HydrogenAudio

Hydrogenaudio Forum => General Audio => Topic started by: botface on 2012-02-28 19:13:51

Title: 16 bit, 24 bit and Noise Floor
Post by: botface on 2012-02-28 19:13:51
I'm involved in a discussion on a home recording forum. We are not discussing post processing, mixing or mastering just the simple capture of a single acoustic instrument.

I'm suggesting that with a recording environment that yields a noise floor of -60dbfs it makes no difference at whether you capture in 16 bit or 24 bit as you effectively only have 10 bits to work with and so any theoretical benefits of using 24 bit are lost

I'm getting arguments back like :
Quote
Specifically 24 bit capture files will still provide the widest range of dynamic and the greatest resolution for capturing anything


Quote
With 16 bit you have that many steps and with 24 you have that many more ... which means the more bits you have, the less grainy your resolution is


Quote
To give an example: You have a scale of 0 to 100. If you print it on a 10cm piece of metall you barely have enough place to distinguish 1mm distances. If you print it onto a 1m bar, you can have plenty more subdivisions. Of course 100 is still 100 and zero is still zero, but you have many more subdivisions. Same with digital signal. Say you would use a 2 db resolution: then you would have just four different levels to represent the different signal levels which would have to be somehow digitalized onto these four steps.
With 16 bit you have 2^16 possible steps (65536) and with 24 bits 2^24 possible steps (16777216, quite a bit more...). Whether you believe them to be more useful to represent an audio signal I leave to your own ears.I'm not claming that the perceived distance between noise level and maximum level is bigger with 24 bits than with 16 bits. It's just that you have many more subdivisions in between, and that is audible.


Quote
If the full theoretically possible dynamical span gets more quantizing levels per dB so must also the same dynamical span of let's say 60dB gain above recorded noise floor. I' can't see where this argument is wrong ... Maybe we're talking about different things?


Quote
16 bit recordings have a theoretical dynamic span of 96dB; 24 bit recordings one of 144 dB. If I take the number of possible representations of level with 16 bit, then I have 2^16/96 = 682,7 (rounded) steps per 1dB. WIth a 24 bit recording I have 2^24/144 = 116508,4 (again rounded) steps per 1 dB. So it seems that the dynamical resolution is much finer with 24 bit.


Quote
The scale is the same, but with 24 bits we get bigger number of smaller steps than with 16 bit at any part of the scale which allows for better resolution.
You can want to increase the dynamic range by cutting off the noise, and stretching what is left down to the negative infinity level to fill the whole range, and higher resolution of 24 bit will become very handy compared to 16 bit.


I've given examples, I've worked through the arithmetic, I've provided analogies but nobody seems to think the noise floor of the recording environment has any impact on the ability of 24 bit to capture more detail. Can any of you provide any examples or analogies that you've found to work in the past.
Title: 16 bit, 24 bit and Noise Floor
Post by: saratoga on 2012-02-28 19:17:18
You are correct and those people you quoted do not know what they're talking about.
Title: 16 bit, 24 bit and Noise Floor
Post by: greynol on 2012-02-28 19:18:29
As a simple experiment, you can create a pure tone that is quieter than the noise floor and you'll likely find that you can still hear it.
Title: 16 bit, 24 bit and Noise Floor
Post by: itisljar on 2012-02-28 19:21:21
Oh, just work in 24 bits so you can have more precise capture. I don't see anything wrong with that. Any postprocessing will be more precise with 24 bit audio.
Theoretically, you can capture 12 bit and still be above room noise. But, why?
Title: 16 bit, 24 bit and Noise Floor
Post by: DVDdoug on 2012-02-28 19:26:36
Let's look at the "ruler" example -
Quote
To give an example: You have a scale of 0 to 100. If you print it on a 10cm piece of metall you barely have enough place to distinguish 1mm distances....


As you mentioned, the problem is noise.  Noise is randomness and uncertainty.  It's as if those 1mm distances are are jumping-around 10mm.    What good does it do to have 1mm worth of resolution when you have less than 1mm of accuracy or repeatability. 

People make the same arguement about analog tape or vinyl having "infinite" resolution.  But, the noise means that you have less resolution than you get with 16-bits.

On the other hand, all profesional interfaces and most simi-professional interfaces are 24-bits, so you might as well use all 24 bits.
Title: 16 bit, 24 bit and Noise Floor
Post by: greynol on 2012-02-28 19:43:30
Please, before submitting any more abject analogies, generate white noise clip with an RMS power of -48dB and then superimpose a 1kHz sine wave with an RMS power of -60dB and tell me if you hear the 1k tone.

If you want, I'll gladly provide a sample.
Title: 16 bit, 24 bit and Noise Floor
Post by: KMD on 2012-02-28 19:46:45
greynol is right, sounds below the noise floor are audible. Noisy recording processes are additive so using 24 bit  to record a noisy enviroment is valid. Anyway the enviroment may be considered to be part of the programe material to be captured.
Title: 16 bit, 24 bit and Noise Floor
Post by: saratoga on 2012-02-28 19:51:01
Please, before submitting any more analogies, generate white noise clip with an RMS power of -48dB and then superimpose a 1kHz sine wave with an RMS power of -60dB and tell me if you hear the 1k tone.


If I understand correctly, the question is not about hearing below the noise floor, but rather if more bits are somehow "better" independent of the actual dynamic range available.  My answer to this is no.  Once the quantization error drops significantly below the noise floor of the equipment, its essentially irrelevant. 

Apologies if I've misunderstood your point.
Title: 16 bit, 24 bit and Noise Floor
Post by: greynol on 2012-02-28 20:22:08
You're right.

Case in point: create a 1k tone in 24 bits that peaks at -103dB and then convert it to 16 bits with dither.
Title: 16 bit, 24 bit and Noise Floor
Post by: mjb2006 on 2012-02-28 20:22:56
Most explanations—even those for technical audiences—of the effect of sample rate and bit depth in sampling generally boil down to the idea of "capturing more detail". I think this is a big part of the problem; everything that people learn about digital audio gets integrated with this basic tenet. You can talk until you're blue in the face about noise, Nyquist, and everything else, but you're still going to run up against this perception: there is a squiggly line which is being segmented along the X and Y axes with a certain degree of precision in order to represent it as a series of numbers, and if you crank up the knobs that control the precision along one axis or the other, you're going to get a better representation of the original squiggly line, so why not do that. When someone's thinking that way, what can you do? You can tell them that if all the music is between 0 and 5, making a knob that goes to 11 instead of 10 is not going to give you "more detail". But they just aren't convinced. So I'm not sure what you can really do. There's still a need for a sampling primer for audiophile myth-busting purposes.
Title: 16 bit, 24 bit and Noise Floor
Post by: saratoga on 2012-02-28 20:30:19
You're right.

Case in point: create a 1k tone in 24 bits that peaks at -103dB and then convert it to 16 bits with dither.


Ah yes I see what you are getting at now. 

botface:  If you're goal is to show these people how quantization works, perhaps a simple way to do that would be a few example files of pure tones in white noise with a given SNR quantized to different bit depths as greynol  suggests.
Title: 16 bit, 24 bit and Noise Floor
Post by: greynol on 2012-02-28 20:39:38
I should have led with my second example.
Title: 16 bit, 24 bit and Noise Floor
Post by: WernerO on 2012-02-29 07:36:24
I'm suggesting that with a recording environment that yields a noise floor of -60dbfs it makes no difference at whether you capture in 16 bit or 24 bit as you effectively only have 10 bits to work with


That is probably not quite correct. Assuming that your '-60dBFS' noise level is noise integrated over a meaningful bandwidth (for instance a reading on a level meter) and that you are in a real room with a real system, the spectral distribution of that noise signal is most likely skewed severely to the lower frequencies. That means that once above a few 100Hz you will need more than 10 bits to capture all, simply because the ambient noise there is much lower than the integrated meter reading wants you to believe.


perhaps a simple way to do that would be a few example files of pure tones in white noise with a given SNR quantized to different bit depths


4 bit demo, including music, here http://www.tnt-audio.com/sorgenti/dither_e.html (http://www.tnt-audio.com/sorgenti/dither_e.html)
Title: 16 bit, 24 bit and Noise Floor
Post by: knutinh on 2012-02-29 10:00:07
That is probably not quite correct. Assuming that your '-60dBFS' noise level is noise integrated over a meaningful bandwidth (for instance a reading on a level meter) and that you are in a real room with a real system, the spectral distribution of that noise signal is most likely skewed severely to the lower frequencies. That means that once above a few 100Hz you will need more than 10 bits to capture all, simply because the ambient noise there is much lower than the integrated meter reading wants you to believe.

A good point. As long as no precise spectral estimate of a)background noise and b)16-bit quantization error is provided and compared, it is difficult to be confident that 24 bits has exacly zero perceptual value.

-k
Title: 16 bit, 24 bit and Noise Floor
Post by: Garf on 2012-02-29 14:48:45
Please, before submitting any more abject analogies, generate white noise clip with an RMS power of -48dB and then superimpose a 1kHz sine wave with an RMS power of -60dB and tell me if you hear the 1k tone.

If you want, I'll gladly provide a sample.


This is slightly surprising to me. Not the effect itself, but the 12dB difference. IIRC, all encoders assume Noise-Masks-Tone thresholds in the neighborhood of 6dB. The white noise will be over the entire spectrum so this shouldn't be an issue of inter vs intra band masking either.

Edit: After looking at this in a spectrogram, it's very obvious. White noise at 48khz with an RMS of -48dB has a spectral floor below -80dB. The tone peaks at -60dB, so obviously it can't be masked. The difference is that the tone concentrates all its RMS power in a single frequency, whereas the noise, while louder in RMS, has to spread that power over all frequencies.

The original post talked about a -60dB noise floor. What does this mean, exactly? If this is a spectral floor, then anything that quantizes from -66dB or better (11 bits) will have no audible advantage. If this is an integrating meter reading, then WernerO is spot on.
Title: 16 bit, 24 bit and Noise Floor
Post by: icstm on 2012-02-29 17:56:55
your point about single tones vs white noise is also shown on THIS (http://www.audioholics.com/education/audio-formats-technology/dynamic-comparison-of-lps-vs-cds-part-4/dynamic-comparison-of-lps-vs-cds-part-4-page-2)thread when they are looking CD vs the black stuff.
Title: 16 bit, 24 bit and Noise Floor
Post by: Garf on 2012-03-01 07:40:29
your point about single tones vs white noise is also shown on THIS (http://www.audioholics.com/education/audio-formats-technology/dynamic-comparison-of-lps-vs-cds-part-4/dynamic-comparison-of-lps-vs-cds-part-4-page-2)thread when they are looking CD vs the black stuff.


If you mean that only looking at the RMS level of the signal doesn't tell you much about the noise floor, yes. You don't need to look any further than at DSD to make that very obvious.

The author makes some severe mistakes that void the entire comparison: he's assuming that the dynamic performance of his LP is constant over the input level. This is, AFAIK, absolutely not true due to the physical characteristics. This can be seen clearly at the end of the linked page, and to the authors credit, he admits this in his conclusion (but somehow claims its an advantage - quite the unbiased comparison there). The entire comparison would also be broken if there is so much as a smooth lowpass filter present on the LP, which, given the admitted harmonics distortions, would be quite likely anyway. Lastly, he investigates silence on a digital CD and concludes it has an RMS power of -103dB. I have no words for this. The RMS power of a silent CD signal is minus infinity. Zip, nada.

He's also trying to draw some conclusions from the peak sample level, which is just...sigh.

This all a bit tangential to our discussion. The original posters statement of "a recording environment that yields a noise floor of -60dbfs" just doesn't convey enough information to know whether recording at more than 10 bits makes sense or not. If this noise floor is caused by a 50Hz transformer hum at -60dB, and the spectral floor at all other frequencies is significantly lower, then recording at 24 bits may certainly provide advantages, especially if there are multiple mastering steps involved. If this noise floor exists because of broad-spectrum ambient noises, he might as well use a tape recorder.
Title: 16 bit, 24 bit and Noise Floor
Post by: botface on 2012-03-01 09:13:32
The original posters statement of "a recording environment that yields a noise floor of -60dbfs" just doesn't convey enough information to know whether recording at more than 10 bits makes sense or not. If this noise floor is caused by a 50Hz transformer hum at -60dB, and the spectral floor at all other frequencies is significantly lower, then recording at 24 bits may certainly provide advantages, especially if there are multiple mastering steps involved. If this noise floor exists because of broad-spectrum ambient noises, he might as well use a tape recorder.

Garf,
      I talking about the ambient room noise, mic self noise, preamp noise etc so it just general broadband noise.

@everybody else,
                        Thanks for the responses
Title: 16 bit, 24 bit and Noise Floor
Post by: 2Bdecided on 2012-03-01 09:51:38
Try lossyWAV (http://www.hydrogenaudio.org/forums/index.php?showtopic=90104) with --bitdist to see how many bits it thinks you need to keep.

Cheers,
David.
Title: 16 bit, 24 bit and Noise Floor
Post by: Garf on 2012-03-01 10:19:29
Try lossyWAV (http://www.hydrogenaudio.org/forums/index.php?showtopic=90104) with --bitdist to see how many bits it thinks you need to keep.


Preferably with options that don't assume the quantizer is noise-shaping, though  He also can't control the quantization on a block-by-block basis, so he'd have to use the worst-case.
Title: 16 bit, 24 bit and Noise Floor
Post by: Ethan Winer on 2012-03-01 18:59:29
As a simple experiment, you can create a pure tone that is quieter than the noise floor and you'll likely find that you can still hear it.

Sure, but not if the signal is 96 - 60 = 36 dB below the noise, using the OP's example of room noise which is typical. Of course, all noise is not uniformly broadband, but the basic principle is valid. You can hear maybe 10 dB below the noise, but certainly not 36 dB unless the noise is sculpted to remove frequencies present in the source.

--Ethan
Title: 16 bit, 24 bit and Noise Floor
Post by: Arnold B. Krueger on 2012-03-02 00:24:33
That is probably not quite correct. Assuming that your '-60dBFS' noise level is noise integrated over a meaningful bandwidth (for instance a reading on a level meter) and that you are in a real room with a real system, the spectral distribution of that noise signal is most likely skewed severely to the lower frequencies. That means that once above a few 100Hz you will need more than 10 bits to capture all, simply because the ambient noise there is much lower than the integrated meter reading wants you to believe.

A good point. As long as no precise spectral estimate of a)background noise and b)16-bit quantization error is provided and compared, it is difficult to be confident that 24 bits has exacly zero perceptual value.


The general point to be made is that a dB for dB masking of a lower amplitude noise floor only  happens when both noise floors have identical spectral shaping. However, just because the two noise floors have different spectral shapes does not mean that all bets are off.

The right thing to do is to subtract the two on a point-for-point, octave-for-octave or whatever relevant means of comparison you have available , and then apply an appropriate audibility weighting curve to the difference.  Since the levels of noise floors are pretty low, the roll off of high and low frequencies will usually be pretty extreme.  This tends to minimize the original differences in the shapes of the spectral content of the two noise floors.

Having measured a ton of noise floors over the years, i've seen just about every rule of thumb to fall apart. Probably the canonical noise floor of a room will be red or brown noise-shaped, but a heavy dose of turbulence in the air flow of the HVAC system can wash a lot of that out. Some microphones have noise floors that are shaped more like white noise, while other are shaped more like pink noise. Complementary equalization curves (e.g. RIAA) can distort the noise floors of electronics between the pre-emphais and de-emphasis networks.  Finally, a 60 dB noise floor is almost a whopping 40 dB away from a 96 dB noise floor and that tends to wash our a lot of minor differences.

Also, while the noise floor of real world 16 bit systems is usually within a few dB of 96 dB, very few real world 24 bit systems come within 20 dB of 24 bits. Usually, going to 24 bits from 16 is only good for a 10-20 dB advantage and sometimes it is as little as 5 dB.

Title: 16 bit, 24 bit and Noise Floor
Post by: Glenn Gundlach on 2012-03-02 06:36:51
As a simple experiment, you can create a pure tone that is quieter than the noise floor and you'll likely find that you can still hear it.


I have a Nakamichi 580M cassette deck with a factory claimed S/N ratio of 65dB. I also have a Sony test CD that a friend wanted copied to cassette. The disc includes 1KHz tones 0, -10 down to -90dB in 10 dB steps. It was easiest to simply record the cassette in one pass even though many of the tones were below -65. And yes, the machine was properly set up and the reference level was correct. I played it back as a joke and was very surprised to hear ALL the tones start and stop including -90dB. No ABX was needed as everyone could hear it clearly. 1/2 lsb dither produces similar results in digital but I haven't tested at 25dB below the noise floor though it might be an interesting experiment.

Title: 16 bit, 24 bit and Noise Floor
Post by: Garf on 2012-03-02 11:07:37
I have a Nakamichi 580M cassette deck with a factory claimed S/N ratio of 65dB...but I haven't tested at 25dB below the noise floor though it might be an interesting experiment.


A S/N ratio of 65dB does not imply a flat noise floor -65dBfs. I'd expect a cassette deck to have a better S/N ratio at lower frequencies than at high ones. This is why Dolby B/C works.

As far as I know, you can't hear tones more than about 6dB below the noise floor.
Title: 16 bit, 24 bit and Noise Floor
Post by: Arnold B. Krueger on 2012-03-02 13:02:08
This is another one of those situations where your mileage varies all over the map depending on the details. One common-sense requirement is that the tone be audible when there is no noise. 

The spectral content of the noise is a very important parameter. In audio systems composed of just electronics, the spectrum of the noise floor can commonly range from pink to white to the sort of peaked-up just below Nyquist spectrum that results from using perceptually-shaped dither.  The actual listening level used in the evaluation is also very important.

It is common for people to judge these things with the signal level boosted by 40 dB or more, as compared to a normal gain setting based on comfortable listening to normal music. This asymmetry is very popular with people who think they have just invented the perfect new dither that will solve all of the problems of digital.  This throws things so out of whack in comparison with actual practical use that it isn't even a bad joke. ;-)
Title: 16 bit, 24 bit and Noise Floor
Post by: googlebot on 2012-03-02 19:59:34
A noise floor can be considered part of the signal in many situations - and be worth capturing as is.

To add another bad analogy: I always prefer MPEG2 BluRays, because they preserve film grain and noise much more conservatively than certain H.264 encoders, which add some form of softening even at very high bitrates. One could claim that this is a feature, because the latter encoder prefers allocating bandwidth to even more detailed signal than noise. But sometimes I want the noise.
Title: 16 bit, 24 bit and Noise Floor
Post by: knutinh on 2012-03-02 20:09:49
A noise floor can be considered part of the signal in many situations - and be worth capturing as is.

To add another bad analogy: I always prefer MPEG2 BluRays, because they preserve film grain and noise much more conservatively than certain H.264 encoders, which add some form of softening even at very high bitrates. One could claim that this is a feature, because the latter encoder prefers allocating bandwidth to even more detailed signal than noise. But sometimes I want the noise.

I know of this preference within the cinema crowd.

I have never heard of such a thing within audio. I understand that many object to the artifacts associated with noise reduction, but preserving the crackles of vinyl or air-conditions of some recording venue does not seem to appeal to most music-lovers.

-h
Title: 16 bit, 24 bit and Noise Floor
Post by: googlebot on 2012-03-02 20:18:35
As discussed earlier in the thread it is not trivial to decide, what kind of noise can be safely drowned by taking away bits. It's easy to decide for certain well defined types of noise, but you seldom find them as exactly that in the analog domain. There is brownish, pinkish, whitish, but not brown, pink, white. Especially when further processing is involved, you might want to preserve the original floor as good as possible.
Title: 16 bit, 24 bit and Noise Floor
Post by: 2Bdecided on 2012-03-03 18:56:56
This is another one of those situations where your mileage varies all over the map depending on the details.
There's no magic though. Just look at the spectrum. That tells you what you need to know. We stop hearing the signal where the noise power falling into a critical band is about the same as the signal power. You can get stranger results if you modulate the noise.

Cheers,
David.
Title: 16 bit, 24 bit and Noise Floor
Post by: Garf on 2012-03-04 09:12:07
A noise floor can be considered part of the signal in many situations - and be worth capturing as is.

To add another bad analogy: I always prefer MPEG2 BluRays, because they preserve film grain and noise much more conservatively than certain H.264 encoders, which add some form of softening even at very high bitrates. One could claim that this is a feature, because the latter encoder prefers allocating bandwidth to even more detailed signal than noise. But sometimes I want the noise.

I know of this preference within the cinema crowd.

I have never heard of such a thing within audio.


A noisy signal will have less distortion introduced (noise modulation iirc?) when its quantized heavily as compared to a less noisy signal. That's why using dither is recommended.
Title: 16 bit, 24 bit and Noise Floor
Post by: spindle on 2012-03-04 11:12:51
Most explanations—even those for technical audiences—of the effect of sample rate and bit depth in sampling generally boil down to the idea of "capturing more detail". I think this is a big part of the problem; everything that people learn about digital audio gets integrated with this basic tenet. You can talk until you're blue in the face about noise, Nyquist, and everything else, but you're still going to run up against this perception: there is a squiggly line which is being segmented along the X and Y axes with a certain degree of precision in order to represent it as a series of numbers, and if you crank up the knobs that control the precision along one axis or the other, you're going to get a better representation of the original squiggly line, so why not do that. When someone's thinking that way, what can you do? You can tell them that if all the music is between 0 and 5, making a knob that goes to 11 instead of 10 is not going to give you "more detail". But they just aren't convinced. So I'm not sure what you can really do. There's still a need for a sampling primer for audiophile myth-busting purposes.


Forgive me but, through ignorance, I fall pretty well into that category and I certainly can't follow the technicalities of this thread. Could you possibly point me at such a primer (for numpties) that will help me understand why 'finer granularity' is not necessarily better and is not the complete story?

TIA
Title: 16 bit, 24 bit and Noise Floor
Post by: Notat on 2012-03-05 04:45:00
I wish I knew of such a reference. I'll take a quick crack at it. If you trust the mathematics, it mostly has to do with the limits of human hearing. I appreciate that trusting the mathematics is easier said than done. Sampling and dither theory are not exactly intuitive.

Improving resolution in the time dimension (higher sample rates) has the effect of allowing digital audio to represent higher frequencies. It is pretty well established that we can't hear above 20 kHz so increasing resolution in the time dimension beyond 48 kHz sample rate is only of use to dogs.

Improving resolution in the amplitude direction (higher bit depth) has the effect of reducing the amount of noise introduced by the process of converting from analog to digital or vise-versa. In most cases, the microphones and other electronics are generating their own noise and so bit resolution is not the weakest link in the recording or playback chain. A 24-bit recording has a 144 dB range. Our range of hearing intensity from quietest detectable sound to pain is 130 dB.
Title: 16 bit, 24 bit and Noise Floor
Post by: saratoga on 2012-03-05 06:53:44
Forgive me but, through ignorance, I fall pretty well into that category and I certainly can't follow the technicalities of this thread. Could you possibly point me at such a primer (for numpties) that will help me understand why 'finer granularity' is not necessarily better and is not the complete story?


For every bit you add, theres 2x the levels to chose from, and thus on average 1/2 the error for each sample.  Intuitive right?  Put your levels 2x as close together and you'll be wrong by half as much on average.  Simple.

Well if you know the noise on your recording is at least 60dB below peak at every frequency, all you have to do is make sure that the average error is a little bit below 60dB.  Then the error will be smaller then the noise.  At that point making the error even smaller doesn't help very much.  So you can keep adding more bits, but all you're doing is making something too weak to be heard weaker. 






Title: 16 bit, 24 bit and Noise Floor
Post by: Arnold B. Krueger on 2012-03-05 12:27:44
Forgive me but, through ignorance, I fall pretty well into that category and I certainly can't follow the technicalities of this thread. Could you possibly point me at such a primer (for numpties) that will help me understand why 'finer granularity' is not necessarily better and is not the complete story?


This might help:

Good Resolutions (HFN) (http://www.audiomisc.co.uk/HFN/goodresolutions/page1.html)
Title: 16 bit, 24 bit and Noise Floor
Post by: bandpass on 2012-03-05 14:16:47
Here's a wood-working analogy

Suppose that you have a staircase with wooden banister spindles; these things:
(http://s2.hubimg.com/u/231489_f260.jpg)
Suppose that one is broken and that you are going to make a copy of an intact one as a replacement.

For measuring the original, you have at your disposal: plastic vernier callipers with 0.01" resolution, and a steel micrometer with 0.00005" resolution (a factor of 200 'better').

Using the micrometer does not result in a replacement spindle that anyone could distinguish from one made using the callipers.  Though it has ~8 bits more resolution, (for this application) the extra bits are in the noise.  The same is true of 24-bit audio.
Title: 16 bit, 24 bit and Noise Floor
Post by: icstm on 2012-03-05 15:51:57
As far as I know, you can't hear tones more than about 6dB below the noise floor.
so what is the noise floor then, if i can hear below it? I am confused 


A noise floor can be considered part of the signal in many situations - and be worth capturing as is.


I do not understand

A noise floor can be considered part of the signal in many situations - and be worth capturing as is.

To add another bad analogy: I always prefer MPEG2 BluRays, because they preserve film grain and noise much more conservatively than certain H.264 encoders, which add some form of softening even at very high bitrates. One could claim that this is a feature, because the latter encoder prefers allocating bandwidth to even more detailed signal than noise. But sometimes I want the noise.

I know of this preference within the cinema crowd.

I have never heard of such a thing within audio. I understand that many object to the artifacts associated with noise reduction, but preserving the crackles of vinyl or air-conditions of some recording venue does not seem to appeal to most music-lovers.

-h


Isn't that because the orginial format in Cinema includes the grainy picture. In audio, the orginal is a live or studio recording with no noise. In cinema, the orginal was NOT the live performance on stage.

A 24-bit recording has a 144 dB range. Our range of hearing intensity from quietest detectable sound to pain is 130 dB.
So 24bit recording COULD add something to the the listening experience?
(though I assume that this is theoretical, as you and I and others have said the rest of your equipment is unlikely to provide such a range or (without wanting to confuse matters) is unlikely to do so accurately)
Title: 16 bit, 24 bit and Noise Floor
Post by: Arnold B. Krueger on 2012-03-05 16:34:09
As far as I know, you can't hear tones more than about 6dB below the noise floor.
so what is the noise floor then, if i can hear below it? I am confused 


The noise floor that we speak about technically is something that we measure. For example a piece of audio gear might have a noise floor of 1 millivolt or 1/1000 of a volt. The loudest signal that can be handled at the same time might have an amplitude of 10 volts.


The interesting thing is that we can hear pure tones at midrange frequencies whose amplitude is 1/10 of a millivolt.


Quote
Quote from:  link=msg=788245 date=0
A noise floor can be considered part of the signal in many situations - and be worth capturing as is.


I do not understand


Imagine a microphone sits in a room that has both music being played and droning on in the background is noise from the HVAC (heating) system. The following signals are coming out of the microphone:

(1) The music.
(2) The noise from the HVAC
(3) The noise from the microphone's electronics

In that order of sizes.  The signal from the microphone appears to be music plus a noise floor, but that noise floor has contributions from the HVAC and the microphone itself. The noise from the HVAC is a noise floor that is superimposed on the noise floor from the microphone.



Title: 16 bit, 24 bit and Noise Floor
Post by: Arnold B. Krueger on 2012-03-05 16:45:26
A 24-bit recording has a 144 dB range. Our range of hearing intensity from quietest detectable sound to pain is 130 dB.


So 24bit recording COULD add something to the the listening experience?


Let's put it this way. The noise floor of your living room is maybe 40 dB SPL.  In order to have a sound that is 144 dB above the noise floor in your living room (1OW 184 dB SPL), we'd probably have to set off some explosives and not just a little.

The loudest car stereo at national "Crank It Up" contests might be able to get that loud but trust me nobody sits inside of them during the contest. The windows have to be specially made and held in place!

Listening to sounds this loud would seriously injure you or even flat out kill you on the spot.

The flight deck of an aircraft carrier launching planes (where everybody wears full-body protective gear) is maybe 130-140 dB SPL.

Back in the real world the loudest sound in the seats at the symphony might be 110 dB when they are playing the loudest passage of the loudest musical work they play. The noise floor of the room with musicians and audience, but no music playing might be 35 dB.  That gives you a dynamic range of 75 dB.
Title: 16 bit, 24 bit and Noise Floor
Post by: googlebot on 2012-03-05 19:52:34
Let's put it this way. The noise floor of your living room is maybe 40 dB SPL.  In order to have a sound that is 144 dB above the noise floor in your living room (1OW 184 dB SPL), we'd probably have to set off some explosives and not just a little.


That's maybe a little too simplified. I think I could differentiate a playback of brown from white noise at 40 dB SPL played back in an environment with room noise at 40 dB SPL.

IMHO icstm's questions are valid and have not been answered thoroughly. If both noise and signals below noise can be heard, how is the poster supposed to understand, that it would be sufficient to capture just the range above it?

Music isn't all rock far above noise. In classic and Jazz it alters my experience significantly, if at all, and what kind of noise is present, when no instrument plays. Having near digital silence is not the best matching choice for every performance. That's an artistic choice and no technical one. Although 96 dB should be enough for everybody, you cannot argue by math and our threshold of hearing that 96 dB is provably enough in any case. You just need to live in a quiet countryside and crank up your volume to 126+ dB SPL max.
Title: 16 bit, 24 bit and Noise Floor
Post by: Wombat on 2012-03-05 20:15:54
What about a simple experiment? Create a silent 16bit file and turn up the volume of your system until you can barely hear the noisefloor. Now press play with an average loud song.
Title: 16 bit, 24 bit and Noise Floor
Post by: googlebot on 2012-03-05 20:18:52
What about a simple experiment? Create a silent 16bit file and turn up the volume of your system until you can barely hear the noisefloor. Now press play with an average loud song.


That advice is evil!  Your DAC may output nothing in this case. Better play back a file with the LSB flipping randomly.
Title: 16 bit, 24 bit and Noise Floor
Post by: Wombat on 2012-03-05 20:22:25
That advice is evil!  Your DAC may output nothing in this case. Better play back a file with the LSB flipping randomly.

Or you can combine that with a noise-shaping dithernoise test. Choose a curve your hearing won´t catch the noise that easy and will allow even a louder playback.
Title: 16 bit, 24 bit and Noise Floor
Post by: icstm on 2012-03-06 09:53:00
A 24-bit recording has a 144 dB range. Our range of hearing intensity from quietest detectable sound to pain is 130 dB.


So 24bit recording COULD add something to the the listening experience?


Let's put it this way. The noise floor of your living room is maybe 40 dB SPL.  In order to have a sound that is 144 dB above the noise floor in your living room (1OW 184 dB SPL), we'd probably have to set off some explosives and not just a little.
OK, but if I look back at your other post, you are saying I can hear below the floor. So for 144dB of dynamic range I do not need 184dB SPL (i think). 
Title: 16 bit, 24 bit and Noise Floor
Post by: icstm on 2012-03-06 10:05:13
Here's a wood-working analogy

Suppose that you have a staircase with wooden banister spindles; these things:
(http://s2.hubimg.com/u/231489_f260.jpg)
Suppose that one is broken and that you are going to make a copy of an intact one as a replacement.

For measuring the original, you have at your disposal: plastic vernier callipers with 0.01" resolution, and a steel micrometer with 0.00005" resolution (a factor of 200 'better').

Using the micrometer does not result in a replacement spindle that anyone could distinguish from one made using the callipers.  Though it has ~8 bits more resolution, (for this application) the extra bits are in the noise.  The same is true of 24-bit audio.

Having slept on this I am not sure this is quite right.

The extra bits are not NOISE as you say.

You have a 30cm banister. To fit this, you have a TOLERANCE. If this tolerance can be RESOLVED with the vernier callipers then the micrometer is not needed. To all intents and purposes the callipers provided an ACCURATE  measurement. The micrometer would have also provided an accurate measurement, but with great PRECISION. This precision (afaik) is not in the noise, as it was and can be resolved, however:
1) In normal use, you cannot make out this difference (ie mk 1 eyeball when comparing the other spindles)
2) In normal use any difference can be catered for in the fixings within the arm rail (ie where the tolerance lies)

As I understand it for it to have been noise, I would be saying that these PRECISE measurements were in fact not ACCURATE due to some other reason (namely noise) and thus could well vary should the measurement be conducted again.

Does that make sense?
Title: 16 bit, 24 bit and Noise Floor
Post by: spindle on 2012-03-06 12:36:28
This might help:

Good Resolutions (HFN) (http://www.audiomisc.co.uk/HFN/goodresolutions/page1.html)


Thank you (and indeed to the others who have tried to educate me). The link and the other pages on that site are enough for me to chew on for now.

However, I don't know whether to be relieved or perplexed that even my 'numpty' question is the cause of continuing discussion. I'll continue to lurk and learn.
Title: 16 bit, 24 bit and Noise Floor
Post by: Arnold B. Krueger on 2012-03-06 13:12:35
A 24-bit recording has a 144 dB range. Our range of hearing intensity from quietest detectable sound to pain is 130 dB.


So 24bit recording COULD add something to the the listening experience?


Let's put it this way. The noise floor of your living room is maybe 40 dB SPL.  In order to have a sound that is 144 dB above the noise floor in your living room (1OW 184 dB SPL), we'd probably have to set off some explosives and not just a little.
OK, but if I look back at your other post, you are saying I can hear below the floor. So for 144dB of dynamic range I do not need 184dB SPL (i think). 


The numbers that one sees for hearing below the noise floor vary, which only makes sense because there are a number of dependencies. However, I can't think of a real world situation where you would hear music 20 dB below the noise floor. Even with pure tones, 6 dB below the noise floor can be hard to hear. It surely wouldn't be your preferred music listening situation! 

In reality, real world listening to music below the noise floor relates to things like reverb tails which are the echoes that follow music that ends abruptly. Another situation is fade-outs at the end of songs that are artificially put there during the production process.  As a practical matter, it is far more important that reverb tails and fade outs end without creating bad noises, than that they seem to go on forever. With even reasonably high levels of background noise, these effects can be very musically satisfying. 

In most cases in real world recordings, if you crank your system up to hear fade outs and reverb tails better, you'll hear them disappear into a mass of rumbling HVAC noise.
Title: 16 bit, 24 bit and Noise Floor
Post by: 2Bdecided on 2012-03-07 10:41:29
Forgive me but, through ignorance, I fall pretty well into that category and I certainly can't follow the technicalities of this thread. Could you possibly point me at such a primer (for numpties) that will help me understand why 'finer granularity' is not necessarily better and is not the complete story?
Imagine you're measuring it with a ruler.

"Finer Granularity" means more (closer) tick marks on your ruler. There's always some discrepancy between the distance you're measuring, and the nearest mark to it on your ruler.

The difference between the real value and the nearest tick mark is an error.

You know the maximum error in any case: it's half the distance between the rick marks - because that's the greatest amount you'll have to round up (or down) to hit a tick mark.


In audio, when the audio signal you've stored isn't quite what you started out with, the difference is noise*. So now we know how much noise has been added due to "measuring" - a noise level equal to half the distance between the "tick marks". Calculate this, and that noise turns out to be nearly 100dB below the maximum signal level.

Given the sound levels in real life, and the ones our ears can cope with, that's enough. Pushing the measurement noise further down by increasing the number of bits does give close tick marks = "finer granularity", but the improvement is inaudible. It's also tricky to go much beyond 16-bits because basic electronics have ~20-bits of noise at best - and everything, from microphone through mixing desk to amplifier needs electronics - not to mention the digital/analogue converters.


* - it could be noise or distortion, but we'll use dither to ensure it's never distortion and always noise. We'll probably also use noise shaping to push the noise level further down in the most audible range, at the expense of slightly more noise at the frequency extremes where our ears aren't so sensitive.

Cheers,
David.
Title: 16 bit, 24 bit and Noise Floor
Post by: 2Bdecided on 2012-03-07 10:55:43
OK, but if I look back at your other post, you are saying I can hear below the floor.
You (and most people!) are confusing measurements of total signal power with measurements of the amount of signal at a given frequency.

A pure 1kHz tone (a beep) only contains one frequency, so the measure of total signal power, and the amount of signal at that one frequency, will be exactly the same for this signal.

Whereas noise can contain a spread of frequencies, with the power varying by frequency. If you have a noise signal where the power is spread evenly across all useful audible frequencies (say, 20Hz-20kHz), then for a particular total power, it'll have a lot less signal at any one frequency.

So a -100dB FS pure tone sticks up above -100dB FS noise - because at the specific frequency of the pure tone, the level of the noise is 10s of dB lower than the level of the tone.

Whether you can see/hear a tone in some noise depends on the level of the tone, the level of the noise, and how you split the signal out into difference frequency bands to try to "see"/hear details like this.

The ear has the equivalent of about 24 logarithmically spaced filters (compare this with the typical 5 filters on a classic graphic equaliser). This means you can hear a tone that measures 20-30dB lower than the total power of the noise around it - because most of that noise falls into different filter bands and can be ignored by your ears.

Cheers,
David.
Title: 16 bit, 24 bit and Noise Floor
Post by: icstm on 2012-03-07 12:46:28
I understand that. So I guess the problem is with the defn. of a noise floor, as it only really sence in a given range and 20-20k is maybe too big a range to define a floor?? just a thought.
Title: 16 bit, 24 bit and Noise Floor
Post by: KMD on 2012-03-07 12:53:20
Noise "floor" is an bad phrase as signals don't go over or under the noise thay are added with it. It is best avoided and use the term noise, and the context, instead.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-08 12:46:52
I'm suggesting that with a recording environment that yields a noise floor of -60dbfs it makes no difference at whether you capture in 16 bit or 24 bit as you effectively only have 10 bits to work with and so any theoretical benefits of using 24 bit are lost
i am posting a reply since I found this post helpful
If my only purpose is to record spoken word and the dynamic range of my speech comes around 40dB, All i need is probably 12 bits
giving a 10dB headroom and minimum 10dB noise floor clearance with my signal peaking at -10dB dbfs. I believe this is similar to the situation what is stated here in the original post since I do it in a room where the noise is peaking at around -60dB. For me I believe there wont be any sort of advantage in using a 24 bit recording interface.
Also from what I understand the 24 bit recording and 16 bit recording uses same number of bits to record a 40dB dynamic range speech like mine, so there is no high definition there of any sort contrary to the popular perception of absolute beginners  like me.

Correct me if I am wrong please.

MOD Edit: Add proper quote tags
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: magicgoose on 2020-08-08 15:28:21
16 bits will be probably enough, especially if you don't want to do any further processing, but the reasoning isn't completely correct.

Quantization always adds some noise (unless the source is already quantized to the same or lower bit depth), it's just a question at which point it becomes insignificant for your use case.

Quote
I do it in a room where the noise is peaking at around -60dB

Comparing noises by peak levels won't give a good guidance, the interesting metric here is not peak levels, but the perceived loudness.
Quantization noise has constant loudness, so it'll likely have a smaller difference between peak level vs average perceived loudness. The frequency characteristics of noises are also important to determine if one noise will be completely masked by other noise.

Quote
Also from what I understand the 24 bit recording and 16 bit recording uses same number of bits to record a 40dB dynamic range speech like mine

PCM coding doesn't work like this. All bits are used all the time. Perhaps you can elaborate, what do you mean by "use same number of bits"?
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: ajp9 on 2020-08-08 17:59:03
If we're talking noise floor from analog equipment, that is quite different than noise from lower-resolution quantization. Analog noise doesn't clip information digitally, and thus can be removed with good noise cancellation.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: magicgoose on 2020-08-09 10:43:10
@ajp9 What does it mean - "clip information digitally"?
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: dc2bluelight on 2020-08-09 10:45:52
If we're talking noise floor from analog equipment, that is quite different than noise from lower-resolution quantization. Analog noise doesn't clip information digitally, and thus can be removed with good noise cancellation.
Not if dithering is involved, and it always is.  Once an ADC is dithered the resulting quantization noise is very similar to analog noise, differing in spectrum only.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-09 16:57:09

Quantization always adds some noise (unless the source is already quantized to the same or lower bit depth), it's just a question at which point it becomes insignificant for your use case.



I use a 16-bit interface for this recording. I do the post processing in Audacity in 32 bit float format and I export it as mp3.

I use a Noise Gate and remove all the part below -50dbFS threshold.
So all parts of the recording below this level I am considering as undesirable. This is what I wrongly attributed to needing lesser number of bits.
I think I  can rephrase it as the part of the signal those extra bits in 24-bit capture, are useless to me since I think its just my room noise.

I am not sure if I should be worried about the quantization noise in this situation?

Quote
PCM coding doesn't work like this. All bits are used all the time. Perhaps you can elaborate, what do you mean by "use same number of bits"?

Yes I got your point, it was my misunderstanding.Sorry.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: magicgoose on 2020-08-10 08:53:19
> I am not sure if I should be worried about the quantization noise in this situation?

To say for sure, you'd need to compare and decide. But if you simply remove the quietest parts, I think it's most likely not worth buying another soundcard for this.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-10 13:54:47
> I am not sure if I should be worried about the quantization noise in this situation?

To say for sure, you'd need to compare and decide. But if you simply remove the quietest parts, I think it's most likely not worth buying another soundcard for this.

Yes thanks, I tried out most of the tests mentioned in this post. I understood that its not exactly intuitive and its not good to make assumptions and  I think i have an idea of what the difference might be.
But like you said to know it for sure the difference, I will need a 24 bit interface and test 16-bit mode and 24 bit mode and compare the output.
I tested the 8-bit and 16-bit recordings for spoken words and there is a perceivable difference in terms of noise. But I understand that the audible  difference won't be so stark between 16 -bit and 24- bit.
So my conclusion is I don't need to buy another sound card only for the sake of a 24 bit depth, for my purpose.
This has been very helpful. thanks so much





Title: Re: 16 bit, 24 bit and Noise Floor
Post by: DVDdoug on 2020-08-10 15:20:22
Quote
I think i have an idea of what the difference might be.
But like you said to know it for sure the difference, I will need a 24 bit interface and test 16-bit mode and 24 bit mode and compare the output.
I tested the 8-bit and 16-bit recordings for spoken words and there is a perceivable difference in terms of noise. But I understand that the audible  difference won't be so stark between 16 -bit and 24- bit.
So my conclusion is I don't need to buy another sound card only for the sake of a 24 bit depth, for my purpose.
This has been very helpful. thanks so much
Right!   At 16-bits you can't hear quantization noise under any normal-reasonable conditions.   

As you may have noticed with your 8-bit experiment, quantization noise is different from regular analog noise.     Analog noise exists constantly in the background.   

Quantization noise rides on top of the signal and when there is digital silence there is no quantization noise (something like a noise gate).    But like analog noise, it's most noticeable with a quiet signal and may be masked (drowned-out) with a higher signal.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-11 09:53:48

As you may have noticed with your 8-bit experiment, quantization noise is different from regular analog noise.     Analog noise exists constantly in the background. 
Quantization noise rides on top of the signal and when there is digital silence there is no quantization noise (something like a noise gate).    But like analog noise, it's most noticeable with a quiet signal and may be masked (drowned-out) with a higher signal.
 

This is interesting. Yes indeed I noticed that I cannot remove the noise in 8 bit using noise gate since it seem to be riding on top of my actual desired audio where as the background noise was completely removed adjusting a noise gate threshold.This also removed the quantization noise in the silent part since the whole signal during silent part were removed I presume. But whenever the words were spoken I could hear the noise. With dithering applied during export also this noise was audible.

For original 16 bit recording this noise is not much distinguishable.With noise gate it sounds very clean.  I presume 24 bit does improve on the 'noise riding above the signals', but the difference will be hard to perceive under normal conditions.

This interests me in another question if I may. Where does the SNR of the microphone I use, fit in in this whole scenario?
I use a Shure SM48. The audio interface I use has a SNR  specification of 85 dB. The room I record in the noise peaks around
-65dbFS.( I know this is not apt,but I am yet to figure out how to do a spectral analysis of the noise.)
thanks






Title: Re: 16 bit, 24 bit and Noise Floor
Post by: ajp9 on 2020-08-11 22:06:49
@ajp9 What does it mean - "clip information digitally"?
When binary bits are removed (clipped off) to fit a smaller significant digit resolution. Information is necessarily lost that can't be recovered.

The clear advantage of digital encoding is its ability to reproduce exact copies, but any digital conversion, not just A/D, will lose information to some degree, so accuracy needs to be high to lose the least. Depth is unavoidably part of that accuracy.

Dithering when decimating helps encode signal energy because it takes more than one sample to make a wave, but it's a limited fallback, and not always guaranteed to be accurate. Only randomized dithering works well with audio signals because ordered dithering or delta-sigma modulation can combine with similar signals to produce interference patterns—something where a lowpass would help but a lowpass could interfere with the purpose of encoding for a high bandwidth.

Once an ADC is dithered the resulting quantization noise is very similar to analog noise, differing in spectrum only.
Similar, but not the same. Dithering—especially noise-shaped dithering—can help improve clarity or SNR in more audible parts of the spectrum. But digital decimation always removes information in ways that analog cannot.

Think of how digital information is recorded, overwriting some other bits on a storage medium; that's as good as it gets; there's no more information for that recording itself; you can't pull other layers the way you can restore from tape. That isn't to say analog media can always be recovered, but with more advanced noise prediction/cancellation you could recover information that was previously thought unrecoverable. Digital may add additional security that the FBI can't restore what you decimated or overwrote, but neither can you if you don't have a more accurate copy.

What I'm saying may sound trivial, but the most important content ought to be recorded at a higher resolution if you can afford it to preserve history.

At 16-bits you can't hear quantization noise under any normal-reasonable conditions.
16 bits is generally adequate for masters because there's a point where the brain stops caring beyond a certain signal threshold (apart from overall amplitude). For sight, it was measured that humans can't see more than just over 600 levels per stepping. Even the best ears on HA can't ABX LossyWAV from the original, and that's limiting significant digit precision to 8 bits on a floating scale (upper/lower bound of 16 bits).
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: DVDdoug on 2020-08-11 22:37:49
Quote
I use a Shure SM48. The audio interface I use has a SNR  specification of 85 dB.
That's tricky because you have to make assumptions about the signal.   Speaking/singing louder, or getting closer to the mic, or using a more sensitive mic will increase the signal which improves the signal-to-noise ratio.   A directional mic (such as the SM48) also helps the SNR because the signal comes from one direction and the room noise comes from all directions.    (A more sensitive mic will help with the electrical preamp noise but of course it doesn't improve the acoustic SNR.)    A weaker signal will make the SNR worse.

If you have a USB-powered interface, sometimes noise gets-in through the USB power and the interface might not actually meet it's published specs.

Quote
The room I record in the noise peaks around -65dbFS.
If you are getting a "good signal" that beats the audiobook spec of -60dB and that's very good for home recording!   But if you have to amplify, the noise will be amplified too.

Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-12 17:19:39
  A directional mic (such as the SM48) also helps the SNR because the signal comes from one direction and the room noise comes from all directions.    (A more sensitive mic will help with the electrical preamp noise but of course it doesn't improve the acoustic SNR.)    A weaker signal will make the SNR worse.
Yes , the dynamic mic of around 150 ohm has self noise much lesser than the preamp noise created while amplifying  its signal.

Discarding the background noise amplification and taking into consideration only the preamp noise ,hypothetically speaking if my 16 bit interface noise floor is 85db(got this value from an online review of the model)
and the mic preamp inside the same interface is producing a noise around 16db to drive the dynamic mic, will my effective noisefloor of the interface be at 69db now instead of the 85db?

Quote
If you are getting a "good signal" that beats the audiobook spec of -60dB and that's very good for home recording!   But if you have to amplify, the noise will be amplified too.

Yes, noted. Tell me if I am wrong , If a 85db SPL system plays the dialogues mixed at -10dbFS the loudness will will 75db SPL
and if I have a SNR of 60db, the noise will be played at a loudness of 15db SPL which will go below the noise floor of even a quiet room and will be inaudible.

thanks


Title: Re: 16 bit, 24 bit and Noise Floor
Post by: DVDdoug on 2020-08-12 17:41:03
Quote
Yes, noted. Tell me if I am wrong , If a 85db SPL system plays the dialogues mixed at -10dbFS the loudness will will 75db SPL
dB SPL and dB FS are almost never calibrated so let's forget about "FS" for the moment.   But they are correlated, so if you are playing a file at 85dB SPL and you reduce the digital level by -10dB, the analog also gets reduced by 10dB and you'll get 75dB SPL, and any noise in recording will also be reduced by 10dB.    (There still may be some more-constant noise from the analog side of the playback system, but that's another topic.)

Quote
and if I have a SNR of 60db, the noise will be played at a loudness of 15db SPL which will go below the noise floor of even a quiet room and will be inaudible.
Basically, yes.    If the SNR is 60dB and the SPL is 85dB you won't hear the 15dB SPL noise.

Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-12 18:58:51
Thanks
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: OEFreelancer on 2020-08-12 21:22:54
This gives a totally different picture about the noise floor.
http://www.avisoft.com/tutorials/understanding-microphone-and-recorder-preamplifier-noise-specifications

"The specifications provided by the recorder manufacturers are often difficult to interpret. Due to marketing considerations, many manufacturers tend to publish large dynamic range numbers (e.g. > 100 dB) that are only valid for high-level signals.

For instance, the microphone input of the Marantz PMD671 .(SNR IEC-A weighted : LINE 92db ,MIC 65db) is honestly specified with a “poor” signal-to-noise ratio of 65dB(A), while the M-Audio MicroTrack claims a dynamic range of 100dB(A)

The overall inherent noise floor of the entire recording system results from the combination of the inherent noise floors of both the microphone and the recorder. It is important to note that both noise components add geometrically (RMS)

 it turns out that the optionally available 24 bit recording format cannot provide any significant improvement of the noise performance."
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: DVDdoug on 2020-08-12 22:40:12
Quote
The specifications provided by the recorder manufacturers are often difficult to interpret. Due to marketing considerations, many manufacturers tend to publish large dynamic range numbers (e.g. > 100 dB) that are only valid for high-level signals.
Re-phrasing an old saying, "There lies, damn lies, and specifications".  :D

Quote
The overall inherent noise floor of the entire recording system results from the combination of the inherent noise floors of both the microphone and the recorder. It is important to note that both noise components add geometrically (RMS)
Plus the acoustic noise.  ;)

Quote
it turns out that the optionally available 24 bit recording format cannot provide any significant improvement of the noise performance.
Pros often record at around -18dB (for some strange "traditional" reasons or maybe to allow extra headroom).   At -18dB you are "loosing" 3-bits so at those levels maybe you can argue for 24-bits.   But, when you mix you are adding bits (mixing is literally done by summation) so if you are multitracking it's less important.

From what I've read most 24-bit ADCs & DACs are only accurate to about 20 bits, but that shouldn't affect the noise floor or anything we're talking about here.
Title: Re: 16 bit, 24 bit and Noise Floor
Post by: dc2bluelight on 2020-08-13 09:28:03
Pros often record at around -18dB (for some strange "traditional" reasons or maybe to allow extra headroom).  
The -18dB figure comes from the need to calibrate a VU meter (which is a special case, averaging meter, 300ms integration time) to a true peak meter (instantaneous) found on a digital recorder, calibrated in dBFS.  When you put up a tone at 0VU, you can calibrate that tone anywhere on the dBFS meter.  -18dBFS is a compromise, allowing good headroom for true peaks that won't be metered correctly but the slow VU meter and provide headroom for unexpected loud peaks.  A rule-of-thumb true-peak to VU differential is 8 to 10dB.  It's not unusual to "peg" a VU meter during a very loud passage, so between 15 and 20dB of calibration offset is what's required to avoid clipping at 0dBFS.

In the early days of the CD, replication labs specified that there could be no peaks at 0dB(FS) (the FS term was a late addition).  And since adjusting a digital signal in post was a bit difficult initially, we had to be careful, especially if not using any dynamics processing.  Hence, there were several 0VU calibrations: -15dBFS, -18dBFS and -20dBFS. 
At -18dB you are "loosing" 3-bits so at those levels maybe you can argue for 24-bits. 
Bits are never lost, they're just not active.  You don''t loose 3 bits, they're there for a purpose, even if it's for that one peak in the entire record that hits 0dBFS. 
But, when you mix you are adding bits (mixing is literally done by summation) so if you are multitracking it's less important.
You aren't adding bits, the bits are always there.  When you mix the final result still has to fit below 0dBFS on the highest peak.  Mixing is a sum, but that sum is scaled by the mixer, or we'd have to establish a new reference 3dB down every time we added a channel.  Even if bits are inactive for most of the recording, an even if the recording never exercises all bits at once, the inactive ones serve a purpose: headroom, a place-holder in case you need it.   Analog is no different, BTW, we have all the same issues and surprisingly, at the same levels.   Only the metering references change.  We generally don't worry about frequent peaks falling a few dB short of clipping our summing amp, do we?  Are we losing dynamic range doing that because we aren't using up all the analog headroom?  No, because the DR is limited already elsewhere in the chain, at the low level end by microphone self noise, preamp noise, and the big one, room noise.  Hopefully the summing amp is one of the quieter amps in the system.
From what I've read most 24-bit ADCs & DACs are only accurate to about 20 bits, but that shouldn't affect the noise floor or anything we're talking about here.
Almost.  Most 24 bit ADCs have 20 bit noise performance.  When you try to build an ADC that has its own internal noise floor somewhere around the 24 bit level, you run headlong into problems with thermal noise.  Heck, the 20kHz Johnson noise of a 2000 ohm resistor is only -122dBV, so you have to jump through some pretty big hoops to get real 24 bit performance.  I know of only one company that marketed such an ADC, and they cascaded more than on ADC to get it to work.  Everything else has about 20 bit noise floor.  That's not the same as "accuracy", of course, which isn't measured in bit count.