NAM calibration: find audio interface unity gain level

[jle]

Hi everyone,


Ive been testing NAM lately and Ive been interested in the calibration feature. My understanding (correct me if im wrong) is that if an amp was captured at 12 dbu and my audio interface input max headroom is 20dbu i should basically setup my interface at unity (+0db gain) and set NAM input gain at +8db.


Thats basically what the code of the client is doing here: https://github.com/sdatkinson/Neura...b8/NeuralAmpModeler/NeuralAmpModeler.cpp#L653


Im using an old audiobox 44vsl interface with -30/+50db gain range control. My problem is that unity position (+0db gain) is not marked clearly on the interface knob, so im wondering, is there a simple way to establish it with certainty using some calibrated test signal somehow? Ive searched the web and thought about it but i cant figure it out, am i missing something obvious here?

At first i was thinking about generating some sine wave at a certain dbfs level (lets say -30dbfs) in my daw (ex: tone test in ableton), get it out of the interface with main output volume at max position (+0db), plug it back in and find the knob position where I have the same -30dbfs input level in daw.
But im not sure that this makes any sense, since to my understanding different interfaces will produce different outputs for -30dbfs depending on interface output level (+10dbu in my case).

Any help/explanation will be deeply appreciated.

 

[2dor]

@MirrorProfiles has a few vids on his YouTube channel that walk you through finding out your instrument input "maximum input gain " / headroom if you don't have a spec sheet for the unit.

 

[jle]

I have the specs of my audio interface, maximum input gain is 21dbu. What i dont know is where the gain knob should be positioned to be at +0db.

 

[MirrorProfiles]

Hi everyone,


Ive been testing NAM lately and Ive been interested in the calibration feature. My understanding (correct me if im wrong) is that if an amp was captured at 12 dbu and my audio interface input max headroom is 20dbu i should basically setup my interface at unity (+0db gain) and set NAM input gain at +8db.


Thats basically what the code of the client is doing here: https://github.com/sdatkinson/Neura...b8/NeuralAmpModeler/NeuralAmpModeler.cpp#L653


Im using an old audiobox 44vsl interface with -30/+50db gain range control. My problem is that unity position (+0db gain) is not marked clearly on the interface knob, so im wondering, is there a simple way to establish it with certainty using some calibrated test signal somehow? Ive searched the web and thought about it but i cant figure it out, am i missing something obvious here?

At first i was thinking about generating some sine wave at a certain dbfs level (lets say -30dbfs) in my daw (ex: tone test in ableton), get it out of the interface with main output volume at max position (+0db), plug it back in and find the knob position where I have the same -30dbfs input level in daw.
But im not sure that this makes any sense, since to my understanding different interfaces will produce different outputs for -30dbfs depending on interface output level (+10dbu in my case).

Any help/explanation will be deeply appreciated.

You can set your gain to minimum, run a sine wave into the input, and then increase the gain knob until the sine wave dBFS level is 30dB higher.

Or if you have a multimeter you can do things more accurately.

 

[2dor]

I have the specs of my audio interface, maximum input gain is 21dbu. What i dont know is where the gain knob should be positioned to be at +0db.

Well then with the gain knob at minimum your interface Hi-Z input should be able to take in about 8.691 Vrms until you see 0 dBFS in the DAW.

I used this to convert the dBu rating to Vrms:

Volts RMS to dBu Converter

Convert Volts RMS to dBu and dBv. Convert dBu to Volts RMS and dBv.

benchmarkmedia.com

 

[2dor]

If you just want to play through NAM profiles but are worried about the best experience, look for profiles which have calibration metadata embedded. Tone3000 has a filter that you can use to only display those profiles.

You can then use NAM plugin player 0.7.12 or later as it has the logic for using the calibration metadata and with your interface's gain knob at 0 (where you know you have 21 dBu headroom) just type in "21" in the "Input Calibration" textbox within the NAM plugin.

More info here & example here:

NeuralAmpModelerPlugin v0.7.12 is released

The next version of NeuralAmpModelerPlugin is available! Download the installer from the users page.This release is all about signal level calibration. If you've been noting the input and output calibration levels in your models as you create them (more on that in this blog post about the...

www.neuralampmodeler.com

 

[jle]

You can set your gain to minimum, run a sine wave into the input, and then increase the gain knob until the sine wave dBFS level is 30dB higher.

Or if you have a multimeter you can do things more accurately.

You can set your gain to minimum, run a sine wave into the input, and then increase the gain knob until the sine wave dBFS level is 30dB higher.

Or if you have a multimeter you can do things more accurately.

Fuck, of course. I know that with minimum gain im at -30db, so yeah, whatever the input signal, i just have to turn it up till i reach "original value + 30db" and i know im at +0db.
Sorry guys, i feel stupid now, i guess i got lost in complicated scénarios. Thanks a lot for your answers.

Just for my knowledge, what would be the more accurate version using a multimeter?

 

[2dor]

Just for my knowledge, what would be the more accurate version using a multimeter?

 

[2dor]

..and you can sprinkle a little bit of this on top if you want to know more about how NAM does it (or how NAM profile authors should go about cooking the calibration metadata in their profiles):

How to determine the calibration levels of your recording setup — neural-amp-modeler 0.12 documentation

 

[2dor]

@jle have a look here. I had a very similar conversation with a few folks yesterday & decided to put together a quick vid to help make the most out of NAM & the new cablibration metadata:

 

[volkan]

i got that there is no perfect or best input gain value for anything until its clipping but regularly input gain should be higher than -18db will be enough other than that its your adjustment and depends every peace you imagine your picking your pickups your amp type your style your interface at the end its your ear.

 

[MirrorProfiles]

i got that there is no perfect or best input gain value for anything until its clipping but regularly input gain should be higher than -18db will be enough other than that its your adjustment and depends every peace you imagine your picking your pickups your amp type your style your interface at the end its your ear.

Not so. If you have your peaks at -18dBFS but the bulk of your signal is way lower, you could well be losing out on having a noisier signal than you need to.

Having your peaks at -18dBFS isn’t recommended for anything (besides constant voltage test signals). The reference level of +4dBu=-18dBFS is for converters with 22dBu of headroom, and that allows for 0VU=+4dBu to give you 18dB of headroom. This means you could use a VU meter where things are around 0VU and you should have enough headroom to avoid clipping. The signal peaks would be anywhere within the range of 0dBFS and -18dBFS depending on how loud your peaks are vs the average.

I made some videos on this topic as it’s often misunderstood:

 

[volkan]

Not so. If you have your peaks at -18dBFS but the bulk of your signal is way lower, you could well be losing out on having a noisier signal than you need to.

Having your peaks at -18dBFS isn’t recommended for anything (besides constant voltage test signals). The reference level of +4dBu=-18dBFS is for converters with 22dBu of headroom, and that allows for 0VU=+4dBu to give you 18dB of headroom. This means you could use a VU meter where things are around 0VU and you should have enough headroom to avoid clipping. The signal peaks would be anywhere within the range of 0dBFS and -18dBFS depending on how loud your peaks are vs the average.

I made some videos on this topic as it’s often misunderstood:

i am using two notes genom with namm profiles i just use the input and output sources shown in the plugin i mean i try to not clipping input and output in the plugin interface this will be enough for adjusting gain stage or should i use vu meter something ???

 

[volkan]

There is also input and output section shown in the namm interface these aren't enough for the gain staging i am trying to prevent clipping on both sides

 

[volkan]

There is also input and output section shown in the namm interface these aren't enough for the gain staging i am trying to prevent clipping on both sides

i want to learn that every plugin has own input and output charts so these are enough for adjusting gain stages ??
Do we need to do another adjustments etc??
i am adjusting input gain level on my sound interface and checking out if there is clipping on the plugin if i dont see any clipping its ok for me and do the same for the ouput also.

 

[[Nathan]]

The combo of nam and input levels seems to be the most confusing for people to grasp (fair enough tbh, the most variables). There’s many videos out there and if you want to go into the details then I’d watch those.

I always think of it with amp sims first. Amp sim plugins are calibrated to a signal level. Don’t think of this in terms of guitar pickups where some are high output and some are low output, none of that matters. It’s calibrated to some kind of fixed level. Then you have your audio interface input level. The high level goal is to match your interface input level to the amp sim level. More than likely when these match then whatever guitar you’re using probably won’t clip in the daw, most interfaces have enough headroom. If it did clip then you would essentially turn the interface down by whatever, say 4db. If you did that you’d turn the plugin input level down by 4db so they still match.

All of this applies to NAM it’s the same thing. The difference is there’s a bunch of nam vsts and also many people making nam models and some put the input levels in the metadata and some don’t. If you’re using the official NAM player you’d tell it your audio interface input level. When you choose a NAM profile if it has the metadata in it, it will auto adjust and you don’t need to think about it.

The problem is if you’re using another nam vst they might not have this functionality so you have to do that manually
Also if you’re using NAM captures with no input metadata then you’ll also need to do that manually
Most people don’t want to sit there and do these mental calculations so they just approximate the input level when playing random nam captures.

For me personally I use the stock nam player because it has this feature, and I only use nam captures with the input metadata in them so it auto adjusts.

 

[2dor]

@jle have a look here. I had a very similar conversation with a few folks yesterday & decided to put together a quick vid to help make the most out of NAM & the new cablibration metadata:

If anybody's interested, I did a follow-up to my last one (on calibrated NAM profiles) which goes into daisy-chaining / stacking calibrated NAM profiles for the most accurate response:

 

[volkan]

İsim ve giriş seviyelerinin birleşimi, insanların kavraması en zor olan kısım gibi görünüyor (doğrusu, en çok değişkene sahip olması da anlaşılabilir). Bu konuda birçok video var ve detaylara girmek isterseniz onları izlemenizi tavsiye ederim.

Ben her zaman önce amfi simülasyonlarıyla düşünüyorum. Amfi simülasyon eklentileri bir sinyal seviyesine göre kalibre edilir. Bunu, bazılarının yüksek çıkışlı, bazılarının düşük çıkışlı olduğu gitar manyetikleri açısından düşünmeyin, bunların hiçbiri önemli değil. Bir tür sabit seviyeye göre kalibre edilir. Sonra ses arayüzünüzün giriş seviyesi var. Yüksek seviyedeki hedef, arayüzünüzün giriş seviyesini amfi simülasyon seviyesiyle eşleştirmektir. Büyük olasılıkla bunlar eşleştiğinde, kullandığınız gitar büyük olasılıkla DAW'da kırpılma yapmaz, çoğu arayüz yeterli boşluk payına sahiptir. Eğer kırpılma olursa, arayüzün sesini örneğin 4 dB kadar kısmanız gerekir. Bunu yaparsanız, eklentinin giriş seviyesini de 4 dB kadar kısmanız gerekir, böylece yine de eşleşirler.

Bunların hepsi NAM için de geçerli, aynı şey. Fark şu ki, bir sürü NAM VST eklentisi var ve ayrıca birçok kişi NAM modeli yapıyor; bazıları giriş seviyelerini meta verilere eklerken bazıları eklemiyor. Resmi NAM oynatıcısını kullanıyorsanız, ses arayüzünüzün giriş seviyesini belirtmeniz gerekir. Bir NAM profili seçtiğinizde, eğer meta veriler içeriyorsa, otomatik olarak ayarlanır ve sizin düşünmenize gerek kalmaz.

Sorun şu ki, başka bir isim VST eklentisi kullanıyorsanız, bu işlev mevcut olmayabilir, bu nedenle bunu manuel olarak yapmanız gerekir.
Ayrıca, giriş meta verisi olmayan NAM yakalamaları kullanıyorsanız, bunu da manuel olarak yapmanız gerekecektir.
Çoğu insan oturup bu zihinsel hesaplamaları yapmak istemediği için rastgele isim yakalama oyunları oynarken giriş seviyesini yaklaşık olarak belirler.

Şahsen ben bu özelliğe sahip olduğu için varsayılan NAM oynatıcısını kullanıyorum ve sadece giriş meta verileri içeren NAM yakalama dosyalarını kullanıyorum, böylece otomatik olarak ayarlanıyor.

yes man thats right
what i mean and at the end it will be your ear.

 

[FrankyFire]

@jle have a look here. I had a very similar conversation with a few folks yesterday & decided to put together a quick vid to help make the most out of NAM & the new cablibration metadata:

I saw your video and got confused. You turned your Interface Input as low as possible and never turned it up again. No wonder there's tons of hiss. I was hoping to get a guide on how to set it up so that I can still add gain on the Interface input to have a better SNR.
Still I do appreciate your effort in trying to shed light on this topic!

The combo of nam and input levels seems to be the most confusing for people to grasp (fair enough tbh, the most variables). There’s many videos out there and if you want to go into the details then I’d watch those.

Sadly many of them are even more confusing or just outright wrong or lead to bad results (like lots of hiss, as just stated).

If it did clip then you would essentially turn the interface down by whatever, say 4db. If you did that you’d turn the plugin input level down by 4db so they still match.

Is this a typo? You probably mean to turn up the Plugin Input level up by 4dB when turning down the Interface Input by 4dB. Anything else doesn't make any sense, as it would just turn your Input signal down by 8dB.

If you’re using the official NAM player you’d tell it your audio interface input level

Which I would have to adjust when turning down my Input as well in this case. But how do I even know how far I've turned it down? It would require me to have anything measuring the gain difference or tell me somehow. My interface has no scale or software to check the gain level.

As far as I understand, the Input gain is not an easy thing to figure out. I would have to have a Signal with an exact level, say -10dB. I could plug that into my Interface, set the gain and from now on know how much gain I'm adding by checking the level inside my DAW. But I don't know the output level of any source to begin with. Which is my problem with this solution too:

You can set your gain to minimum, run a sine wave into the input, and then increase the gain knob until the sine wave dBFS level is 30dB higher.

So the only alternative in my eyes is to measure the output of any source using a potentiometer and do the same thing. Last time I checked that's basically what the manual for creating NAM profiles with calibration states on that topic too.
But I haven't found a video on that so far.

Am I wrong here somewhere?

 

[MirrorProfiles]

But I don't know the output level of any source to begin with. Which is my problem with this solution too:

Make a sine wave, measure the voltage in the analog domain with a multimeter. Then feed this sine wave into your input, note down the level in dBFS. You can calculate your headroom with that.

Turning to 0 doesn't mean extra hiss, unless your signal is extremely quiet and the noise floor of the interface is loud enough to be heard over the other noise in the signal chain. For most interfaces in 2026, you will not have an issue with hiss, nor will you benefit in SNR by turning your interface up by a few dB (as they'll already be hot).

If you have issues with low signal/high noise, then you'll need to measure your levels and set the gain accordingly. All the info you need is in this forum and on YouTube.