Topic: Do recording gain is so important if we use 24bit quantization?

[Dave_Scream]

Hello.
The rules are sayin than before recording we need to setup recording gain to average loudness of sound and then record.
But often during recording process some abnormal loud sounds happen and in that cases we have clipping or limiter effect appear in recording.

Today I think. Why all manuals say that it is better to record sound on maximum recording gain (but not clipping)  Why?   Maybe when recording is in 16bit - the dynamic range is not so big (65000 amplitude values) and  too low singal when normalized will have BIG quantization noise.

BUT if we record on 24bit depth  do we need to afraid this quantization noise?  Maybe if we have 24 bit depth, we can record on low recording gain and then just normalize whole record after process? and in this case even abnormally loud sounds well be recorded without any distortions or limiter effects.

Maybe 24 bit depth allow us to leave more space to extra loud sounds? and allow us to make "standart gain" lower then before? after recording we can just normalize it and there is no quantization noise because we use 24bits

[Dave_Scream]

Also I found some interest text here

Hopefully you're still with me, because we can now go on to precisely what happens with bit depth. Going back to the above, when we add a 'bit' of data we double the number of values available and therefore halve the number of quantization errors. If we halve the number of quantization errors, the result (after dithering) is a perfect waveform with halve the amount of noise. To phrase this using audio terminology, each extra bit of data moves the noise floor down by 6dB (half). We can turn this around and say that each bit of data provides 6dB of dynamic range (*4). Therefore 16bit x 6db = 96dB. This 96dB figure defines the dynamic range of CD. (24bit x 6dB = 144dB).

So, 24bit does add more 'resolution' compared to 16bit but this added resolution doesn't mean higher quality, it just means we can encode a larger dynamic range. This is the misunderstanding made by many.

24bits 144dB dynamic range is more than dynamic range of professional microphones

[page]

As a result of 24bit audio and high quality converters, the standard peaking for many has changed, from -3dbfs or -6dbfs to around -12dbfs or -15dbfs without noticeable ill effects.

That doesn't detract from the skill and experience involved in using of perfectly good gain structuring as quantization noise isn't the only noise that sits at the bottom of a recording.

[Dave_Scream]

allright I make some tests.

Recorded at normal gain and then on zero gain.  After recording I normalized zero gain to peak value. And get too many hiss noise. Also in soundforge I see that after normalizing there is "boxes" or "squares"  so 24bits is not enough for zero gain recordings and normalization after recordings - there is very big quantization error so waveform aplitude is squared. Also very big hiss appear after normalizing

[page]

yeah, one or two members here did tests and found that the ratio for signal:noise started to climb when they were having to add over 30 or 40db of gain digitally via normalization.  I can't remember the exact post, but I remember seeing tests done. If you're interested, that's what I'd start looking at.

[DSatz]

To the original poster: You're looking at the specifications in the wrong way. You always need to look at the dynamic range of the recording system as a whole, from the source (which usually has its own, non-negligible noise floor) to the analog input circuitry of the A/D converters.

There are real-world audio signals that come from real-world microphones in the real world, where the signals have such a large ratio between their noise floor and their peak amplitude that those signals can't be quantized in 16 bits without (a) overload, (b) potentially audible noise being added in the process, or (c) both.

But there are NO such audio signals for which 24 bits are required. You can synthesize them and you can dream them up in thought experiments all you like, but you can't get them from real-world microphones placed in the real world and connected to real-world preamps. And that's actually a good thing, in a way, because no real-world 24-bit A/D converter has a 24-bit dynamic range, either.

As a result, in any 24-bit recording, some number of the lowest-order bits will contain nothing but noise. The question is not whether, but how many of those bits are wasted. It's always at least two of the 24, but usually it's several more; often it's all eight of the added bits that 24-bit gives you, plus even a few of the 16 that you might have started with.

--best regards

P.S.: The signature line quoted above, "Nostalgia isn't what it used to be," is the title of the late, great Simone Signoret's autobiography, but I'm sure the line wasn't even original with her.

[Gutbucket]

Two people far more qualified than I just answered while I was typing this, so much of the below is a rehash of what they said worded slightly differently, but I'll post anyway.  My numbers my be off slightly but the basic idea is sound-


Your basic concept is correct.  More bits = more dynamic range.  A 24 bit file can encode 144dB of range but not even the very best analog equipment is capable of 144dB.  The limit is the analog circuitry in the recording chain before the signal is digitized.  Good studio equipment can manage about 20bits of usable dynamic range. The relatively inexpensive gear most of us use around here is probably closer 18bits when set to write 24bit files, most of it probably closer to 17bits, some no better than 16.  The 24bit files will simply have multiple bits of random analog noise at the bottom.

Dynamic range is limited by clipping distortion at the top and dither noise at the bottom.  The noise at the bottom is not quantization noise in any modern digital recorder, it is dither, which is more benign and sounds like analog hiss (because it is) and which is used to eliminate any quantization noise which is not benign and sounds terrible.   But that dither noise is going to be much lower than the analog noise of the circuitry in almost all cases, and most of the time, that analog circuitry noise is going to be lower than the ambient noise of the environments in which we record.

So in the 'real world' of concert recording, if you figure out how to set things properly you can do pretty much exactly as you suggest.  I now do that myself with one of my setups-

I still adjust the gain to peak at about -15 or -18dbFBS at the start and leave it there without worries when I'm running mics and recorders onstage or whatever, but with my worn surround rig, I've figured out where to set things so that I can leave the gain set the same for everything I record, partly because it's a pain to match the gains across all 4 channels evenly if I needed to change it.  Instead I leave it set low enough to handle the loudest stuff, and let the quiet stuff peak lower than most people would want it, then amplify it digitally later.  Even for very, very quiet music in very quiet rooms, the ambient noise floor of the room is higher than the analog noise floor of my recording equipment which is maybe getting about 17-18dB of range if I'm lucky (but I'm guessing about that).  And I'm talking things like certain sections of classical material with sound levels hovering around the threshold of hearing in a couple halls designed for performance with quiet HVAC and isolation from outside noises. Some of those recordings peak at best around -30dB, which used to freak me out, but they sound fine once amplified.  I've only noticed the analog noise on super quiet nature recordings using this setup, which isn't it's intended purpose anyway.

[page]

P.S.: The signature line quoted above, "Nostalgia isn't what it used to be," is the title of the late, great Simone Signoret's autobiography, but I'm sure the line wasn't even original with her.

Yeah. I've heard a couple use it before, but I grabbed it from one of his posts over on GS. Even taken mildly out of context, it's still one of my favorite lines.

Typing on a tablet sucks sometimes, glad folks filled in with a respectable answer.

[Dave_Scream]

Thanks for answers. I understand that dynamic range of my analog to digital converter is far from 144dB (24bit) and that noise of electronics have the best ratio at ~50% of gain.

So because of it optimal is to set recording gain to 50%. But if sound is too quiet on 50% of gain, then to avoid quantization effect (of not real 24bit electronics), need to move up the recording gain. So quantization effect is lower, but electronics noise because of higher gain is louder.

[bombdiggity]

It's also too simple to say signal to noise is always best at 50% gain.  Dynamic range is best at minimum gain; equivalent input noise is usually best at maximum gain.  Dynamic range will be degraded by nearly 1dB for each 1dB of gain above a certain point.  What that point is will vary by mic, preamp, and converter combination.

^ Recording is all about estimating and managing the tradeoffs in real time...