situation is over thinking it. A quick search shows studio based 32bit float A/D or D/A conversion. Only Sound Devices and Zoom field recorders, despite the tech being around for some time.
When I record my acoustic guitar with the Zoom H4 essential (about 12" from the neck/body joint) and bring the Zoom file into Reaper, the signal level is so low I have to boost that channel to its max (+24db) to get peaks anywhere near 0db. Is this normal, Am I missing something?
What format were you recording in? If 32 bit float, you can normalize it in Reaper. If not, you need to make sure your levels are set appropriately when recording. Are you using the built-in mic?
@@simonpeck Yes, 32bit with the built in mics. Near the end of your video I saw how to normalize the levels in Reaper, and set for 0db peaks i get 30db+ gain bringing it up to a usable level.
Sadly there's no market yet, no big player stepped into this.
Probably because they think it's cheaper if you're careful not clipping the input.
A good comparison would be a camera's focus (or f-stop/ISO/shutter for that matter).
Imagine shooting a 30 minute interview with the camera out of focus. Imagine being able to pull that focus in post. That's what 32 bit float can theoretically offer.
Most professionals manually pull focus in the same way audio engineers gain stage, There's a specific person that solely operates and pulls focus and they get paid LOTS.
Most resistance to utilizing auto focus was purely due to it's poor and inaccurate implementation, usually resulting in undesired, adverse, effects.
People need to experience 32 bit float in order to trust it, in the same way you would need to experience and play around with good quality, modern, auto focus.
Auto focus works well for people who don't have the in-depth knowledge or patience to dial in exactly what they want, 32 bit float satisfies that exact notion.
Someone who's been working in audio for ages would have zero need or use for 32 bit float because they know exactly how to capture what they want, the same exact way a photographer uses manual mode because they've learned exactly how and why to configure the settings to accommodate the desired outcome.
Technology like 32 bit float audio recording is such a massive step, that it's hard to even justify the amount of freedom it could restore to an artist. We've grown so accustomed to only being able to capture a hollow representation of audio and that it's been proven time and time again that any step further gives us diminishing returns.
So despite the heavy resistance in standard use cases, 32 bit float can and should thrive within it's own, separate, classification of audio capture. The same way we have stereoscopy, or 360 action cam videos.
At the end of the day it's just simply a different way of representing captured information, and therein lies the root of the problem; since it's a different way of representing audio, the thought process revolving around it should also be reconsidered.
Woody, I'm thrilled to see this post. This is an excellent analogy. I used to work in the film industry and have witnessed professional camera assistants pulling focus first-hand! Interestingly, when I interviewed Zach, he mentioned the comparison to light field photography where you can focus an image after the fact. I cut that part out of the video, but perhaps I could post the uncut interview if there is interest.
The other problem with autofocus is that it's hard to use when you want to pull focus in a certain way or keep some things slightly out of focus. Also most AF systems can't be programmed to keep two points within the DOF or go slowly between A and B. The closest I can think of doing something like this are the broadcast servo lenses where you can control the focus servo speed. Decades ago Canon had a two point DOF AF in some of their film cameras but IIRC this feature was never carried over to the digital models and it was not a very good implementation anyway.
Hello. Blind guy here… Not setting gain his music to my ears! I am hearing that zoom is making a recorder that is accessible by a blind guys… Right!
We don't need that Kashmiris Sarcasm to be floating around to be sure of what we're doing 24/7. I don't know, I never had those extra hours without taking flights. Namastè.
Equipment shut have opition to listen to it any formate 16 to 32 bit
I am confused, just casually saying to normalize the audio like its nothing. Isn't normalization similar to compression? Its bringing up the quiet parts and bringing down peaks? Thats going to change the sound? If it didnt change the sound then it would just be gain and not called normalization.
Normalization measures the difference between the loudest peak and the desired level of the loudest peak, and adjusts the whole recording by this difference to make the peak match the desired level. Either the whole recording is turned up or down, there is no difference for quiet or loud parts, so normalization is not similar to compression.
The question of whether it has anything to do with Led Zeppelin depends on how you spell it... Fabric? or Geographical Region? :)
That's why I asked! Sound Devices spells it "Kashmir." 🙂
@@simonpeck AHA! Cool man.. I didn't notice if you mentioned the spelling in the video, but I have the attention span of a gnat when I am listening at work lol.. This was very informative and answered quite a few questions I had about 32 bit recorders, but was too lazy to research! :)
Hmmmm for me is simpler than that, the highest resolution in Logic is 32 bit, so is like 11, it just goes to 11, simple 😊
i think 32bit float sounds so much better for some stuff at least
It is different
320 bits 32bit
The brave thing would be to make recording devices with no gain control whatsoever.
The gain control doesn't set the "zero point" but the "one point": you're deciding how much dynamic range is between the number zero, and some other number that is not zero.
Also, your introduction of the term "dithering" here is inappropriate, it has nothing to do with the range selection/combination process.
Also, having more than one ADC is not strictly necessary, and for all intents and purposes 24-bit integer PCM is also fine for gainless line level ADCs... and in practice probably also fine for gainless mic preamp-ADCs, but there are extreme outlier microphones for which this is not the case.
These devices effectively don’t have gain control when recording in 32FP. No matter how you set the gain, the exact same waveform is recorded. It only defines how the waveform appears when you initially import it into the DAW. I imagine it’s some metadata stored in the wav file. Whether it’s called a “zero point” or “one point” is a matter of convention, as I understand it.
I could be wrong about whether or not dithering is used in combining the data from multiple ADCs. That was speculation based on research I did on possible implementations. Sound Devices doesn’t reveal their implementation details.
You’re right that they could use a single 32-bit ADC, but they don’t. From what I read on these converters, there are technical challenges in doing so accurately. I can dig up the source if you’re interested.
At any rate, it was their design decision to use multiple 32-bit converters so I trust they did that for a reason.
The most popular microphone in the world is the Shure SM57/58, which has a max SPL of 180 dB, so it wouldn’t need to be an outlier microphone to produce a wide dynamic range.
@@simonpeck they do have a gain control though, the knob is there, that's what I mean.
As for calling it the zero point... That makes no sense, setting the zero point is setting the DC offset, not the scale...
Also, it's not necessarily a "32-bit ADC" that you would use, since ADCs in production do not produce floats regardless (though theoretically there is a way to design a self-ranging ADC that produces floats). If it's a 32-bit ADC today, it's still only producing integers.
As for their decision to use several ADCs, that will depend on their particular architecture, but it is not necessary in principle. I know the limitations on single ADCs in these applications.
@@microcolonel Ahh, yes, these do have knobs, but there are 32FP recorders that do not. Setting the zero point refers to setting how the peaks will appear relative to 0dB, not DC-offset.
When I mentioned 32-bit ADCs, I was referring to Sound Devices' implementation, not a theoretical one. It's what they use. How they convert the output of those ADCs into 32FP is proprietary.
24 bit integer precision makes zero sense for listening to music. 24 bits are only needed for writing music because you can record your signals low and then boost them without boosting the noise floor caused by imprecision errors. You need to record your source at a very low volume, like barely hearable, much lower than whispering, for 24 bits to start having issues - but why would you even do that?
24 bit precision is 256 times more precise than the default 16 bits and a finished track will be indistinguishable between two formats.
32 bit float precision is also quite questionable because floating point numbers can be very imprecise and result in errors. To me the format always seemed like an internal way for very old DAWs to work around the hard limitation of very old audio interfaces that could only do 16 bits at a cost of extra CPU power.
It depends on what your goal is. It makes total sense if you want to save time and reduce the possibility of human error while recording.
Floating-point rounding errors are inaudible, because we’re talking about extremely small differences in amplitude.
@@simonpeck Most plugins process in 32bit float, if rounding errors were an issue we'd know about it by now.
It all boils down to: digital gain vs analog gain.
What do you mean by “digital gain?”
@@simonpeck Well, I mean gaining it up/down, digitally, in post...vs gaining it the "traditional" way - pre quantization. :)
2024 shut have all type up to the people if won't to listen to it
Auto Align 1 is great on drum tracks. Auto Align 2 is even better. th-cam.com/video/K3DOCTx_qS8/w-d-xo.html Of course wav 16, 24, 32 float or 64 float bit ...it is an upgrade but these things like Auto Align give a lot of depth and sound to the drums that is far beyond the improvement going from 24 to 32 bit float. Phase and alignment is important/essential for drums and for a mix.
🙂👍
Well? That wiring scenario. Will cause your speakers. To mimic, the motion of the microphone, diaphragms. And which way are those microphone diaphragms moving? When a sound pressure wave, blasts into the microphone diaphragms? Which way is that microphone diaphragm moving in? That's right. It's moving in. It's being compressed into the body of the microphone. As the sound pressure wave, smacks into it.
And you absolutely positively, don't want your speakers. Moving like your microphone diaphragms. Now would you? Isn't that kind of backwards? Shouldn't the speakers be moving in the other direction out toward you? Because they are not. They are sucking inward, away from you. The speakers are sucking away from you in front of you. They are not protruding out, toward you. To provide you with a Positive Pressure Wave Immersive Experience. No. It's sucking sound away from you. And it sounds all wonky and crappy. Because it's in Negative Absolute Polarity. Unless you turn both wires around on both speakers.
As connecting the positive terminal on the amplifier to the positive terminal on the speakers and the negative terminals together. Isn't the right way to connect it. No. Positive does not get connected to positive. Positive gets connected to negative. So current can flow. And negative gets connected to positive. So current can flow. Because current cannot flow. When positive is connected to positive and negative is connected to negative. Current cannot flow in a DC circuit that way.
So the electrical engineers know, DC Theory 101. Positive gets connected to negative. You are not jumping a dead automotive car battery. When you connect the positive to the positive. And the negative to the negative. In that parallel wiring. That's to jump a car battery. That is not how to connect speakers. Where positive gets connected to negative.
Because the electrical engineers that have designed amplifier circuitry. And the speaker itself. Had to follow DC theory 101. With positive getting connected to negative. So current can flow. And then when they get to the output of the audio power amplifier. They go against that wiring? And they connect the positives up? Why that inconsistency? Since that is not DC theory 101. What's the matter with them? Why would they change at the output? And cast their DC theory 101 out the window? But they in fact, have. A gross inconsistency. In technical engineering theory. One of those what the FUCK, moments. That makes no sense. No engineering sense.
The reason for this anomaly, inconsistency and downright incorrect way of wiring your speakers. It is due to, grossly ignorant fools. Who have demanded. From the manufacturers. THEY DON'T WANT ANY OUT OF PHASE ANYTHING IN THEIR SIGNAL CHAIN!
Yes but what about where the polarity has to be inverted to obtain Absolute Positive Speaker Polarity? At the output of the audio power amplifier to the input directly to the speaker. Which is the only place. Where Absolute Positive Polarity can be technically Established. Polarity cannot be inverted. At the output. By simply inverting the phase of the waveform, at the input of the audio power amplifier. That does not change the absolute polarity at the output. Which will remain the same. And nobody understands this. Or what is occurring? Because they think they have their wiring right. They refused to listen to it the other way. Because they don't want to appear, stupid. So they shun. Even listening to it. Where you will know, instantly. You have been always listening to it the wrong way. For many years. From the get go.
Nobody ever teaches anybody this. And nobody can think this through, for themselves. Though JBL did. Back around 1948 I reckon. It was certainly by 1968. When I first came into that scene.
Now I have fixed and corrected a lot of big expensive, well-designed, recording studios. That just never sounded right. Because of miswired speakers they thought was right. Because of ignorance and stupidity. And the ability to exercise one's, gray matter, synapses and neurons.
So if you flip the two wires around. On the back of both of your speakers. You will then force your speakers into Absolute Positive forward Polarity. And you will hear a whole new listening experience.
For those of you. With those fine, Active, self powered, Speakers. That you also have to plug into an electrical wall outlet. For each speaker. You are going to require a, screwdriver and soldering iron.
You will remove both the woofers and the tweeters. You will use your soldering iron. To reverse the connection. Both on your tweeter. Then on your woofer. Then you will reinstall the speakers into the box. Do not over torque down those screws. You can warp the speaker frame if you do. Just make it snug. No Hulk action death grip tightening. No. Just make it snug.
This will correct your speakers into Positive Absolute Polarity. You will now hear and experience a vast difference. Like you've never heard before. It will sound the way you always thought it should. But never quite, did. Now it will.
This is all part of my Control Room Design Concepts. For a Positive Pressure Wave Control Room Design. I make certain of that. I'm known, for that.
And then your lousy acoustics in your room. And we both know they are lousy acoustics. Will seem a lot less, lousy. They may mostly disappear. They may sound different. They may not be as wonky and resonant. It changes the whole feel of your room. It's not like you've heard it before.
For those of you fortunate enough to have chosen, JBL Speakers, for yourselves. You are good to go. Do not change anything. Do not rearrange your speaker wiring. They were designed to be, Correctly, Miswired. That's right. JBL actually did that. So you would be hearing their speakers correctly. As they marked the positive and negative connections, backwards. But they aren't backwards. They are correct. To ensure. The speaker will protrude out toward you. To provide you with a Positive Pressure Wave. Surrounding you. Enveloping you. Sending you two new listening heights.. You will hear height. You will physically hear the lead singer. Standing at your height. Just feet in front of your face. Where you can hear their voice. Coming out at you. Right there in midair in front of you. It can seem a little eerie. A little strange. Kind of weird. There is a voice hanging in mid air in front of you. You can almost reach out and touch. You've not heard it this way before. You will hear stereo like you've never heard stereo. It will have a very wide stereo soundscape. Wider than you have ever heard it, before. It won't matter where you sit or stand. You will hear glorious stereo.
The other strange psycho acoustic phenomenon. The timing of everything arriving at your ears. Will be almost precisely. The way the producer, engineer and artist. Wanted you to hear their creation. Their work of art. That way. My way. The JBL wiring way. That all other speaker manufacturers in the world got backwards. Beyond stupid, clueless and totally ignorant.
You haven't lived. Until you've heard Positive Pressure Wave wired monitor speakers. You have not lived.
RemyRAD
Yeah but this one goes to 11
DAWs use 32 bit internally anyway.
It is not about the file format. It’s about the fact that there’s no need to set levels when recording.
40 bit float make new system 32 bits
The best analog stage is barely 20 bits (122 dB) signal to noise ratio, the claim of 1528 dB is complete BS. 32 bit linear is theoretically 194.4 dB.
You’re missing the point. It has nothing to do with a dynamic range increase. Rather, it’s about optimizing the bit stream so that the 16 or so bits actually needed can’t possibly go unused. Two preamps with very different sensitivities permit encoding of so broad a digital range that the full required range cannot fail to be encoded and be decodable.
@@artysanmobile I'm not missing the point, the claim is BS, 32 Bit is capable of recording 194.4 dB, that cover everything from a pin drop to an explosion. A jet fighter on full afterburner at 10 meters is 150 dB, that's more than enough for any situation without making a ridiculously false claim.
@@hwirtwirt4500 Yada yada yada. Glad you’re entertaining yourself.
@@artysanmobile If you don't understand the underlying technology stay silent and not expose your ignorance.
Releasing 32bit was a smart marketing ploy to generate sales, and is still primarily being marketed towards videographers (and anyone) who ever distorts in 24bit, (and even 16bit under normal real life recording, mixing, playback), which indicates they don't know how to gain stage yet. If someone says they need 32bit because they can't be there to monitor levels does not understand proper gain staging. Even a rock concert can be recorded on stage perfectly fine in 24bit and even 16 bit...because it's not about max level, it's all about DYNAMIC RANGE.
-Not allowing the sound engineer to adjust gain is like putting a race car driver in a self-driving car.
-Bit depth does now equal sound quality, PERIOD
-Bit depth = theoretical dynamic range PERIOD (some manufacturers quote specs that are not possible in even the coldest temperatures on earth. Also specs are quoted that are component based (in theory in space) and not based on full circuit design.
-Human pain threshold (frequency dependent) starts around 120dB
-An electric drill at 3-5 feet is approx 96dB
-Loud motorcycle at 10-15 is approx 96dB
-Rock concerts measured front of stage 110 -120dB
-An extremely loud scream at 1 foot away is approx 120dB
-Full stack guitar cabinet (ie: 4 12-inch speakers in a 4x12 cabinet) at 2 inches away is approx120-130 dB
Max dynamic range available on BlueRay and all 24 capable streaming services is 144dB
-32 bit is for those recording levels that exceed +/- 144dB
-Professional mic can handle spl often up to 140-160dB spl without distortion (for use touching the guitar cabinet grill) Can and should, can and do are all different things.
-A Boeing 747 at 6' away would be somewhere in the 150-160 range
- instant deafness occurs around 160-180 (very close explosions/gun shots)
Earth's atmosphere pressure (at sea level) limits us to around +/-194dB SPL ,
This means it's impossible to produce sound levels louder than that, because you can't have less than a vacuum in terms of pressure. Pressure beyond +/- 194dB is no longer sound, it's shock wave. (ie:explosions)
Death would occur instantly at 240-250dB, because again, it wouldn't be sound beyond 194dB, it would be you standing in an explosion an d dying from a shockwave.
PUNCH LINE: 32Bit floating point recording has a theoretical dynamic range limit of 1500dB.
I challenge ANY manufacturer to demonstrate how 32bit is "better for the music and recording industry", vs a money grab, for all outside of scientists and sound designers and OSHA etc.
It's not adding quality, it's simply adding more seats to the bus for potential loud volume. Now instead of taking a family trip in a 144 seat but, the industry is forcing a 1500 seat bus that messes with the audio levels I'm getting from TV and film video editors. What a mess!
Instead of doing the right thing and educating the market how to actually use their devices, they reboot their entire line with designed to record more dynamic range than anything anyone will ever want to subject themselves to.
If ANY manufacturer disagrees, I will very gladly produce a nice 32bit recording taking advantage of the technology and provide a listening room for the executives to gather in for a full dynamic range listening experience to see what they've so generously provided the industry.
Marketing departments are advertising 32bit as an "upgrade" in sound and the best the industry has to offer etc etc etc.
Misleading and false, PERIOD
Happy to discuss with anyone who genuinely wants to get past the marketing smokescreens and the "tech support" and company representatives who have "pro sound" experience who ever say you need 32bit capability because it will improve your sound.
When talk of preamps and/or any other part of the device is discussed along with bit depth you can immediately know the person actually doesn't understand the function of bit depth.
Bit depth are the seats of the bus and have nothing to do with the quality or capabilities of the engine of the bus, or the tires, or the radio, or the windows or the seat belts, or , or , or....., beyond how they are given the info from the audio path PERIOD.
-----
28 years in the professional audio industry.
TV, Film post, sound design, film composing for 3D4D motion ride , immersive audio (incl Dolby Atmos) for streaming and cinema.
Music and audio systems consultant and designer/integrator.
More than 850 hours of Direct face-to-face education and hands on experience and side by sides coming from a broad majority of the worlds top audio equipment manufacturers producing AD/DA and sound recording devices, mixers, monitors, amplifiers and processors etc.
Not needed at all, and to my surprise it even defeats its purpose, taking into account what Zach said. LoL. Besides, those ZOOM preamps are really a piece of crap, they re really bad man
Well? I worked in Pro Audio Manufacturing. Starting over 40 years ago. And I'm what you call a Studio Fixer. As some of us have, well seasoned, extremely acute, superior, hearing apparatus. In other words. I hear things nobody else hears. And I'm not an audiophile. I don't believe in, audiophiles. They haven't a clue what they're hearing or what they are listening to. Neither, do you.
Yeah… In a double-blind listening test. With over 30 other well seasoned top engineers. Apparently. I'm the only person. Who could tell the difference. Between an original source recording. And that of, 24-bit, 192 kHz, high-end, PCM, digital converters. Yes, it's true. Everybody else got it wrong. What did they get wrong?
Everybody else thought. All of what I heard as artifacting. Was the original source material. But it was all artifacting.
This made me realize. In all of the other high resolution digital audio demonstrations. None of it sounded like original source. None! It was all loaded with artifacting. At any bit depth. At any sample rate. It continues to sound all, fizzy and grainy as PCM. Because it is, PCM. A flawed digital format.
So recording a 32-bit float. Makes no difference. It's still PCM. It might be PCM that has a theoretically large, dynamic range. Spanning more than 150 DB. Well Lah Dee Dah. None of the input or output electronics can deliver, anything like that. Nothing like that. It's a phallus, see?
This is such idiotic mumbo-jumbo. That it hinges upon, wacky techno-audiophile gobbledygook. That actually has no bearing in, Audio Fidelity. It's more gobbledygook garbage because it's still PCM. And it will sound like PCM. Because it is. It does not sound like original source.. Because it's, PCM.
You don't even know. That you're in phase, left and right channel monitor speakers. Are both in, Negative Absolute Polarity, together. Because you think it's the other way. You are convinced. You have a test. You can prove you are right. You are not right. You are absolutely wrong. As this is one of the biggest, acoustic, Technical FUCK up's in Pro Audio History. Because everybody is working on in phase but Negative Absolute Polarity Monitor Speakers. Because they think it's the other way. They can prove, it's the other way. No you cannot. You cannot prove anything. I can. I have. I do. And I can prove how stupid you all are. And you can hear for yourself. How stupid you have been. When you follow my advice and suggestion. And what is that? What is the fix for what I'm talking about? Of the way your monitor speakers currently are. In phase but in Negative Absolute Polarity, together, in phase.
Nobody understands that Polarity and Phase are actually 2 different things. Though based on similar concepts. They are vastly different from one another. In the land of Pro Audio.
Unfortunately. Most other audio guys are also reaming idiots. And don't know what they are hearing. Don't know what they are listening to. Don't know the way things are supposed to sound. But they have college degrees. That says they are supposed to know. Unfortunately, they've never figured it out.
All of the electrical engineers and all of the acoustic engineers. Have been getting this issue wrong now. Going back well over 70 years. It's always been wrong. Though it changed. For some. Back in 1948. From a single speaker company manufacturer. JBL of Los Angeles California. This is not an advertisement. This is not an endorsement. This is a Technical Fact. They figured this out. No other speaker manufacturer. Has ever figured this out. Never. None. Because they think they are all, right. And they have a flawed test. To prove it. With a 9 volt, battery.
The 9 V battery. When connected to a speaker. Will cause the speaker cone to either move out or to suck in. You wanted to go out. And that indicates. The speakers Absolute Polarity, for connection. And that would be plus polarity.
Now when you connect that speaker up to your amplifier. The instruction manual for your amplifier and your speaker. Indicates you should connect the positive terminal of the output of the audio power amplifier. To the positive terminal on the speaker. And the negative terminals get connected together. And that is absolutely wrong. And here's why.
When you do it that way. To get everything In Phase. It will be in phase in Negative Absolute capillary. Meaning that. The speaker is not protruding out toward you, the listener. No.. The speaker is sucking inward. Away from you. It's sucking into the box. It is not protruding out toward you.
It is sucking inward into the speaker box. Because it is in phase. With the microphone input. And that's just plain wrong. Why is it wrong?
(More Happy Cluelessness in following post)