What is 32 Bit Float Audio (and do you need it)??
ฝัง
- เผยแพร่เมื่อ 27 ส.ค. 2024
- I've made a better, updated version of this video. You can find it here: • BEFORE You Use 32-Floa...
No, you don't. Set your levels properly. However, it IS really neat.
I've since made a follow up video to this after using 32 float for a few months:
• I WAS WRONG - 32 Float...
---------------------------------------------------------
If you're into Instagram, you can find me @colonelshaun
/ colonelshaun
Hire me at boomreactive.com
Kit information at kit.co/colonel...
I liked this video cuz there was no intro and you answered the question immediately. I stayed for the info and out of respek. Much love man.
facts
I really appreciate that! Though since making this video, I've bought an F6 and haven't recorded without 32-float in over a year. :-)
"A volcano eruption is the loudest sound ever recorded."
Wedding DJs: Hold my beer
Hahaha.
**Wedding DJ proceeds to unplug speaker cable and plug it back in without turning anything off first.**
dads watching football: gimme my beer
32 bit is indespensible for one man band sound recordists as it negates the need to dive for the gain knob when an actor starts getting over dramatic. The amount of times I've lost dialogue to clipped audio when the actor screams or shouts. 32 bit is god send. You can just set the gain and forget about it. The problem with most field recorders they come with digital compressors making them of no use. I'll always record in 32 bit. Just set your gain and let it fly.
Since making this video, I've become a convert. I've purchased the Zoom F6 and always use 32-float. I still stand by the fact that no one NEEDS it, but for one-man-band stuff (especially now during pandemic times), being able to setup the recorder and then essentially walk away is pretty handy. I really like how FCPX can handle the files natively, too.
@@ShaunHautly So can pro tools. Its had 32 bit support for years. I just do my mix sessions 32 bit now. It handles 32 bit internally so really theres no such thing as a 32 bit session as it treats like a 24bit session. The only difference is you can turn down clipped files.
@ShaunHautly hey bud, I would like your opinion on something. I've never been a fan of the sm7b, considering the re27n/d and its minimal 50.5 super clean gain requirements, a far more Superior, and beautiful sounding microphone.
Unfortunately I do have to buy an sm7b as it is being requested by way too many people. I have read that you need a minimum 65 DB to power that hungry guy. Do you think if I get the budget tascam Portacapture X8, and it's very subpar signal to noise of 100.5 DB, it won't matter as much because of the 32-bit float? I know the industry professional standard is 120 to 128 s/n with 24bit & 48,000. But, since the Tascam has an essentially unlimited range with 32-bit float, and only records at a weak 50DB, perhaps that meager 101 signal to noise ratio won't matter? Or, will the very low 50 DB, 100 s/n preamps add noise no matter what, when I boost it to a reasonable level?
This is why I always recommend 32 bit float. It's just an extra safety net that won't back you into a corner, especially if you're a mix/master engineer getting crazy sessions from random clients that don't necessarily know what they're doing when recording.
@@markowalski1 32bit still requires a knowledgeable soundie. I've just had a director reach out to me to assess the production sound on a film she's just completed. Her short film has been ruined by a cowboy soundie. 32bit would not have helped, I had to give her the bad news that her films production sound is trash . Bad booming, no use of lavs, mic too far away. You can't fix any of that.
Whilst I praise 32bit I still find that the infrastructure in the post chain remains a problem. Video editors still don't understand the concept and import the audio into their editors as 24bit (or even 16bit) without checking. I always get films back for sound editing with clipped rendered audio which I can't then get back the peaks. Forced to dig out the clips from the original field recordings which slows everything down. I got so fed up I went back to 24bit recordings and using a limiter.
They'll be sure to set the frame rate and aspect ratio of their project but not think to set up the audio parameters. Most editors now support 32bit but they just don't know to do it, even after I explain it to them.
Of course the answer is to use a software like Ediload.
Thanks for this, it was well explained. I've only just come across 32bit float and find it really fascinating. I work in two fields, as a photographer and as a musician (who also does field recording). As a photographer, the flexibility of a RAW file is the safeguard that if you mess up, there's a chance or recovery, of some sort. But, all competent photographers know that getting exposure right in camera is essential, as it saves a lot of work in post later. You could argue that if you do this, why use RAW? But, your workflow will thank you for it! I imagine, for me in particular as a field recording artist, 32bit float gives me that safe guard. If I'm recording the band for example, I can't always be monitoring, if I'm field recording you have no control whatsoever of external sounds. So it definitely benefits getting the setting right, as best you can in any given situation, and the 'safety net' of 32bit float gives you certain safeguards should something occur beyond you control.
Good example! nicely put
32 bit float recording at 192khz is actually good for sample creation. One example, to record a bee up close. Taking that sound, and slowing it down, while still retaining characteristics of the flapping wings... the things you can do nowadays. Very comparable to those Slowmo Guys channel.
So, your explanation for an average musician is very good, just have to think about possibilities.
Yeah, I totally agree. Since I started using 32-float, I’ve fallen in love with it. For video production there isn’t a reason NOT to use it.
I haven’t tried recording a bee, yet, though.
10/10 A + Thump's up Bose Speakers! :) Woooooo lol
Thanks for getting to the point. This is the first video that just explains 32 bit float in a way that makes sense. Great Job!
Thank you!
I love how you explain stuff bro . Best I’ve ever watched on this platform
Daymaker. Thank you!
As soon as you said it allows the ability for one to go in the booth and deliver RAW performance (not ever having to worry about levels until after) I immediately saw the difference. Just set the gain and let her rip? Oh yeah that’s priceless man. I totally agree that it’s not needed to make amazing music, but it seems to be one of those little golden nuggets that help tremendously if used properly. I’ll be making the switch soon whenever I build the rig for it, Thanks for the video!
If you didn't see my follow up video, I'm a total fan now. I haven't filmed without it in months and it has saved quite a few podcasting moments where laughter suddenly spiked things.
Thank you for that informative video! Recently, I started using IRs of 32-bit for my guitar amps and I can see there's a difference comparing to 24-bit ones. 32-bit drove the sound to become brighter, detailed, and louder.
Rock on! Where can we hear the music?
You are correct, most people do not need it. But you are incorrect in some of your explanation. What you are describing is reminiscent of 32-bit recording. 32-bit recording usually requires of 64.1hz (32x-2) or more to avoid any aliasing at the boundary. Float technology is not the same thing as 32-bit audio. There are slightly different ways, but most used is a 16-bit main and 2-3 additional tracks. We will use the former example . The top track, or peak track, while oscillating at the full 32 bits will will not add data unless there is a loud peak encroachment. Similarly, any extremely quiet sounds are permanently recorded, roughly an additional 8bit. 16 + 8 = 24. This is of course too push up any extremely quiet sounds. The cool thing about this tech, you don't have to set your levels at negative 12 peaks or negative six peaks, you can go the old school analog way and set them at zero peaks if you want. This way you don't have to gate recorded sound , which can occasionally introduce artifacts. And because you are maximizing your preamps and hopefully a low noise, clean gain microphone like At2035 (20 bucks more than the 2020 with a 1 inch capsule instead of a 11.5mm electrtt) or EV RE-20 (greatest dynamic broadcast, and universally the greatest dynamic of all time). 32 bit is wasted on hungry dynamics like the 60db with only a 38 clean gain... yikes! If you get that thing, whatever you do, don't put a cloud lifter on it , watch Julian Krause, and do not get it unless you're ready to put about $4,000 to 10,000 chain behind it. Yes, it's that thing. Most engineers do not understand, the SM7b or the new DB which is hilarious. I promise you, even sure says it's in a sm57 with a shotgun design. there's a long roller coaster history with this ugly old boom television mic callrd the SM5 (just a blimp with an SM57 inside) and it has everything to do with old analog static No longer present in radio studios. By the way, you don't need more than 16 bit with a very clean preamp *MIN 126ein , you'll stop as soon as you hear the noise and have enough headroom for two of those things. I say all this because we are really talking about clean dynamics with 32bit float, with a minimum 80 (preferably 95) clean gain and 105 DB range between the microphones and amps. The biggest use for 32Bit? Lavaliers. Sure enough some guy is going to scream right into it and you will lose audio. you have to opt for a very expensive Sennheiser or behringer system, which some of us know still hiss when you have more than a few people on, and youre going through two to three conversions. If you're recording music, there's almost no use for 32bit. If you are broadcasting, there are many video microphones which come with a safety track option. Or you can use a stereo microphone and lower one of the tracks by 10 dB.
This is genuinely my favorite comment ever. Thank you for being polite but also for being SO THOROUGH! I have a few videos that have gotten way more views than I ever expected them to get. My research is never as deep as it needs to be, and I often get excited to make the video once I’ve learned 10% of what I need to know, likely. I’ve since made two follow ups to this video. However, after reading your comment, I know there are still tons of errors in the updates.
In the next few months, I’ll revisit this and see if I can’t get it more accurate this time around. Thanks so much, again!
Great explanation! I use 32-bit Float on my Sound Devices MixPre-6 II because sometimes I cannot sit and control the levels. I shoot dance recitals, pageants and other events like that... and I usually leave my soundbag hooked to their sound system backstage. I *do* set my levels in rehearsal... but I have peace-of-mind knowing the audio can't get clipped if something changes. It's a great insurance policy. :)
Yeah, if you're not able to actively monitor your audio, it can be a nice, emotional buffer. I've had a few situations where clients abroad have sent me audio files to incorporate into a project and I think 32 Float would have been a GODSEND for these people setting their own levels with no experience. However, that's a big purchase for someone who would just be doing voiceover sound bites...
Very true. I have the mixpre II. The older hands aren't convinced. I got into a heated debate with them a while back. They can't see the benefit. Telling me "you should record in 24 idiot!" "your adding an additional step!". They don't understand how 32bit float actually works in a DAW. Ric Viers said himself that riding the gain is actually bad practice as you are noticeably changing the background noise which makes clips inconsistant which just makes more work for the sound editor.
The only argument I see against 32-bit, and why I leave my MixPre 6 II on 24-bit, is for the sake of having insanely nice analog limiters. I'm exclusively recording dialogue, and I'm going to be compressing in post anyways, so I'd rather have the analog limiters helping with a very small amount of peaks. If you aren't taking advantage of the limiters, the ZoomF6 is arguable a better bang for your buck as you get more XLR channels at 32-bit float. (just some food for thought)
@@johnmellor932 Agreed about the luxury of having the consistent noise floor. I'm intrigued by the simplicity of 32-bit for instances when you are capturing a project solo. Meaning, having to concentrate on the shot (video), the audio, the person being interviewed, the lighting, etc., and can't focus solely on the audio levels.
I brought a Zoom F2-BT and it saved my ass only yesterday. One of my lav mics has an amplifier in the middle of the lead. It sounds great but that battery is about flat and the level was that low I thought shit I’m going to have to record my intro all over again. Like the wave form was only just visible. I thought well 32 but float do you thing so I increased the volume to max and it sounded great. It saved me another 30 mins to an hour re-recording the intro to my latest video.
Great explanation on how 32 bit float works. Thanks.
Happy to help! I recently redid this video because I've been using it for a few years now and have a new perspective on what's great about it and the mistakes a lot of people make when they first try it. Glad you're on board!
@@ShaunHautly if you get a chance try the Zoom F2 it is a seriously cool bit of kit. Especially for motovloggers on motorbikes. It is tiny and the chance of it causing you injury of you have an accident I think would be minimal. You would have to take a hit on the recorder itself.
Thanks! I I listen to orchestral music with 32bit levels, and it definitely helps pick up nuances of the different 80 instruments, but sometimes it picks up the unwanted noises like breathing, page flipping, etc... GREAT explanation.
From what i’ve learnt about 32 bit floating point system from a maths perspective, the precision of the number is far higher close to zero, so setting proper gain levels gives you a far higher precision in your signal than if you are at either extreme of the range.
Interesting. The zoom recorders don't even have an actual gain level control. You can adjust the playback level-- which will be baked into the file-- but the actual conversion of the signal is constant.
I wonder if the sheer abundance of available data means those extremes are never really utilized by the relatively narrow range of sound.
Thanks for the comment! I just made a follow up a few months ago (linked in the description), but when I revisit, I'll look into it. Cheers!
Neat, can`t think of a reason to use, but neat :D This might be something useful for science when you try to record sounds outside of the usual spectrum, as the extra dynamic range and data might make it possible to isolate those sounds. I seem to remember something about a infrasounds that causes people to feel anxious and nauseated, maybe this type of file would be better at recording that (that`s probably more a Khz thing but I could imagine the extra data might make it easier to edit the file to isolate those sounds).
Yeah. I'm trading my f4 for an f6 which has 32-Float, but I don't anticipate needing it. It might be a nice-to-have when I'm a one-man-band and conducting an interview while recording myself, but mostly it's just a ton of extra data. Watch me redact this video in two weeks...
@@ShaunHautly :D
Its useful because they've rolled it out in affordable mid range field recorders that almost never come with analogue compressors. Or id your a one man band foeld recordist. I've seen so many short films with clipped audio because not everyone has $5000 for a Sound Devices 633
@@ShaunHautly the F6 doesn't actually use 32bit float. It uses dual adc, kind of like a cross over on a speaker. One adc handles low frequency and the other high frequency. Zoom have kept this technology secret. It acheives the same result. The SD MixPre II uses a single adc.
Great breakdown, thanks man!
I swear I'm not just trying to self-promote, but I redid this video recently with a lot more perspective and experience. If you're interested, you can find it here:
th-cam.com/video/TEG6ePEGxq4/w-d-xo.html
This was a useful explanation. Thanks!
I'm glad it was helpful. I've learned a LOT since making this video and made an updated one that's linked in the description. You may enjoy it as well!
You make really good points. I think 32Bit float will be great for documentary filmmakers. But for sets, I think everyone else should be good.
I was wrong. Turns out 32-Float is incredible and there's NO reason not to use it. Its ability to capture such range makes it possible to remix everything to sound perfect. After I made this, I tried it a few times, and I'm never going back. Especially since FCPX has native support for the files. I can add and subtract gain directly in my timeline and it can save or recover anything. Holy cow.
@@ShaunHautly I can see an argument for not needing it as general listening use-cases, but for creators: I'm glad you've seen the light.
If you're working with inexperienced or unpredictable artists, then that safety net is just for peace of mind... But you trade that peace of mind for concerns about data storage for the MASSIVE file sizes...
That's true. However, for me, the massive file sizes aren't that massive. Especially when we're capturing 4K video from three cameras simultaneously. Even an hour is only around 1GB. Compared to the 500GB of video footage.
I still don't always trust it, though... Ha.
Man I need 32bit float for my musical instruments
It's SO good. I've since made a response to this video after using 32-float for over a year. The link is in the description. I'm a believer now.
this reminds me of the exchanges among photographers about should one shoot RAW or JPEG. Sure, when all is perfect and stable then record 24bit. But in reality, many of us on the run need the flexibility.
Totally. After making this video, I got a recorder that does 32-float and while I don’t need it, it’s a really really nice nice-to-have. Being idiot proof is a great thing to be.
lovely video very informative thanks to shearing the knowledge
My pleasure! I've since begun using 32-float, and I've updated my thoughts in the video linked in the description... spoiler alert: I'm a convert.
The best AD/DA audio converters are only around 125db of dynamic range. Full 24 bit audio (144db) still to this day has not been realized.
True. The F6 just put TWO AD converters into it so that even more dynamic range is packed into it. It's absolutely overkill.
@@ShaunHautly Why stop there? Put three or four in it. Maybe eight.
I wonder if this is what people said comparing RAW photos to JPEGS.... Do you NEED it? No, but that argument is a slippery slope lots of time with the introduction of new tools.
I think this has a place in the arsenal. Especially in shops where multiple people use shared gear in different ways, or 1 person production crews, who are juggling lighting shooting interviewing and audio simultaneously
Yeah, it's not QUITE as pivotal as RAW. More like just being able to set your ISO afterwards. It can't fix color balance, etc. In practice, it's amazing for when I'm filming by myself and need sound, but it doesn't overcome the need for microphones to be selected and placed properly, etc.
If you're trying to record someone whispering, and there are gunshots going off, you should probably run away.
This made me laugh aloud. That's a very good point.
This is such a good explanation, clipping audio levels is very easy to do when playing multiple synths at the same time. I think I need it! btw the H6 doesn't suport 32bit wave. "Up to 24-bit /96kHz audio in BWF-compliant WAV"
I've since made an updated version of this video that I think is way more helpful. The link is in the description. But your'e right about the H6, my bad!
Thank you for saving me the money
If you didn't watch the update I put in the description, you may want to...
After researching and making that video, I ended up buying a recorder capable of 32-Float, and I'll never, ever go back to life without it. My apologies!
@@ShaunHautly oh ok cool thanks. I'll watch
Great video, definitely helped understand 32 bit float much better. I use 32 bit float in the Zoom F6 every time I record. I actually record 24/32 just in case. When I first heard about 32 bit float, it was explained to me as if I was shooting RAW video files. Which is pretty close to what you went over. Thanks again for the info!
Im learning that in 24 bit mode even if the trim is way off, if the fader brings the level down it won’t be clipping. That is bonkers. I need to experiment more, but as foreign and strange as some of that is, I can’t help but feel like that’s how audio SHOULD be treated by recorders.
Good stuff. Many thanks. (You did great).
Glad you enjoyed it! I've learned so much more since making this video. There's an updated version in the description if you run out of other things to watch. Cheers!
Much appreciated! Thanks.
Glad it helped!
Great video. Love the insight. Especially the tip at the end.
Appreciate it!
5:27 the zoom f6 levels are just for monitoring, the recording is preset gain. I could be mistaken...but I got one a last year and I thought I remember finding that in my research.
Thanks again homie
Awesome stuff! This was exactly what I was looking for. Thanks for the information and great video. :)
Glad it was helpful! I've got a few follow ups that I'm excited to make as I continue to learn more about 32-Float.
THANK YOU SO MUCH
You're welcome! Though I’ve since made a couple videos updating my thoughts about 32-float. I’ve learned a LOT more than when I made this first video.
(32 bit float recording) You made it easy to understand .. Thank you Shaun
Happy to help!
I agree that 32 bit is not for everyone, but the definition of dB is misunderstood (and by many, judging from the comments). dB in dynamic range (bit-depth) and dB of loudness (air pressure) are completely different units so it's not an apple to apple comparison. Furthermore, 32 bit FLOAT (which is 24 bit integer and 8 bit exponent) is a whole other beast.
Yeah, since making this video I've learned a lot more and filled in a lot of blanks for myself. Not just in the actual science of how it all works, but in my embrace of the format to my surprise. I ONLY record in 32-float now, and I can't imagine going back.
@@ShaunHautly Thanks for the reply. Yeah, I still appreciate you stepping up to express your thoughts on such a technically complex topic, it gave me the chance to read up on the technical aspects of this myself.
thank you for this video.
My pleasure! Thanks for the happy comment!
I think you got the dynamics thing on PCM wrong. Yes, you have a higher dynamic range at more bits. But the maximum signal possible with PCM WAV is +/- 0dBFS, no matter if it’s 8, 16, 24, 32, 48 or even 64 bits PCM WAV. The loudest tone possible to record is determined by the max SPL the microphone can take, the output of the microphone and the gain set on the interface. The higher the SPL limit of the microphone, lower the output of the microphone and the lower the gain, the louder a recorded sound can be. The real benefit of higher bit depths with PCM WAV is the lowest possible noise floor. The lower the noise floor of the recording is, the quieter a signal you want to record, without being swallowed by the quantization noise. 8 bit PCM = -48.16dB, 16bit PCM = -96.33dBFS, 24 Bit PCM = -144.49dBFS, 32 Bit PCM = -192.66dBFS; 48 Bit PCM = -288.99dBFS, 64 Bit PCM = -385.32dBFS. But take these numbers with a grain of salt. Practically, even the lowest possible noise floor of PCM WAV is always a bit higher than the quantization noise floor. And: Most audio interface Inputs are nowhere near the limits of 24 bits PCM.
There's a lot that I got wrong with this video. And I'm still learning more! I'm hoping to revisit this video and issue another update later this year. It's all fascinating! Thanks for the info, I'm excited to dive deeper into it.
Due to the limitations of atmospheric pressure, the loudest possible airborne sound on earth is 194dB*
*this is at sea level and when we still can think of it as audible sound waves. It is possibly for higher decibel levels, but at that point you are looking at destructive levels of sound pressure fluctuations involving basically pressure vacuums in the rarefractions.
That's fascinating. I loved making that video and I've learned so much since (including quite a few things from comments like these). The funniest thing is that I've since upgraded to the F6, and have quite quickly become quite fond of 32-float audio. FCPX in particular handles it so seamlessly that I don't see myself going back to 24 unless I really had to. It's a really cheap insurance policy. It's great for podcasting too. Guests have gotten excited and clipped, and now that's no longer a problem. Do I NEED it? No. Have I learned that I like it more than I thought I would? Yes.
Just a comment. As you noted, this is the atmospheric limit. You can get louder sound in metal, for example. I'm not sure how often I'll be around metal ringing that often. :-)
Great, very clear, thanks
Glad it was helpful!
Expensive equipment to edit it?? Why are people so afraid of something like this. You explain it really well. It's a ton of resolution. So why not use it to prevent surprises. Like do you need a fire extinguisher? Sure you can use every safety check there is and no. You don't need a fire extinguisher. But why not have one anyway?
Exactly. Since making this video I bought an F6, and have never recorded without 32-Float since. Just peace of mind.
Thank you for the understanding, bro!! 👍👍👍
No problem 👍
I came across your TH-cam video and was impressed with your explanations.
Everything you said is true, but you left out a few points. Even with "only" 24 bits at 144dB DR, most analog electronics, like your mike pre-amp, can't match that, so you still have to ride the gain to keep it from overloading. Then there's the mike's max SPL as another limit. This is true even with a "digital" mike's AES42 output.
The main advantage of 32-bit FP is that unlike 16- or 24-bit, which have linear intervals (e.g. the LSB is the same size over the entire range), 32-bit FP is logarithmic, with smaller intervals between data points at low levels, and larger ones between higher levels. (Just like human hearing.) If you need to raise a whisper recorded too low in 24-bit, it will become more granular than the same situation in 32-bit. Whether anyone can hear the difference is another matter.
Another way to go is with dual 24-bit ADCs. One digitizes the low levels, and the second does the highs. Then their digital outputs are merged to give a single 24-bit output, with no danger of overloading the analog portions of the ADCs. Zaxcom uses this, in comparison to Sound Devices' 32-bit FP, in the professional recorder arena.
Finally, don't confuse the recorder's dynamic range (the difference between its noise floor and clipping point) and the dynamic range of the program material (there needs to be enough footroom between the noise floor and the lowest audio so the listener isn't aware of it). 20dB is acceptable, but 30dB is better.
Quite a difference from the Nagra III (1/4-inch reel-to-reel mono tape recorder) I started with in the late 60s.
I love this. I made this video before I owned an F6, and I’m eager to do a review now having 32 float at my disposal. My anticipated use is for recording with children who tend to have quite the range between their loudest and softest volumes even in a sound bite.
I'd say that 32-bit/floating point is pretty usable, and the files aren't that much larger than 24-bit. The reason has to do with clipping, where 24-bit is in more danger of having clipping issues, especially if you have the need to bounce out and use the result in a new mix.
192 kHz on the other hand, is mostly a gimmick the way I see it. Yes, I'm aware of the cinematic use - for the issues with driving it (rip, my beloved cpu) and the fact that some plugins won't work properly with it, makes it too niche to appeal to me, at least.
I agree. I've made an updated version of this video in the description after having used 32-float for two years since making it. I'm a full convert now.
And you're right, I'm at 48 kHz guy these days.
32 bits only allows 2^32 different unique values, 4.294.967.296 (2147483648 positives values, 0 and 2147483647 negatives values ). 32 bits have holes or granulation in comparison to continuous 24 bits.
Since making this video, and the updated version (in the description), Ive learned SO much more about 32 Float. I don't imagine the math will affect most users, but as a math minor, I'm always eager to get it right. Thanks for the bump!
It was very well explained! Great video!
Glad you enjoyed it! Cheers!
If I record an instrument (for example a guitar) using a 32 bit audio interface does this make the hum noise more quiet?
Nope. If the hum is in your signal, it's going to be in your signal in 32, 24, or any other bit depth. It might be SLIGHTLY easier to correct in post, but I wouldn't count on noticing a difference. Your best bet is to hunt down the hum and avoid it at the source. Turn off any air conditioners, refrigerators, etc., if you can. Unplug anything you don't need for your recording. See if that helps. Bad cables can also be a culprit, but it's more likely some electrical interference.
This was helpful, and super interesting. Thanks
Glad it was helpful! However, I've since started using it and LOVING it, and never wanting to record without it again...
@@ShaunHautly Really? I was asking because it is an export option in Ozone9 which I use for final mix/master. Would you recommend I export with that? If so, is there something I'd need to do in Ableton Live (settings) when I export out of it, so that I could take advantage of the 32 bit float in Ozone?
Great video
Thanks!
well done
Thank you!
First things, first, thanks for sharing. Is it suitable for documentary? (for unplanned scenarios?) regards mate
I mean, i dont want to be worry for peaks, or levels...
I've made a few videos since posting this one. YES! I LOVE IT NOW!
I'm primarily a documentary-style, interview-based filmmaker. I use 32-float for absolutely everything I record. You still need to set your mics up properly, but if they're setup right, you can adjust your levels in post and fix any clipping that may have occurred, or bring up your levels without bringing the noise floor up, too.
Everybody keeps saying you don't need it. Are you crazy? First, the file size is only 30% bigger so why not? Second: Not having to set levels at all is huge! It takes time to do that when you have a large channel count, and it can go wrong why should I not skip it? Third: Yes nobody needs to record a gunshot and a whisper. But in the end, when recording music, stuff gets put through compressors, distortion units and Limiters a lot. This can amplify quiet noises (like natural room reverb tails) multiple times, why not record everything with the best resolution possible? And Lastly (this is more of a question) doesn't this mean we can ditch Preamps altogether since, we can record quiet sources with such great resolution? People pay thousands per channel to have "clean" preamps that don't color the sound, not having to do that would be crazy good for audio interfaces.
I agree with you, 100% Since making this video, I started using 32 Float regularly. I made an apology video with my change of heart. The link is in the description of the video. 32-Float 4EVA
@@ShaunHautly Oh nice :D Sorry now my comment seems way too angry
Hahaha. I delete the mean/angry ones. You kept it in bounds. :-)
Wow, thank you for making this video!
My pleasure! Thanks for watching it. And thanks for leaving a nice comment!
That was beautifully explained! thanks for sharing :)
Glad it was helpful! Though you should know I've changed my mind on whether or not a creator NEEDS it: th-cam.com/video/mP5H4_HJGAw/w-d-xo.html
Very helpful
You taught me something today - so thank you for explain it so well and in layman's terms. What would be handy to know is when looking at Zooms F1 and F2 - the latter records in the 32 bit float thingy. Both products take a micro SD card. So how much extra data capture "room" would be required comparing a 24bit WAV file against 32 bit float? - Thanks in advance, Peter.
In theory, exactly 25% more. Each sample is being recorded into 32 bits instead of 24. For most of my 32 Float 48 kHz recordings, I'm landing right around 1 hour = 1 GB.
@@ShaunHautly Thank you so much.
What would you recommend setting my speaker setting on audio high fi? 24bit/192 khz?
While I don't fully understand the question, there's no harm in going to 192 kHz if you've got the space for the larger files, but it's also really difficult to hear the difference between 192, 96, and even 48. But just give it a listen and see what you like. I always match the sample rate to whatever the project is going to be. For most of my video production, 48 kHz is the standard. If you're doing music, a higher sample rate might be better for you. I can't say.
Good stuff, man!
Thanks!
I record in 32 bit float for the same reason I shoot raw files for photos and also video when possible: flexibility in post.
Same. Since making this video, I’ve gone 180° and I love it. I put an apology video in the description.
How did they record 180db in the 18 hundreds?
I don't know. They probably just guessed.
thanks brother for the clean explanation about it. God bless you
My pleasure! Good luck.
My focusrite scarltt solo is only 24 bit, but my fl studio lets me record in 32 bit…is it really recording in 32 bit float resolution using fl studio software?
I'm not familiar with either of those devices. However, I assume that the system will capture the lowest option. It'd be like taking a REALLY good, RAW photo of a printed our JPEG. It'll be a good photo of that lower resolution photo. It's not like the 32 float will find extra information hidden in the 24 bit files.
That's my best guess!
At the end of the day, try it. If it sounds good, then great!
@@ShaunHautly i don’t know why but when i use the 32 bit float on fl studio the track sounds better when i add effects like gain and compression
Ok - so ..... Great explanation - great video - So I'm a guy who learned sound on a Nagra back in the day - When they came out with DAT - everyone said: "Do you need a DAT"? And the resounding answer was "No". The telegraph operator wasn't a big fan of the telephone either - "Do we really need the telephone"? To this day some may still say "No". Anyway... I think you get the point. This 32 bit floating recording is a game changer. It gives the Indy film maker the opportunity to eliminate one more thing in his production which could go wrong. As an indy filmmaker - I know "if it can go wrong - it does go wrong". If you don't believe me you either haven't done an indy or haven'r seen Living in Oblivion. As a gratuitous plug - you can also catch my indy I did back in the day: th-cam.com/video/AH2V32vDSS0/w-d-xo.html CHEERS!
I've learned quickly that I was wrong to hate on it. I haven't recorded WITHOUT 32-Float since I got the F6. Here's my updated apology: th-cam.com/video/mP5H4_HJGAw/w-d-xo.html
And I dig the indy! Need to find the time to watch all of it. Cheers!
I was looking this up because I am looking at the Zoom F2. I wanted to be sure it was a gimmick or actually works. It’s small and portable.
I've made a follow up to this video. I have become a huge fan of 32-float and even when audio is my focus on a job, I love having the forgiveness it provides to save sudden laughter or outbursts into sensitive microphones.
my microphone records at 16bit, is it recommended to record at 24bits or 32bits in my DAW?.
Your MIC records at 16 bit??
Setting your DAW any higher than your lowest bit depth won’t improve anything. It’s the whole “chain is only as strong as it’s weakest link” thing.
@@ShaunHautly Ok, thank you very much!!!!
Quickly... what's the diff between 32bit and 32bit float? Thanks
In audio, all 32 bit is 32-bit float. What it means is that it's a "floating point integer" which is to say that instead of just whole number values (like 'Linear' 24 bit and 16 bit) which would have amplitudes of "104" and "44," instead as a floating point, they have long strings of a decimal with an exponent and that exponent can be positive or negative. So your actual value could be 1.487392 x10^-4. By structuring the numbers that way, you can achieve a MUCH wider range of numbers from lowest to highest with the same actual number of digits, and that's how the format can preserve so much data.
The good recorders need to have the right electronics to capture that audio well, but if they can, 32-bit float has so much theoretical range that it's going to be the gold standard for a long long long time.
@@ShaunHautly thanks for that. Yeah there's the options to export @ 32 and 32 float in my daw (cb). Thanks
True, unless if you want to hear the whispers of spirits.
And the spirits say "use 32 bit float....."
So how about mixing down in 32....any difference?
For me anyway, it's REALLY tough to hear the difference between 24 Linear and 32 Float. Some people say the high end is better on the 32 Float, but I don't hear it in video production. I do love its ability to add flexibility to adjust things before my final file is exported, that's where I notice the difference. Once it's balanced and mixed down, I can't tell the difference between the 24 and 32 files.
@@ShaunHautly Thanks. I think I notice a little more of a softness in the hi end but who knows.
I'd REALLY like to hear an explanation of how recorders that are 32 bit float capable (and you "don't have to worry about gain") avoid clipping in the analog preamp stage before the A/D converters.
While even the dual AD converters can't harness the FULL potential of 32-bit, most pre-amps can handle a LOT more than pervious AD converters and 24-bit could harness. So the quality had to go up before we need to build better pre-amps. I assume that will be next, but honestly they're so good right now that most improvements won't be noticeable to our ears.
Silly hypothetical: what if you're trying to interview a group of people while wearing (say) a Zoom F2 with the lavalier mic on yourself. Would recording in 32-bit float help in that situation? That is, could your normalize that recording in post-production to make it all sound good?
I'm going to say no.
While I think you could normalize some audio levels to make things sound better than they would without 32-float, it's still not a replacement for good mic placement. If a person is far away from a mic, that mic is going to pick up a bunch of noise and ambience. That noise and ambience will be recorded into the signal, and no amount of flexibility from 32-float is going to make that person more clear.
The ONLY thing 32-float does is allow you to adjust the gain in post. So I'd still get a few mics out and try to get them as close to people's mouths as you can.
@@ShaunHautly Thanks.
very good video. and very entertaining as well. :)
Thank you!
Great Content Bro! QUESTION.... I have a recording studio and I'm creating a Podcast/Vlogcast network. I do not want to use the typical Shure SM7B mics. i want to use all Lav mics. And afterwards I can great outboard gear and software to EQ the podcast vocals to sound damn near as good, if not better than using a condenser mic or shure sm7b. Because I want to do it this way, do you think having the Lav mic with 32bit float ability is the best way to go?
Yes. Especially if you're recording other people. In my experience podcasting, myself and others will suddenly speak louder, burst out laughing, or otherwise significantly change the volume of their delivery throughout the podcast. Unless you have someone constantly feathering the gain controls to follow the ups and downs of someone's delivery, 32-bit float is the only thing that will protect your audio from clipping or being lost into the noise floor.
After making this video, I made another all about how much I now LOVE 32-Float after trying it for a while. th-cam.com/video/mP5H4_HJGAw/w-d-xo.html
@@ShaunHautly Bro.... that was EXACTLY the info I needed! Thanks for that direct comprehensive reply! I TOTALLY understand what you are saying. Gonna watch that video link you sent me RIGHT NOW! Thanks again my brother.
I’m a video creator who films a lot of concerts/musicals & I’ve been debating on getting a recorder that records 32 bit float for gigs where I need a live feed from the board at a concert or play but can’t be monitoring the audio the entire time. I feel that the 32 bit float will give a bit more peace of mind in case of hard peaks in audio from either poor mic technique or drums. Am I thinking about this correctly?
Mostly. 32-Float will protect you from differences in a drum performance from the quieter parts to the louder parts. However, it can't overcome poor mic placement/technique. If your mic is too far from the sound source, yes you'll be able to turn it up, but you'll turn up any background noise along with it.
It basically gives you the freedom to adjust the gain knob after you've recorded, but as you know, the gain knob can't correct for mic placement, the wrong mics, or faulty mics. Triple check your setup (as always!), but know that quieter songs and louder songs won't require you to change the gain/trim settings.
Analog is the limit
Everybody to the limit!
(That was a HomeStar Runner reference.)
You're wrong for creators who are doing a lot and can't focus on everything all at once - it's seriously an audio-saving insurance policy. Especially for my use-case and others like it: one-man crew's with multiple cameras NEED it. I can't always monitor a mix of audio while operating 3 cameras and assume that talent/multiple talents with multiple inputs/circumstances/men-on-the-street/wind/cars will promise to stay in the same gain area without clipping (or needing to Denoise/Noise Compression because the noise floor is a little too hot, no matter what). Clipped audio is the worst and even if you catch it, there are some things that can't be re-shot, even if you're still on a shoot... but maybe I just need to give a crapton of headroom.
You are 100% correct. Like a dummy, I made this video before I had used 32-bit float. Since making this video, I've bought and fallen in love with an F6. I made this apology video: th-cam.com/video/mP5H4_HJGAw/w-d-xo.html
@@ShaunHautly "I was wrong. And I'm sorry for the shame this brought to me, my family, and my company." Hilarious!
Today it's cheaper to buy second mic (even with less quality than main) set his gain lower and in extra cases use this recording. When interfaces with 32 bit (float) will coast less than 500$ than it will be an option. But not now
Some recorders have a dual channel recording that allows recording of the same mic at two different levels at 24-bit. I used this with the f4 every time. Two mics would also work, but the setup gets a bit heavier and more cumbersome to setup. For me, I'm all about that lean kit. Since recording this, I've started using a recorder with 32-float and I'm enjoying it more than I thought I would. Especially because FCPX can handle the files natively.
motu M2 does this I believe, for $180..?
good breakdown
Thanks! Since recording this video I have started using 32-float recording a bit and while it creates a few extra steps, it has been a nice-to-have, but absolutely not a need-to-have.
sure but for example even at 0 db to 110 db range, 24 bit and 32 bit don't have the same resolution ! you get more details at same dynamic range with 32bit. same as a 100mp image and 24mp image, sure the 100mp image is much bigger but still you can zoom in and crop your image more ! as you can see some audio interfaces have 32 bit resolution but the dynamic range of their preamps are just up to 110 db, which means their not stupid to give you the ability to record at 32bit with just 110db preamp coverage ! the 32bit audio is to make samples and to make VSTs and audio tracks for games.
Yeah. With the zoom f6 that I've been using for a few weeks, the recorder has two AD converters. One for higher volumes and one for lower. So you get more dynamic range than a single AD recorder could give you AND the 32-float to put it all in the same file.
For humans, our eardrums will blow at 140+dB, so for recording sound for playback; it makes no sense to record those 180dB sounds lol. unless your experimenting on a device that allows human ears to surpass listening to that without destroying our ears. 24 bit is plenty and more than enough for our fragile little ears to experience sound. leave 32-bit for the bats and dolphins haha.
Ha! Since making this video, I started using 32-float. I'll never go back to 24. It's less about the LOUD noises (which can be captured in 24 bit anyway), it's about simultaneously capturing the louder noises and the quieter noises.
We just did a huge talking head project, educators talking to kids. As these person's dynamics changed while they were being expressive, it was nice to have that data captured to where we could level it all later without being worried about bringing up the noise floor.
Give it a shot, it may surprise you! Also, I may look at training a dolphin to run sound....
Just one question...
If you play a 16-bit float file and have the DAC set to 32-float, is it gonna make a difference vs setting the DC to 16-bit float?
Just to be clear, you probably do not have a "16-bit Float" file. It's a tiny distinction, but 16 bit files are linear files. So I just want to make sure we're on the same page.
If you're playing a 16 bit audio file and have the DAC set to 32-bit Float, you're not going to increase the audio quality beyond what was already captured in the 16 bit file. However, if you're capturing an analog signal from the 16 bit file and the DAC is converting it from analog back to digital, in theory, the quality of that capture would be better at 32-float than at 16 bit.
Did that answer your question or make things more confusing?
@@ShaunHautly Yes that answered it, thanks.
One question What should I export my songs with 32bit or 64 bit or which bit
That depends on where you going to export them. Exporting in too high of a bit-rate may cause compatibility issues. When capturing, higher bit rates can help preserve quality for editing, but when you're done editing, you can export at 24-bit at 48kHz, and that'll sound as good as anything you hear on Spotify, TH-cam, etc.
@@ShaunHautly That`s the answer :)
When I lower the gain in either Audacity 3.0.2 (confirmed can edit 32bit float) or AudioDirector 11, it just flattens all the peaks to one equal level. Using a Rode VideoMic Pro+ with a Zoom F2 BT. I specifically got this combo since I record very loud concerts. Any suggestions? I see in your video you just adjust the gain and it works as expected. Thank you!
It's because of the Video Mic Pro+. That microphone has its own limits which are being reached before the F2 has a chance to save anything. You need to use a different mic to see the benefits of 32-Float.
Helpful Info. Can you help me to understand if a 32bit float .wav file produced by the Zoom F3 is supported in LumaFusion?
It's called insurance ♥️
Yep! I've since become a full convert. I linked a video in the description to my change of heart. :-)
@@ShaunHautly we are #blesses
Another question: are there any mics that can capture the range of 32 bit float?
Short answer, no. 32-Float is overkill.
However, most modern mics can capture the crazy amount of dynamic range that exists in most real-world situations. So with any decent mic, this recorder will capture everything around. And I'd rather have the recorder handedly eclipse the mic than get too close.
@@ShaunHautly thanks
So what’s the difference between 32 bit and 24bit
32 bit has a magnitude more information in it. Beyond what our ears can even perceive. It’s strength is in capturing in an essentially foolproof way before you convert it to 24 bit for your project delivery.
Thanks man👍
My pleasure! Cheers.
But that’s a 5 with 41 zeroes after it… not 38….
HAHAHAHAH THAT’S WHY I REDID THE VIDEO!! I’ve been waiting for YEARS for someone to count that and call me out. I wish I had a prize I could give away. There’s an updated (and accurate) version in the description.
@@ShaunHautly didn’t count, just saw the groups and did simple multiplication and then added 5. Easy to spot but we all make mistakes, so don’t worry. You can finally relax.
well thats actually wrong
Yeah. I know. There are a couple things in this video that I've learned more recently aren't accurate. That's the problem with learning...
...not the zoom H6
Yeah F6, not H6.
At 00:29 You meant Zoom F6, not Zoom H6.
Yes. Whoops.
Do you need it? NO! (Deletes few albums and saved space) Thanks!
Shaun: 16? 24 bit? 32 bit float You don't need it!!
Cubase: Hold my bits. 64 bit float.
#djrtnyc #futureproof
Hahaha. Future proof until.... GarageBand128!
You might want to educate your viewers between 32 bit integer vs 32 bit float. Not the same.
Is 32 integer and option on anything? I haven’t heard of that one. This video was all about float. I didn’t know there was an alternative 32 format readily available.
@@ShaunHautly There are companies making 32 bit DAC now. Stupid I know. But people in the comments seem to be confusing 32 bit integer with 32 bit floating point. 32 bit integer referes to the realtime audio recording resolution while 32 bit floating point referes to editing precision or math precision. 32 bit integer has it's dither while on the other hand ....when converting a file 24 bit file to 32 bit floating point it's still a 24 bit integer file with 24 bit dither.
Man, 32 bit is nothing, 64-bit floating is the thing)
I cannot even imagine. :-)