A lot of people have asked how I determined the amount of gain I added when turning the gain up on the interface, since the knob isn't labelled with any values. I did this by using a loudness meter plugin, and playing open strings as hard as possible for 10-15 seconds. I then compared the loudness before and after I adjusted the gain, and the difference was 24dB. Hope that makes sense.
2:07 There's a much better example then using pixels. With pixels you still have the resolution issue that will always stay the same unless you add pixels. Instead of 100 pixels you use 200 pixels. But, you are always limited. The better way to show arbitrary unit is o use vectors, that are also exist in photoshop. Vectors don't have size but they are mathematical unit. there for, you can make them larger then the universe and everything will still stay sharp and clear.
I would add that many people are confused and they think they have to follow the same procedure to set their gain. That part was to make a fair comparison, then every card is different and everybody will have to find where it clips. I would also add that sometimes you will have to pad, trim reduce the input gain via the internal software, if you have too much gain even when the knob is at zero.
@@sqlb3rnmeasure the S/N ratio and see if it is inline with 24bit digital recordings or way off and closer to 16bit or worse. It could also be they are recording IN 16bit, but that would also need addressing, so it's good to check this on recorded clips regardless of if they are using an amp sim or not tbh. Just make sure to ask them to leave in some silence at the start and/or end of the track, so you have something to compare against (It's also just good practice to avoid accidental truncation of the begining or end of the taget performance to be recorded).
What a great video. I’ve always advocated for zero on my input because anything above it I would hit red. I’ve always been doing this on my Apogee and UAD interfaces. Definitely going to try this out.
I'm running an AXE IO interface and with a petrucci guitar I can maybe turn the gain up like to 10 O'Clock which is like +10db increase in signal... so my neural plugins are running like -12 input and its amazing how QUIET the plugins are with the noisegate OFF rofl. Only pain in the ass obviously is everytime I switch guitars I gotta redo the settings since the pickup output differs
I'm glad I missed it. It's the same all over. Misinformation is established as truths because someone learnt a new word and don't know what it means, but still comment like a pro :). Whether you are recording audio or restoring a car or building stuff, it's always the same. :)
@@LifeOnHoth Or commenting on politics, apparently, looking at all the pro-Palestine rally attendees and the blatant propaganda and lies they swallow hook line and sinker, then propagate without a care in the world.
@@GhostNoteAudio I think it's mostly that this is the "simplest" solution that works OK for most people. E.g. my focusrite scarletts (I've got two generations) just clip immediately with any modern humbucker. The first generation could not be dialed down _enough_ , the 3rd generation one that I've got just about scrapes by at 0. I think NDSP just assumed that proper engineers would actually know that if the interface has enough headroom, you use the headroom.
@@GhostNoteAudiothey should add a SNR detection feature in their software that tries to figure out if your analog gain is set high enough before ADC. Maybe some optimisation tool that says basically: don’t play, now strum hard, decrease your preamp gain etc.
Or raise it!! Me too but also, I've been screaming "use that input gain to make your "amp" higher or lower gain as you see fit! It's like an amp mod! You can have a hotter Soldano or a cleaner 5150, which is fucking amazing when you think about it! No need for pedals! Get a boost that sounds exactly like the amp already does! Use it to make music! It's amazing flexibility"
That actually makes more sense than you realize. Most pizzas are 8 slices because they're not large enough to cut into 12. A 12 inch pizza (for instance) cut into 12 slices wouldn't be slices at all. They would be more like slivers. To get 12 slices out of a pizza, the pie would need to be at least 18 inches, and upwards to 24 inches. In some rare cases larger, but it's safe to say it was a rather large pie. When someone tells me a pizza is 12 slices, I immediately have a visual representation of its approximate size.
Another thing I'll add is that, there is no way to convert dBu to dBFS. It's like asking how to convert from inches to percent - one is an absolute measurement, the other is relative.
But why didn’t you make this video earlier? It kind of says to me you didn’t actually believe any of this until this dude made it. It’s wild to me for you to see apparent misinformation, but only deciding to speak up after someone else does an in depth video
As a scientist and engineer, this has driven me nuts for decades. The other one was Zero Latency I got into a back and forth with some people on here about *ZERO* *LATENCY* It doesn't exist. It's a compensated state achieved with programs and behind the scene (displays zero but really isn't) slight of hand. If something travels from point A to B....there will be latency as it took time from trigger to travel to sound from speakers.
I do digital audio for fun, and IT networking for a living. People always flub the difference between latency and throughput. Then they flub identifying bottlenecks as well. Say it again. There is no such thing as zero (absolute) latency. You can only match two systems (like in real time recording) by delaying the lower latency signal.
@GhostNoteAudio, from a plugin OEM: thank you!! Finally a video that explain the only correct way to set up an audio interface, and most importantly, why that is the correct way. We do not make copies of analog amps, so we do not have "accuracy" problems related to a wrong input level. We still have an handy input meter that turns blue if the level is too low, green if it's ok, red if it's too hot, but that's more of a suggestion, just a matter of taste. In the end, if the audio interface is set up correctly, then you can just trust your ears. With our approach, if all our users set up their plugin in the "green area", then they can share presets among each other and have the exact same experience, without the need of spreadsheets, measures and math.
I have a confession to make regarding this video. I have about a dozen audio interfaces in my possession, ranging from 25 year old M-Audio PCI units, to cheap Behringers, mid-range Focusrites and top of the range MOTUs and RMEs... you name it, I probably have it (except Antelope, that stuff is crazy $$$ :). I chose to use the Behringer because it has objectively poor noise performance, and a high instrument input level - Every interface has the same issues that are discussed here, to some degree, but the Behringer highlights them. On my Scarlett 2i2 2nd gen, following this advice still results in a good SNR improvement, but it was probably closer to 10dB than 19dB. The main point is, however; you are optimising two different things, and that's what many people didn't realise, and was the cause of all the confusion.
Ok, and the Scarelett G2 is also legendary for a truly terrible Hi-Z pre, to the point it could be/frequently is called defective. How about a decent modern interface? Not much difference left then is there?
@@mycosys I wouldn't say it's terrible, where do you get that from? The 1st gen was known for having very low headroom, and struggled to cope with active pickups, but they've done a pretty decent job since then.
So with an excellent interface that has a very good noise floor, going with 0 gain is still ok if you’re willing to get a bit of noise just to not go through all this?
@@devnull5109 With loud pickups, and an instrument level input that peaks around +10dBu, you're going to be pretty close to running at full level anyway at 0dB. It becomes more critical to adjust the gain when you have low output pickups, a cheap interface, or using super high gain ampsims. The main thing to understand is that optimizing the analog signal going into your interface, and optimizing the digital signal going into the plugin, are two separate things.
Fuck me. Been using the 0db gain approach for months because of those videos. This is brilliant. Tones sound wayyyy better now. Less noise, tone is less brittle, more detailed waveforms coming through. Thank you so much! Subbed.
Yes! The dynamic wave form was a very clear indicator for me that there was more signal coming from the instrument and less padding from the amp sim. Sounds much more like a real amp now.
@@MikaTarkela turn up the signal on your interface, turn down the input on your amp sim .. adjust to taste. Use your ears. Those goofballs at neural want your waveform to look nonexistent. 😂I knew from the beginning they were full of it... just common sense really.
If that were true, rock'n'roll would never have existed because no one would ever have overdriven their amp. The art is to know when it's right to be technically correct.
Thank you!! I have Hatred this advise (set input to 0db) for years. When re-amping, the signal to noise ratio issue is blaringly obvious and exasturbated when going through boost pedals before the amp. Thank you for keeping Gain Staging alive!
@@drrodopszin Ahh, true,,, lol. But he did provide reference well enough to agree it is 1" x 1". Also 100 dpi should have been at least 300 dpi in my opinion, some would choose 600 dpi, lol. But then he would have to have gone into resolution dpi, and basically explain SD to 8K lol... Ooo,,, and 4x3 instead of 16 x 9 lol...
I’m pretty sure his analogy is incorrect. You can convert pixels to inches. While things like dpi affect the actual size, there IS a measurable size. That’s why his image printed that way, which I found ironic.
Modellers have their input modeled to amps as well, interfaces are general purpose. But you still tweak input in modellers with input impedance, pads, etc
@@mserranomwell the point to set the gain to 0 is to have some sort of baseline for the guitar sim. But your audio interface must be set to INST/hi-z. Line level of course doesn't work.
@@cirisirpula152 Until you do haha. I had a guitar with Seymour Duncan blackouts that needed -6db on the input to not clip my Mooer GE300. And I have a vintage strat that needs about +2db to sound any good with the GE300.
I could never get a decent clean sound from my Marshall plugin like I could from my real physical Marshall amp, then I saw those videos telling everyone to set the interface to zero, I tried that and it cleaned up BUT it seemed silly to have such a weak physical signal so I put the level back up on the interface and introduced an eq plugin before the Marshall plugin, ran the eq flat but turned the gain way down and (I think) solved the problem in the same way you did. Thank you for confirming what I had suspected.
@@josemelrose5465 Ah yes, that's the old Softube plugin. Many older plugins didn't have an input level control. You may want to try the Amp Locker by Audio Assault, it comes with a free 1959 Plexi amp sim, called Prestige.
@@josemelrose5465 No, it doesn't. It's an old Softube plugin. I recommend trying out Amp Locker by Audio Assault. It comes with a free 1959 Master Lead amp sim. Neural Amp Modeler is another good option.
Thanks for setting things straight! The misinformation thing can be frustrating on so many different levels. I remember when the term "stems" became super popular to throw around, but it became obvious that everyone had a completely different idea of what a stem is, and none of them ever got it right. So if the client was of a certain age I would have to clarify what they meant when they asked for "stems". I had people think it meant playlists of individual tracks, the whole multitrack, pro tools session files, the process of mixing, a type of plugin, or even a mastered/printed vinyl complete with all of the album art, distribution, tour dates, and a handful of groupies. So having to ask what they meant by stems would lead to the awkward "You seriously don't know what stems are dude? Aren't you an engineer? A stem is a hi-hat sample you moron...". Thankfully I haven't run into that issue much anymore these days, but for a while there I would throw up a little every time I heard some kid say stems...
Finally! :) Thank you. Another advantage is that you record a reasonable DI signal which can be edited properly. At 0 input gain my Motu M2 gives me a nearly flat waveform.
@@EthanRom well you record clean signal. But you also use the zoom of your daw when you need it it's a keypress like every zoom it's not a "step". Also depending on the interface distortion can happen in the preamp before the clipping of the ADC.
Thanks for helping me to realize that I had not completely lost my mind by insisting all those theories of lowering the volume at the input were nonsense
It's a contentious issue because a lot of amateurs (like myself) are trying to use this fairly sophisticated pieces of equipment. And most of the equipment manufacturers provide terrible documentation. That means we all go to TH-cam to try to solve our problems. Which means...unless we find someone with this guy's background...we're probably getting bad advice. I have a TONEX One pedal. You can literally find videos from 3 different "experts" telling you how to set the Trim on it and the approach for all 3 is different. And of course, the TONE One manual is worthless because it doesn't provide much more than 2 sentences on setting the Trim.
I was not even aware this debate was going on until I clicked this video and now I'm mad because this is like Day 1 gain staging, how is everyone getting it so wrong? Great video btw, well explained
I didn't got the point of the debat too, except maybe everyone is putting too much input gain on their plugins ? I mean, everyone knows that on the interface the input must be set to maximum before clipping, right ? ... Wait... RIGHT ?
I can explain that: its how internet knowledge spreads. Itll start by one person posting a comment or a video or something in which they oftentimes literally disclaim 'yo guys, i have know idea what im talking about, but ima talk about it anyways'. Now people coincidentally not being affected by any outside influence the initial poster forgot to consider, are 'successfully' applying the 'knowledge' not noticing that the bunch of a few extra problems they got was actually caused by that 'solution'. So they go ahead with the opener 'yo guys i have no idea what im talking about' still in their minds and post a video or something themselves, further spreading the false knowledge. Eventually people not even involved with the domain of the problem will start commenting on it, further entangeling truth, believes, self promotion and stupidity into one big mess. This is how you end up with such discussions in a nutshell.
Yeah. I didn't know this was some sort of argument. Also when switching guitars , I have to fool with input gain for each one. When I change modeled amps and effects same too. mostly on the sim instead of the interface though. I don't know the science behind it, I just ...do it.
Thank you so much! The whole 0dB input always sounded incredibly dumb however the other alternatives presented didn't yield better results. This is the definitive way to do it
@@alrecks619 24 is already more than enough. Hell 18 bits would already be more than enough (17 bits is twice as many values as 16, etc) there are tons of interfaces and field recorders with 24 bit recording and no gain knob at all, because they simply don’t need them. The idea that you need as hot a signal as possible is one that originated with tape, held traction with 16bit recording and noisy preamps, but makes little to no sense with most modern equipment. This guy demonstrated a well known and tangential audio concept with his potato of an interface from Behringer. His preamps are garbage and his adc is likely working at 16bits. It is important to know how to minimize your noise floor for your particular gear, but the conclusions he’s come to based on his simple experiment are…. Wild
@@threepe0 totally. He is theoretically correct, but the kind of optimization he is advocating for as something essential seems to have already been taken care of at the hardware level in most decent modern interfaces, to the point where the difference (even though still exists) is negligible for most use cases.
@@alrecks619 Even 32 bits audio cards have a noise floor. Now, is it as obvious as a 70s tascam mixer? No, but it's there, so it's good practice to adjust the gain as high as possible *before* the conversion.
@@threepe0 for a modern metal guitar tone the signal chain might include compression, distortion pedals, overdrives, hi-gain amp sim, tape distortion, transformer emulation, and a boatload of other non linear effects that can easily reduce the SnR by 10s of dBs, so doesn't it makes sense as a general principle to optimise the SnR at the input even if the self-noise of the ADC is not that high? Not to mention that when using a noise gate it results in a better sounding sustain for long-sustaining single notes, where one would like the decay to sound as natural as possible.
This is how I've always approached setting levels for amp sims. Didn't have a handy-dandy chart, but by ear has worked well enough. It's the same as how I'd record anything else. If I'm recording vocals, I don't turn down the input gain on my interface to match the level the compressor plugin and EQ plugin I'm going to use wants. No, I record a strong signal on my vocal and then adjust the input gain on the plugin or the clip gain as needed so my compressor is happy. Digital recording basics. (Yes current technology is more forgiving that of years before, but still.) Good video and nice technical explanation. Looks like you're a fellow lefty as well 👍
Appreciate this information. I would propose amp sim devs add the basic process and information to their user manuals as well as include either auto-gain function or a “green light” for the setting the input gain knob in their plugin.
A lot of the issue stems from ToneX users, where the physical pedal version (opposed to the plugin) has a global input level that is both the AD level and the input level feeding the captures. So if you optimize the input for “more bits” you can’t then lower the optimized input signal before feeding the next part of the chain. So a lot of the blame should go to IK Multimedia not to the users scrambling to find workarounds. In a perfect world you are absolutely correct and for DAW/plugin use one can achieve this even with a Behringer as you showed. But lots of players don’t realize their gear is sabotaging them inside the box.
I wouldn't consider the global input level to be "both". The AD conversion happens before and the pedal itself has an instrument input (like an amp). That's why you shouldn't adjust the global input level when switching guitars. Then what would be the point in having different pickups. Speaking of tonex the important thing is to set up the pedal so it corresponds with your experience in the app where you download and tweak your profiles.
I saw that a few weeks ago. I didn’t elaborate in my response, but the ToneX’s instructions tell people to “turn it up before clipping then turn it down a bit” but there is no scale, no reference. I’ve been setting my presets for plugin and physical pedal using a signal generator. Most players won’t be doing that. And then how was the level set when a capture was done? Was it correct? Lots of tails to chase.
THANK YOU THANK YOU THANK YOU THANK YOU I was always super skeptical and my first thought when I saw these videos was "yeah, but what about the signal-to-noise ratio?!" Having the signal so close to the ground noise is such a weird suggestion! But I said welp, these people seem to have done crazy research, but on top of that a lot of plugin developers confirmed this. So I guess that must be true? Turns out i've been having the noisiest recordings ever since then😭 Especially with some of my very low input guitars. tho I swear if another video comes out proving you wrong i'm gonna be so done with the internet hahahahhahaha 😭 also nice theme ❤
I wasn’t looking for this, I didn’t need to know this, I own / use none of the equipment or tools you talk about, it just appeared in my recommended feed and .. even I understood this perfectly. You may have a like sir
The quantity of misinformation I see thrown around the internet, especially when speaking about gainstaging and impedance is overwhelming. It's depressing to see how technicians (or "sound engineers") with a lot of "experience" are much more prone to believe in audio superstitions than to use their own heads. As an electronic engineer, I can't understand why people mess up so much, as these subjects are based on basic and simple concepts and are the basis to be able to work with any analog audio technology. This video is well made, straight to the point and simple, I hope it will be understood by most people
Just a side note, optimizing the input before clipping might introduce some natural compression from the pre AD circuit. So, I still suggest to try it out and use your ears for best result. Also a buffer guitar pedal (like a BOSS tuner pedal or other good non true bypass pedal ) might be a big benefit to noise and impedance matching! Cheers and thank you very much for your "guts" to stand agains the (stupid) internet! I would have imagined my self doing this 20 years ago, but now i'm just too old :D
Just tried this out! What a difference it's actually made to the noise! Thanks for this. I do wish amp sim makers would just implement the level indicator like ML Sound Lab where you just chug turning their input dial until it goes green. It'd save all this headache! haha
I did the same and was amazed at how much of the higher end fizzy sound was gone and the tone was so much tighter and defined. Amazing so many respected TH-cam musicians hadn't figured this out. There really needs to be a better standard way these plugins all measure input and adjust for us.
@@BeGoodBe On the ML Sound Lab amps? Yeah, all of them have it. Turn the input gain dial up until a little light comes on, you'll see it. That's the sweet spot for the amp.
I'm very glad you used that chart. It's pretty accurate. I've found my MOTU interface's DI is my favorite since it can boost and cut. Some of my really hot pickups require a cut, where lower output ones need a boost. Used in conjunction with the chart it's REALLY a game changer for amp sims.
Thanks for clarifying. Been doing it "your way" for ages, & the recent discussion all over the interwebs had nearly convinced me that I should instead be recording about 12db quieter than I have been, rather than just normalizing the file post performance to my desired peak after tracking. The crazy thing is that I've known better for ages, but when enough voices say the same wrong thing, it can give you pause. The noise floor issues associated with low input signal seem obvious, but when a bunch of "smart guys" say noise isn't an issue with low input gain... Sheesh. Nice video!
I'm not a qualified audio engineer - just a producer. However, i used to think that it's obvious: the more clean boost you got = the more depth (resolution) and more signal/noise ratio you achieve within the interface. It's strange that someone claims the opposite as a community consensus, because all the A/D theory proves that. Thank you for that video. Musicians possess a humanities mindset rather than a technical one, so there's a lot of honest misconceptions like this - it will be very helpful.
@@andymerrett Since I am not an English native speaker, I didn't expressed myself precisely - I meant that musicians are humanities more often than technicians. I have 10 more years of experience of working with artists, so i have some observations. Excuse me or you're welcome, i dunno
I don't know that musicians really skew any differently than the general population when it comes to technical mindset. It's just that _most_ people aren't technically minded, and those that are tend toward fields where this is an asset. But you still get plenty of technically minded people (like Tom Scholz) who also happen to be great musicians.
@@mal2ksc I agree with your point in general, but i don't think that there is a discrete dualism between these extremes. Personally, i have a technical and humanity mindset at the same time - this is common among producers, mixing and mastering engineers. There's no way you can mix your music without any abilities in math and physics, for example. However, to write good music, you need a strong humanitarian background. Music written by engineers is often boring and doesn't excite in any way, because most of good music is about feeling rather that knowledge. In my opinion, only a combination of cultural and technical development factors gives a result. The music industry got it long ago, which is why artists work in tandem with engineers - without them, everything will sound bad. Without the former, it will sound good, but it will not catch. However, it's not a rule - just a trend that becomes mainstream.
Thank you for this! Though I've maintained that we should digitize at the hottest level w/out clipping to capture at the highest resolution, I've gained correct understanding for the justification behind it.
As an engineer who deals in communication systems, and a musician, this made me feel good. Data driven decisions based on solid engineering principles. Great job.
Thank you very much for making this. I was confused about the whole thing this year. I especially hated having such quiet DI recordings in my DAW to work with
This is the recording equivalent of flat earth theory. A bunch of people start believing something stupid for no reason and now we have to re-prove what has already been a proven and established fact for decades.
@@Rodriastral666 while the definition of globe is the earth that surely makes sense! Lol Globe can mean a round object so the globes in a classroom is a spherical representation of the earth and not exactly the pear like shape of the literal shape of the earth. Nobody said the globe maps are exactly.
For all these years I never left a zero gain on my input, but I always corrected the gain on the plugins by ear. Now I've learned how to correctly do the adjustments with precise levels. Will try it out very soon, thanks for the tip!
I’ve been stuck on this problem for a while. I feel like a lot of us beginners have trouble understanding gain staging which end in us getting stuck somewhere along the way in the mix.Thank you so so mich for this video! Subscribed
Really good video, subscribed! Well laid out. Basically, the "set it to 0" approach is just "the easy way out" for the ones not knowing what they are doing. I have to admit that I belonged to that crowd for quite a while, or differently said, I did not care for it. The issue with John Cordy's, and even way more so with Rabeas video, is that they simply are parrotting out what others told them, to make a click-baity video. They do not explain the technicalities behind those approaches, but they just say "Yeah m8 you dumb turn knobz down wow such sound lol" On the other hand, Ed Sokolowski's approach to propagate for setting the gain to 0 (which I would say had the most impact, since John Cordy for example also has grabbed this idea from Ed) has a different motivation: To have a single common denominator and a clear reference from which to work on moving forward. For people to being able to properly assess at which levels they are making their profiles. If you gain up your channel via analog means, you have to evaluate through measurement how much of an increase you are pushing for, before you reach the clipping threshold. A guitar signal due to it's inconsistency is not a good reference to check on that, and not everybody has a sine wave generator they can plug into their interface (to get nice before/after values of their analog gain staging). Also, the Behringer UPHORIA 1820 with it's +18dBu headroom has plenty of room to utilize before you hit the clipping threshold - with the +10dBu of my audient iD44, that might be a different story. However, this is a great approach which I will put to the test on my interface. I mean, I personally did not have too much complaints on noise for my taste by setting all to 0, but I do CLEARLY see, as an fellow electrician (although not within the field of digital signal processing, I shifted sideways nowadays to networking technologies and Crypto-IT), that everything you stated in this video is technically correct and sensefully the right thing. THANKS!
Thanks a lot for this clear explanation. It reminds me of another analogy with the digital photography. There is a lot of difference in the final image file, if I underxpose a shot and then I add luminosity in post production, than if I overexpose the shot to the maximum possibile level but without burning lights, and then I reduce the luminosity in post production. It is always related to the signal to noise ratio and to the ability to exploit, at the best possibile level, all the available bits.
As a person who mainly just plays guitar and doesn't record much, the moment I heard about setting input gain down in general, made me slap the knob to 0 on my dac and bam. suddenly all the presets I avoided using straight in my plugins sounded like how they were meant to sound, and so I just left it like that for ages. Over time I got bugged by some of the noise and started to compensate as you showed in this video, but I wonder if some people just found it easy because it is one single adjustment, vs adjusting the input gain for each plugin they use...
As a recording engineer (who also plays guitar) I just instinctively gain-staged the input of amp sims like I would anything else. Glad to find out I was doing things correctly. Great vid 👍
Dude, you really deserve if not a gold then a bronze monument for sure.. no irony.. I listened to advice for years, shoveled through a bunch of printed material to now find out and hear, see such an obvious thing. Thank the gods and engineers!
Good presentation on this complex topic. And coincidentally, right after seeing this, I watched a Crow Hill video where they called out your Ghost Note Audio Conductor MIDI controller (nice!). I think some addendums are worthwhile re: standards, and whether the optimization you prescribe is the "perfect" being the enemy of the good (or even of the better than "perfect"). First, re: "There is no accepted convention for what is considered peak level for an instrument input." Well, if by "accepted" you meant "universally accepted," I suppose that's right; anyone can ignore a standard if they wish. But in a video attempting to be "rigorous" about this topic, it's probably worth noting that there are in fact standards; it's just that (1) there are competing standards that differ, and (2) there is no enforcement mechanism, so interface manufacturers are free to ignore standards if they choose. In Europe, EBU R68 specifies that 0 dBu (an input power level) should correspond to -18 dBFS (decibels with respect to full scale, a digital level, "unitless" in your terminology, though "relative" would be at least as good a term). In the US, SMPTE RP155 specifies that +4 dBu should correspond to -20 dBFS. So there's a 6 dB difference in the recommended calibration just from "crossing the pond." Wikipedia's "dBFS" entry indicates that Japan, France, and "some other countries" recommend +22 dBu at 0 dBFS (the relevant standards body is not identified). Moreover, +4 dBu is considered to be the "professional" reference level. Some interfaces adopt the lower consumer reference level of -10 dBV instead (and the difference between the final "u" and "V" is important; these reference levels are not 14 dB apart!). So, there are some standards, but their multiplicity means that "There is no accepted convention..." is not a bad description of the situation in practice. The datasheet examples you give are sadly imprecise and thus bad examples (this is not your fault, obviously). A level is called a "maximum" level without saying what that means. Is it describing analog performance of the preamps, specifying when some point in the audio signal path (e.g., just before the ADC) hits some target distortion level (say, 0.01% THD)? Is it instead where the ADC produces the maximum digital signal? A precise datasheet would say exactly what is meant. For example, for my RME Fireface UC (which lets you switch between +4 dBu and -10 dBV input reference levels), they provide a table specifying dBu above the reference level that corresponds to 0 dBFS. For the +4 dBu choice, 0 dBFS is reached for signals at 13 dBu, "the latest EBU recommendation" (which evidently was 1 dB different from the current recommendation). If "dBFS" does not appear in the datasheet, you can't be sure how the analog and digital levels are related. "The accepted wisdom was-and still should be-to set the input gain... as high as possible without clipping the signal." Your discussion of gain staging seems to me to date back to the days of 16 bit bit depth, with only 96 dB of dynamic range. Part of the motivation for moving to 24 bit and higher bit depths was to make it unnecessary to push levels very close to 0 dBFS when gain staging. With the headroom on hand in an ADC with decent, modern preamps and 24 bits or more of bit depth (144 dB!), it is not necessary to set the input gain as high as possible to avoid clipping in order to get a very clean signal. That was always a nuisance, because in the studio (and even more so on stage) the talent would very often perform louder during an actual performance than during a sound/level check. With 24 bit converters, you can set your gain to be well below clipping and still have an imperceptible noise floor-and plenty of headroom in case the performance is hotter than expected (and to enable EQ and other processing without having to attenuate the signal digitally later). You call 90 dB of range "relatively good." Actually, if that's really the signal to noise ratio one is achieving, that's spectacularly good. Unless your listening room is an anechoic chamber, you won't be able to use that dynamic range-it's the difference between the lowest perceptible sound (for perfectly healthy ears in a completely silent environment) and a sound level that actually causes hearing damage. So: No need to push close to clipping; in fact, your recording and mixing life will probably be easier if you give yourself 6 or 12 dB of headroom when you set the analog input gain. But do follow the rest of the advice here, so that whatever analog gain you use ends up hitting the amp sim plug-in at the designed-for level. Or, on the other hand, just do what sounds good to you. There's no law (or even a standard!) saying you need to use the amp sim the way the manufacturer intended! -Tom
Good points. A few thoughts- Not all equipment uses the same full scale voltage reference, so while it may be useful to have a defined level that is "compliant to" some standard. But, practically, I think it would be better for manufacturers to clearly state what the full scale level in dBu or volts (or both) is for a given interface and we all can go from there. Crest factors (peak to average ratios) of the signals also affect where we might set the "average" level, and of course the averaging method affects that back off, as well. More clarity rather than less information would be helpful all around, I would hope. By the way, 90dB dynamic range, about 15 bits (ENOB), in a single source channel is probably not so great when combined with other signals and additional processing. It's about what the old redbook CD standard can do, though. Cheers
You're conflating thermal (electric) noise and quantization noise. 24bits does reduce quantization noise, but you still have the best SNR when setting your gain to just below clipping. That's basic gain staging. Of course you always keep a little headroom when setting your gain. Setting a proper gain is not very complicated either..
@@nolyspe No, the point is that most interfaces that support higher bit depth than 16 bits also have quieter preamps and converters. It would be somewhat pointless to put a preamp with only 90 dB of headroom (say) in front of a 24 bit converter. If you compare the preamp specs of modern interfaces with older interfaces that were only 16 bit, you'll see modern interfaces typically have quieter preamps.
Yeah I found the values for the amp sims in the spreadsheet. Great video and of course everything here is correct and I agree with completely. All I would really add here, is that most interfaces with around 12dBu of headroom at 0 gain are going to produce a signal close to clipping when using humbuckers. So in these cases, it’s already going to be very close to an optimal SNR, and you can easily know your headroom amount (in order to calibrate your specific plugin). The other factor, is that as long as the noise of the interface is sufficiently below the background noise of the pickups, there will be little to no improvement in optimising SNR - increasing preamp gain will also increase EMI/RF interference from the pickups. On noiser interfaces like Behringer, it’s absolutely worth optimising SNR. I think many interfaces in 2024 will have less issues with noise and too much headroom. If noise is not causing issues, for the average home guitarist, easily knowing the level of headroom they are using will be more useful than a louder signal. In an ideal world, the SNR would be optimised, AND each user would be able to easily determine their dBu->dBFS ratio. If SNR is essentially “solved”, then the easiest way to know the dBu->dBFS relationship is to turn the gain to the value provided in the specs.
Agree with all of that! I intentionally used the Behringer because it highlights the issue. It both has poor noise performance, and an unusually high input level. These two facts conspire to make it particularly sensitive to this subject matter 🙂 And thanks again for collecting all that data!
@@GhostNoteAudio For sure! I think generally companies like UA/Audient/Focusrite/RME etc are doing the right thing with “standardising” 12-13dBu as a sufficient level of headroom, while still retaining good SNR. If the maximum is below that, then it becomes easy to clip. I believe some UA and RME interfaces have stepped gain controls which makes it easier to optimise further without the need to measure the headroom manually. Also guitar modellers like the Axe FX and Helix have dedicated DI inputs with extremely good SNR converters and can be a good option for guitarists who might already own one and have a noisy interface.
I noticed that with a Gen 3 Scarlett I just started using! On my Tascam model 12 I would run the preamp at around 11 o'clock where it has 0db marked and my signal would peak inside reaper at around -6. So I was quite surprised that on the Scarlett, turning the preamp all the way down resulted in a similar input signal level
i was about to comment that "finally someone understands gain-staging" and then that whole section on gain staging began. Love this so much. Anytime someone mentions gain-staging incorrectly, or says "gain staging isn't even important now with super high bit rates..." I'm sending them this video
That was a cool tutorial and explanation... I may need to watch it a few times to fully take it all in but the basics of gain staging makes total sense to me...
for the lazy: 1. hit the strings as hard as you would playing and then turn the interface preamp gain up til it clips and back it off a little bit 2. open up a high gain patch on your plugin and play chugs whilst turning down the input gain on the plugin until it sounds good bc that's literally the point of music. inb4: "it won't be the same" yes I know, but also I probably don't have the same pickup as the preset maker so I'll actually get closer to the intended tone this way.
The point is "sounds good" works up to a point. But people don't want presets to sound the same, they want the simulated amp to react as it would IRL (as close as possible) - and that means if the real amp has a gain knob and you put it on 5 you expect the same gain as if you plugged the same guitar into a real amp. You have to note too that some of the software (e.g amplitude 5) have models where it's not got enough gain (compared with the real amp) if you put the input gain on max. It's a mess - and the input gain on zero + spreadsheet was an attempt to solve the mess for some plugins and interfaces (and really are people getting that much noise from good interfaces? Probably not)
I would absolutely care if something marketed as "this thing" behaved like that thing or not. If it sounds good but doesn't sound like the thing you're calling it, maybe call it something else @@alessandroproverbio3411
@@michael1 I hear ya, I just don't care about how "accurately" the plug reacts to a real life amp. If it feels how I want and sounds how I want everything is doing it's job.
Never use a distorted plugin patch for this, you must use a clean patch! Why? select a clean patch and crank up the plugin(!!) input gain. Hear it getting distorted?. Now back it off till it becomes clean, now back it off a little to give you headroom. You will NOT hear this with a distorted patch, thus you will NOT know where to back the gain off to...and if you don't know that, you will NEVER guarantee the correct sound of that patch. I'm very surprised this isn't recognised in the video. When change to a higher/lower power pickup, you'll need to optimise both the audio interface and the plugin gain settings again.
Man, that's one of the biggest, most useful videos ever watched on TH-cam! I'm practicing this thing right now on my plugi setup and my guitar tone has never been so much focused until now! Thank you so much. You've got a new subscriber, content like this is real gold. Beers & pizzas from Italy.
THANK YOU. Also have an electrical engineering background, have worked in DSP, been doing this stuff for almost 20 years. The misinformation drives me crazy!
I have office tomorrow. Its past 23:00 here in my time. Just as I was getting ready to go to sleep I saw the thumbnail. Clicked on it. Was not disappointed. Halfway through your video I was like "holuy shit I need to try this": Told the wife not to disturb me for 35-45 mins and that I will work from home tomorrow. Took my like 5 minutes to do the steps you mentioned in the end. Holy shit dude my life has changed.
I used to work for Audio Note, very arguably the top top end of hifi, and we’d spend a lot of time deprogramming people. Most so called audiophiles aren’t musicians, don’t see live bands, jazz ensembles or orchestras. They talk about statistics, efficiency, freq. res. and dynamics but never soundstage, ambience, acoustic energy, character, tone, catching phrases musically etc etc The average audiophile treats hifi like a digital camera and we all know that having a camera is just one tiny step on the long road to being a great photographer
Great video! I definitely thought I was doing well in following the 0dB on the interface and adjusting in the plugin. One thing is still unclear to me though: how do you know by how much I increased the input gain on my interface? Since it's just a physical knob I can't read it. Do I use my DAW to check? I would need a reliable source of consistent level (so probably not me strumming my guitar as hard as I can) to be accurate in the amount of input gain I added. Thanks!
Being a musician and an electrical engineer, I am happy to watch such a vidéo. The «zero gain» vids made me sad when they appeared and I felt alone explaining exactly that in their comments.
I've always recorded guitars as loud as possible without clipping, and then adjusting the ampsim input accordingly "by feel". It just seemed logical to me to use the interface's dynamic range as good as possible, and adjust the gain afterwards before plugins, especially with ampsims. So I'm glad it seems I did it right without ever thinking too much about it :D Thanks for the great video, hope it helps a lot of people!
I'm genuinely surprised that this is even up for debate, it's like an argument about recording digitally from 20 years ago. I guess youtubers spreading terrible information has got people convinced.
I had never heard of the "set it to zero" bad advice until watching this video. I totally agree with @GhostNoteAudio and his maths and science. That not gaining your input ever advice is just wrong. The zero to worry about is 0 dbFS off your DAC, just stay below it, but let your loudest peak approach as close as you dare to. "Set it to zero" ?? REALLY? Why have a gain knob then? Leaving it down puts your signal closer to the self-noise of the interface circuitry, although turning it up will bring the noise of the source up with it, but it will still be at the same ratio to instrument signal as you are playing. Now all watchers of this video who didn't know before can enjoy cleaner recordings of their guitars. Yay!
I’m baffled that people thought zero gain on the interface was the solution. Am I getting old or something? That’s basically recording guitar DI 101. Thank you for putting this together and bringing some sense into the world 🤘🏻
He used a shit interface. You have to understand the problem they were trying to solve first. Component based amp modelling software has dials like the real amp. People want the simulated amp to work the same as if they plugged their guitar into the real amp. That means, e.g if the real amp has a gain knob from 0-10 that they want gain on 5 to sound the same as the real amp with gain on 5 - for their guitar. Obviously this also tends to make presets marked 'clean' sound clean (modulo that different guitars / pickups etc will affect presets) But the problem is now twofold (a) interfaces were saying to set gain 'just below clipping' and on that setting many of the plugins were massively over gained - because that's not how the plugin authors envisaged the gain to be set and (b) Many of the plugins, especially the ones that come with tons of different amps, are a complete mess in the sense that if you pick n of their models you need to use n different input gain settings to make it match the real amp - and unless you own all the amps how do you begin? Plus some of them the gain is only correct if you turn everything up to max, i.e the clip lights can be on but the gain is still too low - this amp modelling stuff can be an inconsistent mess. So problem (a) was solved by at least making people aware that their input gain matters and giving them settings that work for a bunch of plugins and interfaces. But saying "input gain on zero - set input gain on the plugin from the spreadsheet" has confused some into believing 'input gain on zero' is somehow key to everything or a new religion. Obviously it's not. The big clue is - why would you have an input gain knob if it weren't supposed to be turned? But it works - and it works far better than this video suggests because he used crap hardware to exaggerate a problem. Quite disingenuous really. Problem (b) is not solved and there's no real solution. You really want all people who are simulating a bunch of amps to do so in a way that you only have to set input gain once - where you have things that are more open and 3rd parties can add, well, you can see why it often ends up a mess. But the noise in this video was a bit silly. It's like I'm using a pod xt for my interface - it has no gain knob to turn but the biggest problem with noise I have is if I point the pickups towards the computer running the plugin. Plus of course this video side skirts the issue to get off on calling people wrong. He doesn't offer settings that we can use with our plugins - but suggests we refer to the same spreadsheet these "wrong" people created. So it's like he's saying "They're idiots and are wrong" - but, wait a second, their spreadsheet is the information we need to make our plugin simulated amp gain match (as closely as possible) the real amps that are simulated and the gain on zero really doesn't make it that noisy - so they were actually right. The only people who are wrong are those who think you absolutely HAVE to have your input gain on zero - of course that's not true. Of course you can set input gain on your interface how you want and then adjust in the plugin accordingly - but that misses the point of the whole exercise which is trying to get simulated amps to behave like the real amps.
@@michael1 Well, the problem is actually threefold: (c) 0% preamp gain would result in a different level from manufacturer to manufacturer (sometimes even different between various model lines of the same manufacturer), so it still can't be a baseline.
I feel as though gain staging was understood more in the analog days. The digital age has not gotten the lesson. I hear incorrect gain advice from young sound engineers quite a lot. And when i look through their mixer scenes, oh boy! Gains all over the place. Thanks for sharing this great lesson!
Youre right, but im also shocked by how bad the pre/converter noise on the Behringer is to make it matter that much. Definitely will only be using mine for line signals.
@@GhostNoteAudio Just grabbed one for $200US new, but it will definitely only be doing the synth connecting duties as an ADAT expander i grabbed it for XD - 20yo MOTU died, i doubt it will be worse lol. Theyre really popular in modular groups as expanders cos we need so many channels, but HOT HOT HOT signals only XD
Amazing that so many peoples takeaway is that "none of this matters"... when right at the end of the video you literally set the interface gain to 0, find the accurate plugin correction and THEN offset the interface gain and plugin gain if there's more headroom to be had while not clipping the signal. The comments are a fascinating insight to people wanting to believe that plugin accuracy doesnt exist and they "finally found someone who agrees with them" so they're piling on. Had they actually watched your video they'd see you're agreeing with everything JnC, Rabea, Ed and myself have been talking about but just taking it one step further with SnR optimisation (which should already be obvious to people but is a great thing to point out).
Absolutely spot on and very well explained. I am also amazed at how much rubbish is pedaled on TH-cam from TH-camrs trying to appear professional but are spreading ignorance. Long before there was "digital", we'd max-out analog desk inputs just before the signal on the VU meter hits the red or distorts. That wisdom is also true for digital.
I had no idea this whole conversation was going on - and I'm glad I missed it. I learned gain staging (correctly, as you prescribed) as a fundamental (which it certainly is). It boggles the mind that people would somehow start getting it all wrong, when it's such a fundamental part of audio production. Thank you for this video.
7:11 Judging by your footage you activated BOTH instrument and PAD knobs for your input. The latter attenuates the signal. So the advice about 0 gain might sill hold on simpler interfaces / non padded inputs the INSTRUMENT / Z inputs are already delivering max (or rather optimal) gain for PASSIVE pickups without clipping and changing gain knob even one degree results in clipping (at least with interfaces and guitars that I had focusrite and motu).
I noticed this also. 18 dBu normal and then the PAD adds at least 10 more dBu of headroom. Maybe he will make part 2 with modern style interface and humbuckers with 12 dBu of headroom.
So, I tested your method instead of the "0 Gain method" I was previously using, and yes, it absolutely brings down the inherent noise floor of the interface. One has to be however mindful of what signal or noise you might be boosting when going with the optimized SNR setup you recommend. In my case, I get some crippling interference through my computer, which I dislike more than the inherent hiss, which nicely masks some of that nasty noise. But it definitely is awesome to know your tools and just learn another point of view every now and then! Would my PC be further away from me, then this would be a very valid approach to get the best out of all! I am SUPER HAPPY to see that this video is getting quite some traction and your subscriber numbers are skyrocketing. Off to 1k, buddy!
Agree. Better to have a higher homogenous white noise than an angry "eeeeeee!" getting picked up into the signal. Disabling the CPU turbo boost helps with that high pitched sound.
You don't set it to max level. Usually for microphones and stuff you set it to around -12db or -9dbfs. But for most interfaces a electric guitar can already be -6db to higher than 0db(clipping) with gain at zero.
@@johnyang799 with no gain it peaks around -9dBFS, so there is something to be gained. Also, when I set the speed of the CPU cores statically, then the process explained here works very well, because I get great SNR, but also very less crippling of the CPU noise.
@@PippPriss Did some measuring and testing with a 1kHz sine tone against the pickups to level match and it seems like adding gain on the interface doesn't really add more of that high pitch whine or other noises into the signal. In comparison, by leaving the input at zero you get more noise that ends up masking more of the whine but it's level is the same. Ended up adding 5dB on the interface input, it's not that much but the SNR gains are clear. I guess I am back to the old method of targeting around -6dBFS for the extreme peaks.
This makes so much sense now! I think the trend of sticking high gain pickups in whatever guitar and turning gain knobs up has influenced the 0 gain input dial misconception. Some guitars won't require much of the steps you explained in the video. But a lot will definitely need this and benefit from it. Thank you for the thorough explanation!
27 วันที่ผ่านมา +7
In my modest opinion, the story is a little more complicated with sim plugins, which are supposed to behave exactly like the real counterparts. With this approach (ie, maximising the signal/noise ratio for ANY type of input), the amplitude of a signal generated by a Strat as observed by the plugin in the digital domain will be roughly the same as the amplitude of a signal generated by a Les Paul. So I think there are two options here: either we follow this advice, but then we'll have to use different level adjustments in the digital domain to compensate for the different instruments, or we give up a little bit of noise floor and adjust the interface gain for the highest output pickup we have, and keep the level adjustment fixed in the digital domain. In any case, however, I think the sim plugin manufacturers should report a reference level that the plugin expects to see *for a particular pickup*, otherwise we are still missing one aspect of the equation! (and then obviously you have TONEX/NAM, where the capture process is not standardised, so those numbers should be provided by whoever produced the captures).
I set it this way to my loudest humbucking pickup. And then just leave it at that even if I plug in a single coil. So the lower signal of the single coil behaves how it should with its lowered output. I think that makes sense?
27 วันที่ผ่านมา
It does make sense. To be 100% accurate, we'd still need to know the reference input level that the plugin expects *for a particular pickup*, but the manufactures don't go down to this level of detail, so using the reference levels that were collated in that spreadsheet should put everyone in the right ballpark... and we can finally play 😀
Really interesting point about different types of pickups! Plugin companies providing reference levels for different guitars/pickups seems like a pretty cool idea to me, especially if you're trying to sell people the "sound" of something in the real world they can plug their guitar into and compare it to. You don't want them to think your plugin doesn't measure up to the real thing simply because the input levels need to be a certain way to reproduce the sound they're familiar with, but nobody has this info. Since plugin companies don't really do that, I'm wondering how would I deal with this if I was using the method described in this video to record more than one type of guitar, and wanted the experience you're describing with a plugin reacting like a real amp in that it behaves differently when fed different guitars and pickups, etc.... I think I'd try getting a setting dialed in on the audio interface using the loudest guitar(s)/pickup(s) out of the ones that I have. Then, I'd make a mark on the audio interface, take a picture of the knob's position, make a note, or something so that I could return to that setting whenever I input any of my guitars into the audio interface. I would dial in the settings on my amp sim plugin(s) using the guitar I play the most and the input level I just decided on with the audio interface. I have not tested this (of course), but I'm guessing the difference in the noise floor created by the different signal levels when comparing two guitar pickups might only be a big deal if one of the pickups is noticeably noisier than the other, or if one is passive and the other is active. Ok, enough of this. I'm gonna go record some things.
26 วันที่ผ่านมา
Jason Sadites just posted a video about this - he seems to agree with me 😀 But yes, fundamentally, to avoid constantly adjusting the input gain in the digital domain, the analog gain on the interface will always be a tradeoff between signal/noise ratio and the ability to pick up any guitar we have and avoid clipping. And yes, the signal/noise ratio lost (worst case is for lowest output pickups), if we are not using garbage interfaces, it's likely to be... in the noise (no pun intended).
lol best thing i found was to barely raise it a little at a time until it sounds nice. 0 sounded upsettling weak for me..ended up around 2.3 which ended up being what neural dsp says on their site. (for QC as an interface at least) just go in small increments
It's not hard... 1. Optimise analog signal going in to interface, by increasing it to just below clipping; 2. Compensate digital signal going to your amp sim (i.e., in your DAW, or on the amp sim itself) by turning it down by the same amount.
@@matthijsheblynot quite. 1. Interface to zero 2. Digital gain plugin to chart (interface/amp sim matrix) 3. Increase analog gain to under clipping. 4. Reduce digital gain plugin by the amount you added to the analog stage. (If you added 10 dB in step 3, subtract 10 dB from the plugin you added at step 2.)
@@matthijshebly not true. The whole point of having the chart is to match calibrated levels. That’s what the digital gain plugin step is for. If you leave that out, you’re right back to seeing things by ear.
0db is the dumbest thing I have ever heard! I was not aware this debate was happening, I thought everyone just turned up to the right level 😂 Why the hell would someone want to feed a weak signal into an amp sim?? The internet is insane sometimes ! Great video !
Yeah, I'm not understanding where this idea came from. Like... it's common knowledge to turn up to red and then roll back to green to find best signal strength
@@Citizen_J only thing I can think of is that they have had the software gain up too high, maybe software builders have even been setting up this way so it can handle the 0db “geniuses” 😂
@@Mr-Foad It could be that this information came from metal players who have high output pickups, so they're slamming the input hsrd enough that turning up isn't necessary. I know my passive bass doesn't work like that
I've been trying to clean up my distorted guitar sound for years with out luck and just learned to live with it as it is (since I'm not releasing it publicly). Now I can't wait to go revisit all I have and dust it of. Thank you so very much for this.
As a returning intermediate player, this was incredibly helpful and settled a lot of nonsense online. The difference is in the signal and this is it! Thanks brother
I think im an idiot. Can someone help me and make sure im doing this right? I have a Focusrite Solo 3rd Gen, Reaper DAW, and Neural DSP. (Step 1) Set gain on my Solo to 0 (Step 2) Change the input gain knob on my Neural DSP to 0.3 (Number of 3rd gen Focusrite on spreadsheet) (Step 3) Play my guitar and turn up the gain knob on my interface until it doesnt show yellow or red (Step 4) ????????? This is where im confused, what does he mean by "Reduce digital input gain to compensate" Not sure why this is confusing to me?
He's saying to raise the gain on your interface until its just below clipping the taking the amount of gain you turned up on your interface and subtracting that from your input in the plugin (I think)
@@JP-ir7uw If your interface only has lights and no db meter:(howewer you could also open a daw and watch it. Anyway if you want to be precise: Record while turning the gain from 0 till just below clipping. Measure the recording. The differrence in db is what you subtract. So the beginning when it is set to 0. compare it to the in the recording part where it is set just below clipping. If you still don't know How to compare? Meter in your daw.
How do I know how much I've increased the gain by on my interface? By just estimating the percentage of dial turned up into percentage of maximum gain available from the interface?
The chart he shows has your Interface (I have the Scarlett) on the left. I assume you look across the top of the chart to see which plugin or DSP(I have a Helix) . Looking down the helix column the scarlet has 0.5 which corresponds to 12 o'clock . I may be wrong but it kinda makes sense to me lol
I tried this using NDSP with UAD volt. I used prefader metering in Logic to see what my guitar peaked at with interface gain set to 0. Then I checked the peak after raising interface gain to just below clipping. Get the difference of those values and turn down the plugin input gain by that value. Say prefader 0 peaks at -14 db and prefader under clipping peaks at -4. Turn down input gain in plugin down to -10 and add back the adjustment from the chart (+.3), so -9.7 in my case. Not sure if there’s a more precise way but would be interested.
It's a useless information. Just push the gain until it clips, then turn down a bit. What you have is the audio file in its best possible shape, and how much gain you used to get there is irrelevant from that point on. Every intrument has its different output, every amp sim has its design. Use your ears from that point on like you would do with any analog hardware. If it clips or distorts, the input is too high, just a notch below that you'll have the best ratio.
Thank you for this great video. The problem with those influencers is that people give credits to those assholes without doing their own researches. People want the influencers to think for them.
I think a lot of people start recording on Focusrite Scarletts On those interfaces (at least 3rd gen), simply setting the input to instrument mode and turning the gain all the way down gives you a decent signal for electric guitars
For those who want to keep their workflow, and don't have -20 dB on input in ampsim software: u can always use simplest "gain" plugin to trim signal to correct levels for further processing. It still amaze me how many of myth are functioning in pro-audio sphere, it is just nuts.
it's understandable as most people don't know the technical details of how this all works behind the scenes. that's why the devs can solve this with an automatic digital input gain feature.
So I've got a question - how to determine how much input gain to take off the plugin after setting the input gain on the interface if the said interface doesn't show how much gain it's currently adding?
I had the exact same question. Idk if I'm right but I used a RMS meter using the mVmeter2 plugin from tbproaudio. Now it sounds awesome but I'm not entirely sure if what I did actually measures decibels. So yea, can someone help as well?
A lot of people have asked how I determined the amount of gain I added when turning the gain up on the interface, since the knob isn't labelled with any values.
I did this by using a loudness meter plugin, and playing open strings as hard as possible for 10-15 seconds. I then compared the loudness before and after I adjusted the gain, and the difference was 24dB. Hope that makes sense.
2:07
There's a much better example then using pixels. With pixels you still have the resolution issue that will always stay the same unless you add pixels. Instead of 100 pixels you use 200 pixels. But, you are always limited.
The better way to show arbitrary unit is o use vectors, that are also exist in photoshop. Vectors don't have size but they are mathematical unit. there for, you can make them larger then the universe and everything will still stay sharp and clear.
I would add that many people are confused and they think they have to follow the same procedure to set their gain.
That part was to make a fair comparison, then every card is different and everybody will have to find where it clips.
I would also add that sometimes you will have to pad, trim reduce the input gain via the internal software, if you have too much gain even when the knob is at zero.
How would a mixing engineer figure that out if the guitarist recorded their own DI? I guess the guitarist would have to provide that value.
Did you measure the loudness in Db or LUFS?
@@sqlb3rnmeasure the S/N ratio and see if it is inline with 24bit digital recordings or way off and closer to 16bit or worse. It could also be they are recording IN 16bit, but that would also need addressing, so it's good to check this on recorded clips regardless of if they are using an amp sim or not tbh.
Just make sure to ask them to leave in some silence at the start and/or end of the track, so you have something to compare against (It's also just good practice to avoid accidental truncation of the begining or end of the taget performance to be recorded).
What a great video. I’ve always advocated for zero on my input because anything above it I would hit red.
I’ve always been doing this on my Apogee and UAD interfaces.
Definitely going to try this out.
If you're hitting red going above zero, you're not going to want to......go above zero.
That usually happens because of ultra high volume face melter modern humbucker pickups.
Try with your new Strat and you will notice the difference. 😀.
I just watched your video PLASTIC METAL - Pod Express Black. Which sent me here. 🤘 🤘
I'm running an AXE IO interface and with a petrucci guitar I can maybe turn the gain up like to 10 O'Clock which is like +10db increase in signal... so my neural plugins are running like -12 input and its amazing how QUIET the plugins are with the noisegate OFF rofl. Only pain in the ass obviously is everytime I switch guitars I gotta redo the settings since the pickup output differs
Reply for the algorithm!
Fellow DSP enginerd here. This video is spot on. Subscribed.
"enginerd" I like it 🤣
Yep, Subbed.
Commenting for the algorithm. I work with developing DACs, ADCs, CODECs yadayada and I thought I was going bonkers when this 0dB chat started.
I worry that the engineering department at NeuralDSP feels the same way 😄
I'm glad I missed it. It's the same all over. Misinformation is established as truths because someone learnt a new word and don't know what it means, but still comment like a pro :). Whether you are recording audio or restoring a car or building stuff, it's always the same. :)
@@LifeOnHoth Or commenting on politics, apparently, looking at all the pro-Palestine rally attendees and the blatant propaganda and lies they swallow hook line and sinker, then propagate without a care in the world.
@@GhostNoteAudio I think it's mostly that this is the "simplest" solution that works OK for most people. E.g. my focusrite scarletts (I've got two generations) just clip immediately with any modern humbucker. The first generation could not be dialed down _enough_ , the 3rd generation one that I've got just about scrapes by at 0.
I think NDSP just assumed that proper engineers would actually know that if the interface has enough headroom, you use the headroom.
@@GhostNoteAudiothey should add a SNR detection feature in their software that tries to figure out if your analog gain is set high enough before ADC. Maybe some optimisation tool that says basically: don’t play, now strum hard, decrease your preamp gain etc.
I've been screaming "just lower the input gain in the plugin" for like a year.
I did it too on some on the mentioned videos and quickly felt discouraged !
Or raise it!! Me too but also, I've been screaming "use that input gain to make your "amp" higher or lower gain as you see fit! It's like an amp mod! You can have a hotter Soldano or a cleaner 5150, which is fucking amazing when you think about it! No need for pedals! Get a boost that sounds exactly like the amp already does! Use it to make music! It's amazing flexibility"
😂😂😂😂
Yeah. Also, digital volume pedal post preamp > turning down volume knob on guitar, but, I don't know if most people are ready for that yet.
I've been saying that to Cordy until he shadowbanned me from his comments.
Thank you for clarifiying this. I always thought the original assertion was a bit bonkers. Now we know why. Thanks again!
Yeah, pretty sure I saw someone suggesting it in your comments. Guess they never got raged enough to make Butthurt of the Week!
"these people on reddit are wrong" That just goes without saying.
hahaha, yes you can probably just link to any random reddit thread and there's a 99% chance that statement would hold true 😆
Reminds me of a waitress saying the size of the Large pizza was “12 slices” & my uncle joked “just cut it to 8 slices, we’re not that hungry”.
LMAO !!! Nice analogy right there bud,,, lol. Then wondered why the pieces were so large, lol. It's perfect actually, ;)
Jason McAteer for those who know!
best joke about pizza ever.
Oh, like when you get asked what's the difference between a pizza and a pie.
That actually makes more sense than you realize.
Most pizzas are 8 slices because they're not large enough to cut into 12.
A 12 inch pizza (for instance) cut into 12 slices wouldn't be slices at all. They would be more like slivers. To get 12 slices out of a pizza, the pie would need to be at least 18 inches, and upwards to 24 inches. In some rare cases larger, but it's safe to say it was a rather large pie.
When someone tells me a pizza is 12 slices, I immediately have a visual representation of its approximate size.
I'd debated making a similar video for months, and was hesitant for all the same reasons. From one plugin developer to another, thank you!
Another thing I'll add is that, there is no way to convert dBu to dBFS. It's like asking how to convert from inches to percent - one is an absolute measurement, the other is relative.
Love your plugins! 🙏
@@KazrogPluginsbut you can define what absolutely value is 100%. We just need (actually not, I don't care) one coefficient for audio interface
But why didn’t you make this video earlier? It kind of says to me you didn’t actually believe any of this until this dude made it. It’s wild to me for you to see apparent misinformation, but only deciding to speak up after someone else does an in depth video
@@withinthrall1445 sometimes peoples just lazy to do some free job you know
As a scientist and engineer, this has driven me nuts for decades.
The other one was Zero Latency
I got into a back and forth with some people on here about *ZERO* *LATENCY*
It doesn't exist. It's a compensated state achieved with programs and behind the scene (displays zero but really isn't) slight of hand. If something travels from point A to B....there will be latency as it took time from trigger to travel to sound from speakers.
I do digital audio for fun, and IT networking for a living. People always flub the difference between latency and throughput. Then they flub identifying bottlenecks as well. Say it again. There is no such thing as zero (absolute) latency. You can only match two systems (like in real time recording) by delaying the lower latency signal.
They're referring to the rate at which things travel through their heads 😆
@@fokeyjo I don't think so lol. I genuinely think they just believe it's correct, but it's not
@GhostNoteAudio, from a plugin OEM: thank you!! Finally a video that explain the only correct way to set up an audio interface, and most importantly, why that is the correct way.
We do not make copies of analog amps, so we do not have "accuracy" problems related to a wrong input level. We still have an handy input meter that turns blue if the level is too low, green if it's ok, red if it's too hot, but that's more of a suggestion, just a matter of taste. In the end, if the audio interface is set up correctly, then you can just trust your ears.
With our approach, if all our users set up their plugin in the "green area", then they can share presets among each other and have the exact same experience, without the need of spreadsheets, measures and math.
I have a confession to make regarding this video. I have about a dozen audio interfaces in my possession, ranging from 25 year old M-Audio PCI units, to cheap Behringers, mid-range Focusrites and top of the range MOTUs and RMEs... you name it, I probably have it (except Antelope, that stuff is crazy $$$ :). I chose to use the Behringer because it has objectively poor noise performance, and a high instrument input level - Every interface has the same issues that are discussed here, to some degree, but the Behringer highlights them. On my Scarlett 2i2 2nd gen, following this advice still results in a good SNR improvement, but it was probably closer to 10dB than 19dB. The main point is, however; you are optimising two different things, and that's what many people didn't realise, and was the cause of all the confusion.
Ok, and the Scarelett G2 is also legendary for a truly terrible Hi-Z pre, to the point it could be/frequently is called defective.
How about a decent modern interface? Not much difference left then is there?
@@mycosys I wouldn't say it's terrible, where do you get that from? The 1st gen was known for having very low headroom, and struggled to cope with active pickups, but they've done a pretty decent job since then.
So with an excellent interface that has a very good noise floor, going with 0 gain is still ok if you’re willing to get a bit of noise just to not go through all this?
@@devnull5109 With loud pickups, and an instrument level input that peaks around +10dBu, you're going to be pretty close to running at full level anyway at 0dB. It becomes more critical to adjust the gain when you have low output pickups, a cheap interface, or using super high gain ampsims. The main thing to understand is that optimizing the analog signal going into your interface, and optimizing the digital signal going into the plugin, are two separate things.
@@GhostNoteAudio thank you, this is very helpful
Fuck me. Been using the 0db gain approach for months because of those videos. This is brilliant. Tones sound wayyyy better now. Less noise, tone is less brittle, more detailed waveforms coming through. Thank you so much! Subbed.
Yes! The dynamic wave form was a very clear indicator for me that there was more signal coming from the instrument and less padding from the amp sim. Sounds much more like a real amp now.
What is the right way? I dont have time for the maths atm
Yes! The wave forms were so much better after switching to this method. More signal from the instrument and less padding from the amp sims.
@@MikaTarkela turn up the signal on your interface, turn down the input on your amp sim .. adjust to taste. Use your ears. Those goofballs at neural want your waveform to look nonexistent. 😂I knew from the beginning they were full of it... just common sense really.
There's only one thing better than being correct...being TECHNICALLY correct.
Exactimo!
Yawn
Nah it's attention 😊
The best kind of correct
If that were true, rock'n'roll would never have existed because no one would ever have overdriven their amp.
The art is to know when it's right to be technically correct.
My Audio Engineering heart is so full after watching this. Thanks for the detailed explanation and I will pass this along to anyone who needs it!!!
Thank you!! I have Hatred this advise (set input to 0db) for years. When re-amping, the signal to noise ratio issue is blaringly obvious and exasturbated when going through boost pedals before the amp. Thank you for keeping Gain Staging alive!
You actually printed the picture haha. Amazing video.
But he never proved that the printed image is 1 inch by 1 inch.
Haha!
@@drrodopszin Ahh, true,,, lol. But he did provide reference well enough to agree it is 1" x 1".
Also 100 dpi should have been at least 300 dpi in my opinion, some would choose 600 dpi, lol. But then he would have to have gone into resolution dpi, and basically explain SD to 8K lol... Ooo,,, and 4x3 instead of 16 x 9 lol...
Blimey this video took off didn't it!!
This probably is the actual reason why this video came out one year after all the others.
That pixels to inches opened my eyes to analog to digital. Thanks for the analogy!
I’m pretty sure his analogy is incorrect. You can convert pixels to inches. While things like dpi affect the actual size, there IS a measurable size. That’s why his image printed that way, which I found ironic.
What is the physical size of one pixel?
@@bobbydigitales 1/96th of an inch (approximately 0.26 millimeters
@@ParanormalResponse How are you calculating that?
@@bobbydigitales That is the standard refrence point in digital designing. The size can alter between your screen resolutions.
So you're telling me it's only 2024 and people have already forgotten how to avoid high noise floors lol incredible
Right? The lack of logic in "you don't have to worry about noise because pickups or something" is kinda disturbing.
@@w0mbatinawhat about all the digital pedals, amps and amp modelers. You don't set the input gain on those
Modellers have their input modeled to amps as well, interfaces are general purpose. But you still tweak input in modellers with input impedance, pads, etc
@@mserranomwell the point to set the gain to 0 is to have some sort of baseline for the guitar sim. But your audio interface must be set to INST/hi-z. Line level of course doesn't work.
@@cirisirpula152 Until you do haha. I had a guitar with Seymour Duncan blackouts that needed -6db on the input to not clip my Mooer GE300. And I have a vintage strat that needs about +2db to sound any good with the GE300.
I could never get a decent clean sound from my Marshall plugin like I could from my real physical Marshall amp, then I saw those videos telling everyone to set the interface to zero, I tried that and it cleaned up BUT it seemed silly to have such a weak physical signal so I put the level back up on the interface and introduced an eq plugin before the Marshall plugin, ran the eq flat but turned the gain way down and (I think) solved the problem in the same way you did. Thank you for confirming what I had suspected.
Almost all amp sims have an input level control. Which one are you using?
@ it’s the UAD Marshall Plexi Classic that comes with the Apollo Twin. I’ve looked for an input control but I can’t find one.
@@josemelrose5465 Ah yes, that's the old Softube plugin. Many older plugins didn't have an input level control. You may want to try the Amp Locker by Audio Assault, it comes with a free 1959 Plexi amp sim, called Prestige.
@@ErebosGR I don’t suppose you happen to know if the UAD Marshall Plexi Classic has an input level control? (It would be great if it did).
@@josemelrose5465 No, it doesn't. It's an old Softube plugin.
I recommend trying out Amp Locker by Audio Assault. It comes with a free 1959 Master Lead amp sim. Neural Amp Modeler is another good option.
Thanks for setting things straight! The misinformation thing can be frustrating on so many different levels. I remember when the term "stems" became super popular to throw around, but it became obvious that everyone had a completely different idea of what a stem is, and none of them ever got it right. So if the client was of a certain age I would have to clarify what they meant when they asked for "stems". I had people think it meant playlists of individual tracks, the whole multitrack, pro tools session files, the process of mixing, a type of plugin, or even a mastered/printed vinyl complete with all of the album art, distribution, tour dates, and a handful of groupies. So having to ask what they meant by stems would lead to the awkward "You seriously don't know what stems are dude? Aren't you an engineer? A stem is a hi-hat sample you moron...". Thankfully I haven't run into that issue much anymore these days, but for a while there I would throw up a little every time I heard some kid say stems...
Commenting for the algorithm. Great demonstration and explanation! Now let's hope this gains traction.
I see what you did there, @void_snw...
Finally! :) Thank you. Another advantage is that you record a reasonable DI signal which can be edited properly. At 0 input gain my Motu M2 gives me a nearly flat waveform.
Exactly. I can't believe so many people fell for this set at 0 advice
You can zoom the waveform verically in most daws
@@tutatis96 Seems like an added step. Why not have a good signal to begin with where each transient and note is clear?
@@EthanRom well you record clean signal. But you also use the zoom of your daw when you need it it's a keypress like every zoom it's not a "step". Also depending on the interface distortion can happen in the preamp before the clipping of the ADC.
@@tutatis96 if you record too low sometimes even zooming in won’t show it anymore
Thanks for helping me to realize that I had not completely lost my mind by insisting all those theories of lowering the volume at the input were nonsense
Pro audio engineer here. This video is exactly correct. I didn't even know this was a contentious issue.
It's a contentious issue because a lot of amateurs (like myself) are trying to use this fairly sophisticated pieces of equipment. And most of the equipment manufacturers provide terrible documentation. That means we all go to TH-cam to try to solve our problems. Which means...unless we find someone with this guy's background...we're probably getting bad advice.
I have a TONEX One pedal. You can literally find videos from 3 different "experts" telling you how to set the Trim on it and the approach for all 3 is different. And of course, the TONE One manual is worthless because it doesn't provide much more than 2 sentences on setting the Trim.
Same here...
I was not even aware this debate was going on until I clicked this video and now I'm mad because this is like Day 1 gain staging, how is everyone getting it so wrong? Great video btw, well explained
I didn't got the point of the debat too, except maybe everyone is putting too much input gain on their plugins ?
I mean, everyone knows that on the interface the input must be set to maximum before clipping, right ? ... Wait... RIGHT ?
I can explain that: its how internet knowledge spreads. Itll start by one person posting a comment or a video or something in which they oftentimes literally disclaim 'yo guys, i have know idea what im talking about, but ima talk about it anyways'. Now people coincidentally not being affected by any outside influence the initial poster forgot to consider, are 'successfully' applying the 'knowledge' not noticing that the bunch of a few extra problems they got was actually caused by that 'solution'. So they go ahead with the opener 'yo guys i have no idea what im talking about' still in their minds and post a video or something themselves, further spreading the false knowledge. Eventually people not even involved with the domain of the problem will start commenting on it, further entangeling truth, believes, self promotion and stupidity into one big mess.
This is how you end up with such discussions in a nutshell.
Yeah. I didn't know this was some sort of argument. Also when switching guitars , I have to fool with input gain for each one. When I change modeled amps and effects same too. mostly on the sim instead of the interface though. I don't know the science behind it, I just ...do it.
@@ohcibi “The irony of the Information Age is that it has given new respectability to uninformed opinion.”
― John Lawton
Man, I wish I could give you 100 upvotes.
Finally someone that knows what he's talking about. Congratulations.
Thank you so much! The whole 0dB input always sounded incredibly dumb however the other alternatives presented didn't yield better results. This is the definitive way to do it
either correcting the gain structure digitally or going 32 bit is the acceptable solutions rn, but 32 bit interfaces aren't as common yet.
@@alrecks619 24 is already more than enough. Hell 18 bits would already be more than enough (17 bits is twice as many values as 16, etc) there are tons of interfaces and field recorders with 24 bit recording and no gain knob at all, because they simply don’t need them. The idea that you need as hot a signal as possible is one that originated with tape, held traction with 16bit recording and noisy preamps, but makes little to no sense with most modern equipment.
This guy demonstrated a well known and tangential audio concept with his potato of an interface from Behringer. His preamps are garbage and his adc is likely working at 16bits. It is important to know how to minimize your noise floor for your particular gear, but the conclusions he’s come to based on his simple experiment are…. Wild
@@threepe0 totally. He is theoretically correct, but the kind of optimization he is advocating for as something essential seems to have already been taken care of at the hardware level in most decent modern interfaces, to the point where the difference (even though still exists) is negligible for most use cases.
@@alrecks619
Even 32 bits audio cards have a noise floor.
Now, is it as obvious as a 70s tascam mixer?
No, but it's there, so it's good practice to adjust the gain as high as possible *before* the conversion.
@@threepe0 for a modern metal guitar tone the signal chain might include compression, distortion pedals, overdrives, hi-gain amp sim, tape distortion, transformer emulation, and a boatload of other non linear effects that can easily reduce the SnR by 10s of dBs, so doesn't it makes sense as a general principle to optimise the SnR at the input even if the self-noise of the ADC is not that high? Not to mention that when using a noise gate it results in a better sounding sustain for long-sustaining single notes, where one would like the decay to sound as natural as possible.
This is how I've always approached setting levels for amp sims. Didn't have a handy-dandy chart, but by ear has worked well enough.
It's the same as how I'd record anything else. If I'm recording vocals, I don't turn down the input gain on my interface to match the level the compressor plugin and EQ plugin I'm going to use wants. No, I record a strong signal on my vocal and then adjust the input gain on the plugin or the clip gain as needed so my compressor is happy. Digital recording basics. (Yes current technology is more forgiving that of years before, but still.)
Good video and nice technical explanation. Looks like you're a fellow lefty as well 👍
Appreciate this information. I would propose amp sim devs add the basic process and information to their user manuals as well as include either auto-gain function or a “green light” for the setting the input gain knob in their plugin.
This reduced the noise on my signal drastically, i almost don’t even need a sound gate anymore. Great stuff!
I am more shocked that you had a working printer lol. Very nice video !!
A wild Lucas appears.
A lot of the issue stems from ToneX users, where the physical pedal version (opposed to the plugin) has a global input level that is both the AD level and the input level feeding the captures.
So if you optimize the input for “more bits” you can’t then lower the optimized input signal before feeding the next part of the chain.
So a lot of the blame should go to IK Multimedia not to the users scrambling to find workarounds.
In a perfect world you are absolutely correct and for DAW/plugin use one can achieve this even with a Behringer as you showed.
But lots of players don’t realize their gear is sabotaging them inside the box.
I wouldn't consider the global input level to be "both". The AD conversion happens before and the pedal itself has an instrument input (like an amp).
That's why you shouldn't adjust the global input level when switching guitars. Then what would be the point in having different pickups.
Speaking of tonex the important thing is to set up the pedal so it corresponds with your experience in the app where you download and tweak your profiles.
Ah nice ... I came across this ...
th-cam.com/video/BqbviHm9hrY/w-d-xo.htmlsi=JZcNfDHvvFYgiOMd&t=14m51s
I saw that a few weeks ago. I didn’t elaborate in my response, but the ToneX’s instructions tell people to “turn it up before clipping then turn it down a bit” but there is no scale, no reference.
I’ve been setting my presets for plugin and physical pedal using a signal generator. Most players won’t be doing that.
And then how was the level set when a capture was done? Was it correct?
Lots of tails to chase.
THANK YOU THANK YOU THANK YOU THANK YOU
I was always super skeptical and my first thought when I saw these videos was "yeah, but what about the signal-to-noise ratio?!"
Having the signal so close to the ground noise is such a weird suggestion!
But I said welp, these people seem to have done crazy research, but on top of that a lot of plugin developers confirmed this. So I guess that must be true?
Turns out i've been having the noisiest recordings ever since then😭
Especially with some of my very low input guitars.
tho I swear if another video comes out proving you wrong i'm gonna be so done with the internet hahahahhahaha 😭
also nice theme ❤
This x100.
I wasn’t looking for this, I didn’t need to know this, I own / use none of the equipment or tools you talk about, it just appeared in my recommended feed and .. even I understood this perfectly. You may have a like sir
The quantity of misinformation I see thrown around the internet, especially when speaking about gainstaging and impedance is overwhelming. It's depressing to see how technicians (or "sound engineers") with a lot of "experience" are much more prone to believe in audio superstitions than to use their own heads. As an electronic engineer, I can't understand why people mess up so much, as these subjects are based on basic and simple concepts and are the basis to be able to work with any analog audio technology. This video is well made, straight to the point and simple, I hope it will be understood by most people
It's literally just physics and biology, people love making it so much more than it is.
Just a side note, optimizing the input before clipping might introduce some natural compression from the pre AD circuit. So, I still suggest to try it out and use your ears for best result. Also a buffer guitar pedal (like a BOSS tuner pedal or other good non true bypass pedal ) might be a big benefit to noise and impedance matching! Cheers and thank you very much for your "guts" to stand agains the (stupid) internet! I would have imagined my self doing this 20 years ago, but now i'm just too old :D
Just tried this out! What a difference it's actually made to the noise! Thanks for this.
I do wish amp sim makers would just implement the level indicator like ML Sound Lab where you just chug turning their input dial until it goes green. It'd save all this headache! haha
Never noticed that.
I did the same and was amazed at how much of the higher end fizzy sound was gone and the tone was so much tighter and defined. Amazing so many respected TH-cam musicians hadn't figured this out. There really needs to be a better standard way these plugins all measure input and adjust for us.
@@BeGoodBe On the ML Sound Lab amps? Yeah, all of them have it.
Turn the input gain dial up until a little light comes on, you'll see it. That's the sweet spot for the amp.
@markuskerloch Cheers, I'll have to use that lol.
Helix Native and THU also have input level indicators. PRETTY HANDY!!
I'm very glad you used that chart. It's pretty accurate. I've found my MOTU interface's DI is my favorite since it can boost and cut. Some of my really hot pickups require a cut, where lower output ones need a boost. Used in conjunction with the chart it's REALLY a game changer for amp sims.
Thanks for clarifying. Been doing it "your way" for ages, & the recent discussion all over the interwebs had nearly convinced me that I should instead be recording about 12db quieter than I have been, rather than just normalizing the file post performance to my desired peak after tracking.
The crazy thing is that I've known better for ages, but when enough voices say the same wrong thing, it can give you pause. The noise floor issues associated with low input signal seem obvious, but when a bunch of "smart guys" say noise isn't an issue with low input gain... Sheesh. Nice video!
I'm not a qualified audio engineer - just a producer. However, i used to think that it's obvious: the more clean boost you got = the more depth (resolution) and more signal/noise ratio you achieve within the interface. It's strange that someone claims the opposite as a community consensus, because all the A/D theory proves that. Thank you for that video. Musicians possess a humanities mindset rather than a technical one, so there's a lot of honest misconceptions like this - it will be very helpful.
"Musicians possess a humanities mindset rather than a technical one" Thanks for speaking for an entire section of a population like you know everyone.
@@andymerrett Since I am not an English native speaker, I didn't expressed myself precisely - I meant that musicians are humanities more often than technicians. I have 10 more years of experience of working with artists, so i have some observations. Excuse me or you're welcome, i dunno
@@andymerretthe is of course generally correct, despite the offence you have taken
I don't know that musicians really skew any differently than the general population when it comes to technical mindset. It's just that _most_ people aren't technically minded, and those that are tend toward fields where this is an asset. But you still get plenty of technically minded people (like Tom Scholz) who also happen to be great musicians.
@@mal2ksc I agree with your point in general, but i don't think that there is a discrete dualism between these extremes. Personally, i have a technical and humanity mindset at the same time - this is common among producers, mixing and mastering engineers. There's no way you can mix your music without any abilities in math and physics, for example.
However, to write good music, you need a strong humanitarian background. Music written by engineers is often boring and doesn't excite in any way, because most of good music is about feeling rather that knowledge. In my opinion, only a combination of cultural and technical development factors gives a result.
The music industry got it long ago, which is why artists work in tandem with engineers - without them, everything will sound bad. Without the former, it will sound good, but it will not catch. However, it's not a rule - just a trend that becomes mainstream.
Thank you for this! Though I've maintained that we should digitize at the hottest level w/out clipping to capture at the highest resolution, I've gained correct understanding for the justification behind it.
As an engineer who deals in communication systems, and a musician, this made me feel good. Data driven decisions based on solid engineering principles. Great job.
Thank you very much for making this. I was confused about the whole thing this year. I especially hated having such quiet DI recordings in my DAW to work with
Thank you for clarifying this. With some luck, the hair I've pulled out of my head this year due to this misunderstanding will grow back.
Ahah, you were not alone in that situation !
Thanks for making this video, hopefully it can put an end to the madness!
I didn't even realize this was a discussion. I've been doing it correctly since the beginning.
Thank you for making me feel smart and very cool.
This is the recording equivalent of flat earth theory. A bunch of people start believing something stupid for no reason and now we have to re-prove what has already been a proven and established fact for decades.
Earth is not a globe.
@@Rodriastral666 it's an oblate spheroid.
@@Rodriastral666 while the definition of globe is the earth that surely makes sense! Lol
Globe can mean a round object so the globes in a classroom is a spherical representation of the earth and not exactly the pear like shape of the literal shape of the earth. Nobody said the globe maps are exactly.
For all these years I never left a zero gain on my input, but I always corrected the gain on the plugins by ear. Now I've learned how to correctly do the adjustments with precise levels. Will try it out very soon, thanks for the tip!
I’ve been stuck on this problem for a while. I feel like a lot of us beginners have trouble understanding gain staging which end in us getting stuck somewhere along the way in the mix.Thank you so so mich for this video! Subscribed
Really good video, subscribed! Well laid out.
Basically, the "set it to 0" approach is just "the easy way out" for the ones not knowing what they are doing. I have to admit that I belonged to that crowd for quite a while, or differently said, I did not care for it. The issue with John Cordy's, and even way more so with Rabeas video, is that they simply are parrotting out what others told them, to make a click-baity video. They do not explain the technicalities behind those approaches, but they just say "Yeah m8 you dumb turn knobz down wow such sound lol"
On the other hand, Ed Sokolowski's approach to propagate for setting the gain to 0 (which I would say had the most impact, since John Cordy for example also has grabbed this idea from Ed) has a different motivation: To have a single common denominator and a clear reference from which to work on moving forward. For people to being able to properly assess at which levels they are making their profiles.
If you gain up your channel via analog means, you have to evaluate through measurement how much of an increase you are pushing for, before you reach the clipping threshold. A guitar signal due to it's inconsistency is not a good reference to check on that, and not everybody has a sine wave generator they can plug into their interface (to get nice before/after values of their analog gain staging). Also, the Behringer UPHORIA 1820 with it's +18dBu headroom has plenty of room to utilize before you hit the clipping threshold - with the +10dBu of my audient iD44, that might be a different story.
However, this is a great approach which I will put to the test on my interface. I mean, I personally did not have too much complaints on noise for my taste by setting all to 0, but I do CLEARLY see, as an fellow electrician (although not within the field of digital signal processing, I shifted sideways nowadays to networking technologies and Crypto-IT), that everything you stated in this video is technically correct and sensefully the right thing.
THANKS!
Well said! Isolating independent variables rules!
Thanks a lot for this clear explanation. It reminds me of another analogy with the digital photography. There is a lot of difference in the final image file, if I underxpose a shot and then I add luminosity in post production, than if I overexpose the shot to the maximum possibile level but without burning lights, and then I reduce the luminosity in post production. It is always related to the signal to noise ratio and to the ability to exploit, at the best possibile level, all the available bits.
As a person who mainly just plays guitar and doesn't record much, the moment I heard about setting input gain down in general, made me slap the knob to 0 on my dac and bam. suddenly all the presets I avoided using straight in my plugins sounded like how they were meant to sound, and so I just left it like that for ages. Over time I got bugged by some of the noise and started to compensate as you showed in this video, but I wonder if some people just found it easy because it is one single adjustment, vs adjusting the input gain for each plugin they use...
As a recording engineer (who also plays guitar) I just instinctively gain-staged the input of amp sims like I would anything else. Glad to find out I was doing things correctly. Great vid 👍
Dude, you really deserve if not a gold then a bronze monument for sure.. no irony.. I listened to advice for years, shoveled through a bunch of printed material to now find out and hear, see such an obvious thing. Thank the gods and engineers!
Good presentation on this complex topic. And coincidentally, right after seeing this, I watched a Crow Hill video where they called out your Ghost Note Audio Conductor MIDI controller (nice!). I think some addendums are worthwhile re: standards, and whether the optimization you prescribe is the "perfect" being the enemy of the good (or even of the better than "perfect").
First, re: "There is no accepted convention for what is considered peak level for an instrument input." Well, if by "accepted" you meant "universally accepted," I suppose that's right; anyone can ignore a standard if they wish. But in a video attempting to be "rigorous" about this topic, it's probably worth noting that there are in fact standards; it's just that (1) there are competing standards that differ, and (2) there is no enforcement mechanism, so interface manufacturers are free to ignore standards if they choose.
In Europe, EBU R68 specifies that 0 dBu (an input power level) should correspond to -18 dBFS (decibels with respect to full scale, a digital level, "unitless" in your terminology, though "relative" would be at least as good a term). In the US, SMPTE RP155 specifies that +4 dBu should correspond to -20 dBFS. So there's a 6 dB difference in the recommended calibration just from "crossing the pond." Wikipedia's "dBFS" entry indicates that Japan, France, and "some other countries" recommend +22 dBu at 0 dBFS (the relevant standards body is not identified). Moreover, +4 dBu is considered to be the "professional" reference level. Some interfaces adopt the lower consumer reference level of -10 dBV instead (and the difference between the final "u" and "V" is important; these reference levels are not 14 dB apart!). So, there are some standards, but their multiplicity means that "There is no accepted convention..." is not a bad description of the situation in practice.
The datasheet examples you give are sadly imprecise and thus bad examples (this is not your fault, obviously). A level is called a "maximum" level without saying what that means. Is it describing analog performance of the preamps, specifying when some point in the audio signal path (e.g., just before the ADC) hits some target distortion level (say, 0.01% THD)? Is it instead where the ADC produces the maximum digital signal? A precise datasheet would say exactly what is meant. For example, for my RME Fireface UC (which lets you switch between +4 dBu and -10 dBV input reference levels), they provide a table specifying dBu above the reference level that corresponds to 0 dBFS. For the +4 dBu choice, 0 dBFS is reached for signals at 13 dBu, "the latest EBU recommendation" (which evidently was 1 dB different from the current recommendation). If "dBFS" does not appear in the datasheet, you can't be sure how the analog and digital levels are related.
"The accepted wisdom was-and still should be-to set the input gain... as high as possible without clipping the signal." Your discussion of gain staging seems to me to date back to the days of 16 bit bit depth, with only 96 dB of dynamic range. Part of the motivation for moving to 24 bit and higher bit depths was to make it unnecessary to push levels very close to 0 dBFS when gain staging. With the headroom on hand in an ADC with decent, modern preamps and 24 bits or more of bit depth (144 dB!), it is not necessary to set the input gain as high as possible to avoid clipping in order to get a very clean signal. That was always a nuisance, because in the studio (and even more so on stage) the talent would very often perform louder during an actual performance than during a sound/level check. With 24 bit converters, you can set your gain to be well below clipping and still have an imperceptible noise floor-and plenty of headroom in case the performance is hotter than expected (and to enable EQ and other processing without having to attenuate the signal digitally later). You call 90 dB of range "relatively good." Actually, if that's really the signal to noise ratio one is achieving, that's spectacularly good. Unless your listening room is an anechoic chamber, you won't be able to use that dynamic range-it's the difference between the lowest perceptible sound (for perfectly healthy ears in a completely silent environment) and a sound level that actually causes hearing damage.
So: No need to push close to clipping; in fact, your recording and mixing life will probably be easier if you give yourself 6 or 12 dB of headroom when you set the analog input gain. But do follow the rest of the advice here, so that whatever analog gain you use ends up hitting the amp sim plug-in at the designed-for level. Or, on the other hand, just do what sounds good to you. There's no law (or even a standard!) saying you need to use the amp sim the way the manufacturer intended! -Tom
Good points. A few thoughts-
Not all equipment uses the same full scale voltage reference, so while it may be useful to have a defined level that is "compliant to" some standard. But, practically, I think it would be better for manufacturers to clearly state what the full scale level in dBu or volts (or both) is for a given interface and we all can go from there. Crest factors (peak to average ratios) of the signals also affect where we might set the "average" level, and of course the averaging method affects that back off, as well. More clarity rather than less information would be helpful all around, I would hope.
By the way, 90dB dynamic range, about 15 bits (ENOB), in a single source channel is probably not so great when combined with other signals and additional processing. It's about what the old redbook CD standard can do, though. Cheers
You're conflating thermal (electric) noise and quantization noise. 24bits does reduce quantization noise, but you still have the best SNR when setting your gain to just below clipping. That's basic gain staging. Of course you always keep a little headroom when setting your gain. Setting a proper gain is not very complicated either..
@@nolyspe No, the point is that most interfaces that support higher bit depth than 16 bits also have quieter preamps and converters. It would be somewhat pointless to put a preamp with only 90 dB of headroom (say) in front of a 24 bit converter. If you compare the preamp specs of modern interfaces with older interfaces that were only 16 bit, you'll see modern interfaces typically have quieter preamps.
Yeah I found the values for the amp sims in the spreadsheet. Great video and of course everything here is correct and I agree with completely.
All I would really add here, is that most interfaces with around 12dBu of headroom at 0 gain are going to produce a signal close to clipping when using humbuckers. So in these cases, it’s already going to be very close to an optimal SNR, and you can easily know your headroom amount (in order to calibrate your specific plugin).
The other factor, is that as long as the noise of the interface is sufficiently below the background noise of the pickups, there will be little to no improvement in optimising SNR - increasing preamp gain will also increase EMI/RF interference from the pickups.
On noiser interfaces like Behringer, it’s absolutely worth optimising SNR. I think many interfaces in 2024 will have less issues with noise and too much headroom. If noise is not causing issues, for the average home guitarist, easily knowing the level of headroom they are using will be more useful than a louder signal.
In an ideal world, the SNR would be optimised, AND each user would be able to easily determine their dBu->dBFS ratio. If SNR is essentially “solved”, then the easiest way to know the dBu->dBFS relationship is to turn the gain to the value provided in the specs.
Agree with all of that! I intentionally used the Behringer because it highlights the issue. It both has poor noise performance, and an unusually high input level. These two facts conspire to make it particularly sensitive to this subject matter 🙂
And thanks again for collecting all that data!
@@GhostNoteAudio For sure! I think generally companies like UA/Audient/Focusrite/RME etc are doing the right thing with “standardising” 12-13dBu as a sufficient level of headroom, while still retaining good SNR. If the maximum is below that, then it becomes easy to clip. I believe some UA and RME interfaces have stepped gain controls which makes it easier to optimise further without the need to measure the headroom manually.
Also guitar modellers like the Axe FX and Helix have dedicated DI inputs with extremely good SNR converters and can be a good option for guitarists who might already own one and have a noisy interface.
I noticed that with a Gen 3 Scarlett I just started using! On my Tascam model 12 I would run the preamp at around 11 o'clock where it has 0db marked and my signal would peak inside reaper at around -6. So I was quite surprised that on the Scarlett, turning the preamp all the way down resulted in a similar input signal level
Great Video! Very well explained and the detail was fantastic!
i was about to comment that "finally someone understands gain-staging" and then that whole section on gain staging began. Love this so much. Anytime someone mentions gain-staging incorrectly, or says "gain staging isn't even important now with super high bit rates..." I'm sending them this video
That was a cool tutorial and explanation... I may need to watch it a few times to fully take it all in but the basics of gain staging makes total sense to me...
for the lazy:
1. hit the strings as hard as you would playing and then turn the interface preamp gain up til it clips and back it off a little bit
2. open up a high gain patch on your plugin and play chugs whilst turning down the input gain on the plugin until it sounds good bc that's literally the point of music.
inb4: "it won't be the same" yes I know, but also I probably don't have the same pickup as the preset maker so I'll actually get closer to the intended tone this way.
The point is "sounds good" works up to a point. But people don't want presets to sound the same, they want the simulated amp to react as it would IRL (as close as possible) - and that means if the real amp has a gain knob and you put it on 5 you expect the same gain as if you plugged the same guitar into a real amp. You have to note too that some of the software (e.g amplitude 5) have models where it's not got enough gain (compared with the real amp) if you put the input gain on max.
It's a mess - and the input gain on zero + spreadsheet was an attempt to solve the mess for some plugins and interfaces (and really are people getting that much noise from good interfaces? Probably not)
@@michael1 I actually really don't care if a simulation is equal to something analog, it must sound good to my ears, nothing else.
I would absolutely care if something marketed as "this thing" behaved like that thing or not. If it sounds good but doesn't sound like the thing you're calling it, maybe call it something else @@alessandroproverbio3411
@@michael1 I hear ya, I just don't care about how "accurately" the plug reacts to a real life amp. If it feels how I want and sounds how I want everything is doing it's job.
Never use a distorted plugin patch for this, you must use a clean patch! Why? select a clean patch and crank up the plugin(!!) input gain. Hear it getting distorted?. Now back it off till it becomes clean, now back it off a little to give you headroom. You will NOT hear this with a distorted patch, thus you will NOT know where to back the gain off to...and if you don't know that, you will NEVER guarantee the correct sound of that patch. I'm very surprised this isn't recognised in the video.
When change to a higher/lower power pickup, you'll need to optimise both the audio interface and the plugin gain settings again.
Thank you. Finally someone gets it. Depending on the situation, pushing your converters actually sounds good as well.
I like how Rabea got on there gaslighting everybody like he had been doing the whole interface gain at zero bullshit. Gtfoh.
He turned that video private ;-)
@scottyecora yeah that's not surprising. Lol.
Man, that's one of the biggest, most useful videos ever watched on TH-cam! I'm practicing this thing right now on my plugi setup and my guitar tone has never been so much focused until now! Thank you so much. You've got a new subscriber, content like this is real gold. Beers & pizzas from Italy.
THANK YOU. Also have an electrical engineering background, have worked in DSP, been doing this stuff for almost 20 years. The misinformation drives me crazy!
I have the same background. Those zero gain videos made me really mad!
I have office tomorrow. Its past 23:00 here in my time. Just as I was getting ready to go to sleep I saw the thumbnail. Clicked on it. Was not disappointed. Halfway through your video I was like "holuy shit I need to try this":
Told the wife not to disturb me for 35-45 mins and that I will work from home tomorrow.
Took my like 5 minutes to do the steps you mentioned in the end.
Holy shit dude my life has changed.
If bunch of musicians can't figure out gain, imagine how well an average audiophile understands AD/DA conversion...
I love AC/DC!
These idiots can't tell the difference between their precious $3000 speaker cables and coat hangers - see the Ethan Winer demo.
Asking a musician to explain gain staging is like asking a surgeon to explain the krebs cycle.
@@mezu-e That is painfully true
I used to work for Audio Note, very arguably the top top end of hifi, and we’d spend a lot of time deprogramming people.
Most so called audiophiles aren’t musicians, don’t see live bands, jazz ensembles or orchestras. They talk about statistics, efficiency, freq. res. and dynamics but never soundstage, ambience, acoustic energy, character, tone, catching phrases musically etc etc The average audiophile treats hifi like a digital camera and we all know that having a camera is just one tiny step on the long road to being a great photographer
Great video! I definitely thought I was doing well in following the 0dB on the interface and adjusting in the plugin. One thing is still unclear to me though: how do you know by how much I increased the input gain on my interface? Since it's just a physical knob I can't read it. Do I use my DAW to check? I would need a reliable source of consistent level (so probably not me strumming my guitar as hard as I can) to be accurate in the amount of input gain I added. Thanks!
Being a musician and an electrical engineer, I am happy to watch such a vidéo.
The «zero gain» vids made me sad when they appeared and I felt alone explaining exactly that in their comments.
I've always recorded guitars as loud as possible without clipping, and then adjusting the ampsim input accordingly "by feel".
It just seemed logical to me to use the interface's dynamic range as good as possible, and adjust the gain afterwards before plugins, especially with ampsims.
So I'm glad it seems I did it right without ever thinking too much about it :D
Thanks for the great video, hope it helps a lot of people!
I'm genuinely surprised that this is even up for debate, it's like an argument about recording digitally from 20 years ago. I guess youtubers spreading terrible information has got people convinced.
You're the hero we didn't know we needed. I am so sick of the stupidass "set it to zero" argument.
I had never heard of the "set it to zero" bad advice until watching this video. I totally agree with @GhostNoteAudio and his maths and science. That not gaining your input ever advice is just wrong. The zero to worry about is 0 dbFS off your DAC, just stay below it, but let your loudest peak approach as close as you dare to.
"Set it to zero" ?? REALLY? Why have a gain knob then? Leaving it down puts your signal closer to the self-noise of the interface circuitry, although turning it up will bring the noise of the source up with it, but it will still be at the same ratio to instrument signal as you are playing.
Now all watchers of this video who didn't know before can enjoy cleaner recordings of their guitars. Yay!
I never knew this was a thing, I’ve learnt something today. All my videos have had it set on zero, I’ll definitely be giving this a try.
I’m baffled that people thought zero gain on the interface was the solution. Am I getting old or something? That’s basically recording guitar DI 101. Thank you for putting this together and bringing some sense into the world 🤘🏻
He used a shit interface. You have to understand the problem they were trying to solve first. Component based amp modelling software has dials like the real amp. People want the simulated amp to work the same as if they plugged their guitar into the real amp. That means, e.g if the real amp has a gain knob from 0-10 that they want gain on 5 to sound the same as the real amp with gain on 5 - for their guitar. Obviously this also tends to make presets marked 'clean' sound clean (modulo that different guitars / pickups etc will affect presets)
But the problem is now twofold (a) interfaces were saying to set gain 'just below clipping' and on that setting many of the plugins were massively over gained - because that's not how the plugin authors envisaged the gain to be set and (b) Many of the plugins, especially the ones that come with tons of different amps, are a complete mess in the sense that if you pick n of their models you need to use n different input gain settings to make it match the real amp - and unless you own all the amps how do you begin? Plus some of them the gain is only correct if you turn everything up to max, i.e the clip lights can be on but the gain is still too low - this amp modelling stuff can be an inconsistent mess.
So problem (a) was solved by at least making people aware that their input gain matters and giving them settings that work for a bunch of plugins and interfaces. But saying "input gain on zero - set input gain on the plugin from the spreadsheet" has confused some into believing 'input gain on zero' is somehow key to everything or a new religion. Obviously it's not. The big clue is - why would you have an input gain knob if it weren't supposed to be turned? But it works - and it works far better than this video suggests because he used crap hardware to exaggerate a problem. Quite disingenuous really.
Problem (b) is not solved and there's no real solution. You really want all people who are simulating a bunch of amps to do so in a way that you only have to set input gain once - where you have things that are more open and 3rd parties can add, well, you can see why it often ends up a mess.
But the noise in this video was a bit silly. It's like I'm using a pod xt for my interface - it has no gain knob to turn but the biggest problem with noise I have is if I point the pickups towards the computer running the plugin. Plus of course this video side skirts the issue to get off on calling people wrong. He doesn't offer settings that we can use with our plugins - but suggests we refer to the same spreadsheet these "wrong" people created. So it's like he's saying "They're idiots and are wrong" - but, wait a second, their spreadsheet is the information we need to make our plugin simulated amp gain match (as closely as possible) the real amps that are simulated and the gain on zero really doesn't make it that noisy - so they were actually right. The only people who are wrong are those who think you absolutely HAVE to have your input gain on zero - of course that's not true. Of course you can set input gain on your interface how you want and then adjust in the plugin accordingly - but that misses the point of the whole exercise which is trying to get simulated amps to behave like the real amps.
People will believe anything as long as it allows them to half-ass something.
@@michael1 Well, the problem is actually threefold: (c) 0% preamp gain would result in a different level from manufacturer to manufacturer (sometimes even different between various model lines of the same manufacturer), so it still can't be a baseline.
@@ramziel That's what the spreadsheet is for.
I feel as though gain staging was understood more in the analog days. The digital age has not gotten the lesson. I hear incorrect gain advice from young sound engineers quite a lot. And when i look through their mixer scenes, oh boy! Gains all over the place.
Thanks for sharing this great lesson!
I love it. Thank you for applying a clear method here. Makes total sense.
Youre right, but im also shocked by how bad the pre/converter noise on the Behringer is to make it matter that much. Definitely will only be using mine for line signals.
Yeah they are really bad! 😀 But unfortunately, the UMC1820 is quite a popular interface.
@@GhostNoteAudio Just grabbed one for $200US new, but it will definitely only be doing the synth connecting duties as an ADAT expander i grabbed it for XD - 20yo MOTU died, i doubt it will be worse lol.
Theyre really popular in modular groups as expanders cos we need so many channels, but HOT HOT HOT signals only XD
Amazing that so many peoples takeaway is that "none of this matters"... when right at the end of the video you literally set the interface gain to 0, find the accurate plugin correction and THEN offset the interface gain and plugin gain if there's more headroom to be had while not clipping the signal.
The comments are a fascinating insight to people wanting to believe that plugin accuracy doesnt exist and they "finally found someone who agrees with them" so they're piling on. Had they actually watched your video they'd see you're agreeing with everything JnC, Rabea, Ed and myself have been talking about but just taking it one step further with SnR optimisation (which should already be obvious to people but is a great thing to point out).
if it sounds good then it's good
Absolutely spot on and very well explained. I am also amazed at how much rubbish is pedaled on TH-cam from TH-camrs trying to appear professional but are spreading ignorance. Long before there was "digital", we'd max-out analog desk inputs just before the signal on the VU meter hits the red or distorts. That wisdom is also true for digital.
I had no idea this whole conversation was going on - and I'm glad I missed it. I learned gain staging (correctly, as you prescribed) as a fundamental (which it certainly is). It boggles the mind that people would somehow start getting it all wrong, when it's such a fundamental part of audio production. Thank you for this video.
7:11 Judging by your footage you activated BOTH instrument and PAD knobs for your input. The latter attenuates the signal.
So the advice about 0 gain might sill hold on simpler interfaces / non padded inputs
the INSTRUMENT / Z inputs are already delivering max (or rather optimal) gain for PASSIVE pickups without clipping and changing gain knob even one degree results in clipping (at least with interfaces and guitars that I had focusrite and motu).
I noticed this also. 18 dBu normal and then the PAD adds at least 10 more dBu of headroom.
Maybe he will make part 2 with modern style interface and humbuckers with 12 dBu of headroom.
So, I tested your method instead of the "0 Gain method" I was previously using, and yes, it absolutely brings down the inherent noise floor of the interface. One has to be however mindful of what signal or noise you might be boosting when going with the optimized SNR setup you recommend.
In my case, I get some crippling interference through my computer, which I dislike more than the inherent hiss, which nicely masks some of that nasty noise.
But it definitely is awesome to know your tools and just learn another point of view every now and then! Would my PC be further away from me, then this would be a very valid approach to get the best out of all!
I am SUPER HAPPY to see that this video is getting quite some traction and your subscriber numbers are skyrocketing. Off to 1k, buddy!
Agree. Better to have a higher homogenous white noise than an angry "eeeeeee!" getting picked up into the signal. Disabling the CPU turbo boost helps with that high pitched sound.
@@iursnitram Yep, that's what I do from time to time. Using Ryzen Master with a preset with lower consistent CPU speeds works super great!
You don't set it to max level. Usually for microphones and stuff you set it to around -12db or -9dbfs. But for most interfaces a electric guitar can already be -6db to higher than 0db(clipping) with gain at zero.
@@johnyang799 with no gain it peaks around -9dBFS, so there is something to be gained.
Also, when I set the speed of the CPU cores statically, then the process explained here works very well, because I get great SNR, but also very less crippling of the CPU noise.
@@PippPriss Did some measuring and testing with a 1kHz sine tone against the pickups to level match and it seems like adding gain on the interface doesn't really add more of that high pitch whine or other noises into the signal. In comparison, by leaving the input at zero you get more noise that ends up masking more of the whine but it's level is the same. Ended up adding 5dB on the interface input, it's not that much but the SNR gains are clear. I guess I am back to the old method of targeting around -6dBFS for the extreme peaks.
as an audio tech person, i love this. Thank you.
Youa are so very correct. Matching what the device needs is always the best solution.
This makes so much sense now! I think the trend of sticking high gain pickups in whatever guitar and turning gain knobs up has influenced the 0 gain input dial misconception. Some guitars won't require much of the steps you explained in the video. But a lot will definitely need this and benefit from it. Thank you for the thorough explanation!
In my modest opinion, the story is a little more complicated with sim plugins, which are supposed to behave exactly like the real counterparts. With this approach (ie, maximising the signal/noise ratio for ANY type of input), the amplitude of a signal generated by a Strat as observed by the plugin in the digital domain will be roughly the same as the amplitude of a signal generated by a Les Paul. So I think there are two options here: either we follow this advice, but then we'll have to use different level adjustments in the digital domain to compensate for the different instruments, or we give up a little bit of noise floor and adjust the interface gain for the highest output pickup we have, and keep the level adjustment fixed in the digital domain. In any case, however, I think the sim plugin manufacturers should report a reference level that the plugin expects to see *for a particular pickup*, otherwise we are still missing one aspect of the equation! (and then obviously you have TONEX/NAM, where the capture process is not standardised, so those numbers should be provided by whoever produced the captures).
I set it this way to my loudest humbucking pickup. And then just leave it at that even if I plug in a single coil. So the lower signal of the single coil behaves how it should with its lowered output. I think that makes sense?
It does make sense. To be 100% accurate, we'd still need to know the reference input level that the plugin expects *for a particular pickup*, but the manufactures don't go down to this level of detail, so using the reference levels that were collated in that spreadsheet should put everyone in the right ballpark... and we can finally play 😀
Really interesting point about different types of pickups! Plugin companies providing reference levels for different guitars/pickups seems like a pretty cool idea to me, especially if you're trying to sell people the "sound" of something in the real world they can plug their guitar into and compare it to. You don't want them to think your plugin doesn't measure up to the real thing simply because the input levels need to be a certain way to reproduce the sound they're familiar with, but nobody has this info.
Since plugin companies don't really do that, I'm wondering how would I deal with this if I was using the method described in this video to record more than one type of guitar, and wanted the experience you're describing with a plugin reacting like a real amp in that it behaves differently when fed different guitars and pickups, etc....
I think I'd try getting a setting dialed in on the audio interface using the loudest guitar(s)/pickup(s) out of the ones that I have. Then, I'd make a mark on the audio interface, take a picture of the knob's position, make a note, or something so that I could return to that setting whenever I input any of my guitars into the audio interface. I would dial in the settings on my amp sim plugin(s) using the guitar I play the most and the input level I just decided on with the audio interface.
I have not tested this (of course), but I'm guessing the difference in the noise floor created by the different signal levels when comparing two guitar pickups might only be a big deal if one of the pickups is noticeably noisier than the other, or if one is passive and the other is active.
Ok, enough of this. I'm gonna go record some things.
Jason Sadites just posted a video about this - he seems to agree with me 😀
But yes, fundamentally, to avoid constantly adjusting the input gain in the digital domain, the analog gain on the interface will always be a tradeoff between signal/noise ratio and the ability to pick up any guitar we have and avoid clipping. And yes, the signal/noise ratio lost (worst case is for lowest output pickups), if we are not using garbage interfaces, it's likely to be... in the noise (no pun intended).
I'm more confused now than I was before thanks dude
lol best thing i found was to barely raise it a little at a time until it sounds nice. 0 sounded upsettling weak for me..ended up around 2.3 which ended up being what neural dsp says on their site. (for QC as an interface at least) just go in small increments
It's not hard... 1. Optimise analog signal going in to interface, by increasing it to just below clipping; 2. Compensate digital signal going to your amp sim (i.e., in your DAW, or on the amp sim itself) by turning it down by the same amount.
@@matthijsheblynot quite.
1. Interface to zero
2. Digital gain plugin to chart (interface/amp sim matrix)
3. Increase analog gain to under clipping.
4. Reduce digital gain plugin by the amount you added to the analog stage. (If you added 10 dB in step 3, subtract 10 dB from the plugin you added at step 2.)
@@Hexspa You're basically saying the same thing, just with way more words and with lots of unnecessary confusion added...
@@matthijshebly not true. The whole point of having the chart is to match calibrated levels. That’s what the digital gain plugin step is for. If you leave that out, you’re right back to seeing things by ear.
0db is the dumbest thing I have ever heard! I was not aware this debate was happening, I thought everyone just turned up to the right level 😂 Why the hell would someone want to feed a weak signal into an amp sim?? The internet is insane sometimes ! Great video !
Yeah, I'm not understanding where this idea came from. Like... it's common knowledge to turn up to red and then roll back to green to find best signal strength
@@Citizen_J only thing I can think of is that they have had the software gain up too high, maybe software builders have even been setting up this way so it can handle the 0db “geniuses” 😂
@@Mr-Foad It could be that this information came from metal players who have high output pickups, so they're slamming the input hsrd enough that turning up isn't necessary. I know my passive bass doesn't work like that
I've been trying to clean up my distorted guitar sound for years with out luck and just learned to live with it as it is (since I'm not releasing it publicly).
Now I can't wait to go revisit all I have and dust it of.
Thank you so very much for this.
As a returning intermediate player, this was incredibly helpful and settled a lot of nonsense online. The difference is in the signal and this is it! Thanks brother
I think im an idiot. Can someone help me and make sure im doing this right? I have a Focusrite Solo 3rd Gen, Reaper DAW, and Neural DSP.
(Step 1) Set gain on my Solo to 0
(Step 2) Change the input gain knob on my Neural DSP to 0.3 (Number of 3rd gen Focusrite on spreadsheet)
(Step 3) Play my guitar and turn up the gain knob on my interface until it doesnt show yellow or red
(Step 4) ????????? This is where im confused, what does he mean by "Reduce digital input gain to compensate" Not sure why this is confusing to me?
He's saying to raise the gain on your interface until its just below clipping the taking the amount of gain you turned up on your interface and subtracting that from your input in the plugin (I think)
@@MortalFear-mt4ov Ok, that makes sense. I guess i just wouldnt know how to get that exact number
@@JP-ir7uw If your interface only has lights and no db meter:(howewer you could also open a daw and watch it. Anyway if you want to be precise: Record while turning the gain from 0 till just below clipping. Measure the recording. The differrence in db is what you subtract. So the beginning when it is set to 0. compare it to the in the recording part where it is set just below clipping. If you still don't know How to compare? Meter in your daw.
@@KadriYounes ok, since the difference from zeo to just below clipping is 7 db, what do we do next?
How do I know how much I've increased the gain by on my interface? By just estimating the percentage of dial turned up into percentage of maximum gain available from the interface?
I feel like this is a crucial bit of information that the OP left out for some reason.
I have the same question
The chart he shows has your Interface (I have the Scarlett) on the left. I assume you look across the top of the chart to see which plugin or DSP(I have a Helix) . Looking down the helix column the scarlet has 0.5 which corresponds to 12 o'clock . I may be wrong but it kinda makes sense to me lol
I tried this using NDSP with UAD volt. I used prefader metering in Logic to see what my guitar peaked at with interface gain set to 0. Then I checked the peak after raising interface gain to just below clipping. Get the difference of those values and turn down the plugin input gain by that value. Say prefader 0 peaks at -14 db and prefader under clipping peaks at -4. Turn down input gain in plugin down to -10 and add back the adjustment from the chart (+.3), so -9.7 in my case. Not sure if there’s a more precise way but would be interested.
It's a useless information.
Just push the gain until it clips, then turn down a bit.
What you have is the audio file in its best possible shape, and how much gain you used to get there is irrelevant from that point on.
Every intrument has its different output, every amp sim has its design.
Use your ears from that point on like you would do with any analog hardware.
If it clips or distorts, the input is too high, just a notch below that you'll have the best ratio.
Thank you for this great video. The problem with those influencers is that people give credits to those assholes without doing their own researches. People want the influencers to think for them.
thank you for making this video. i’ve been trying to explain this exact thing to other guitarists for years. hope this videos blows up
Makes perfect sense to me. Seems like a no brainer. I mean, you can't argue with facts like this. Thanks man. Great video!
I think a lot of people start recording on Focusrite Scarletts
On those interfaces (at least 3rd gen), simply setting the input to instrument mode and turning the gain all the way down gives you a decent signal for electric guitars
For those who want to keep their workflow, and don't have -20 dB on input in ampsim software: u can always use simplest "gain" plugin to trim signal to correct levels for further processing.
It still amaze me how many of myth are functioning in pro-audio sphere, it is just nuts.
it's understandable as most people don't know the technical details of how this all works behind the scenes. that's why the devs can solve this with an automatic digital input gain feature.
So I've got a question - how to determine how much input gain to take off the plugin after setting the input gain on the interface if the said interface doesn't show how much gain it's currently adding?
I had the exact same question. Idk if I'm right but I used a RMS meter using the mVmeter2 plugin from tbproaudio. Now it sounds awesome but I'm not entirely sure if what I did actually measures decibels. So yea, can someone help as well?
Wow that was so clear, loved this approach. Thanks for clarifying this
Most of this stuff is over my head, so I appreciate the way you break it down.