Skip to main content

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Distinctive benefits of 24bit recording? (Read 21247 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Distinctive benefits of 24bit recording?

Reply #25
Bit depth directly affects you headroom. So if you want a deeper dynamics of recordings (e. g. want to record explosion without limiting peaks with all beauties) with any most botique and vintage mics you'll got better details capturing in 24 than in 16 bits.

Do you have any data to back this up, or should we just take your word for it?

Distinctive benefits of 24bit recording?

Reply #26
Of course 24 bits requires more storage space. However, the particular topic was whether or not samples sold as 24 bit were likely to be a fraud, recorded at 16 and converted. Of course there is no shortage of people who commit fraud, but for this purpose there would be no extra cost to the producer/seller to offer the real thing. Surely the fact that it doesn't make an audible difference has been done to death and need not be repeated when it isn't part of the question.

The fact that some program may automatically convert to and work in floating point is irrelevant. The post said ALL modern DAWS, and that just isn't so. Some give one a choice.

And, it would seem rather inconsistent for a program to allow 16 bit recording without then being able to open in 16 bits. If the recording were automatically in float, where is that 32 channels for an hour and a half? If you don't have the resources to capture it, how can you work on it?

Distinctive benefits of 24bit recording?

Reply #27
Recording everything at 24 bits is like saying "I have lots of money so I will always pay twice list price for everything that I buy". If it floats your boat, by all means do it!

So the music industry has money to burn? 


You;'ve got talent that you're paying how many millions of dollars per year, and they'll make you a millionaire several times over if you can just coax them to make another CD...

Quote
Thank god those engineers don´t build bridges.


Different strokes for diffferent folks.

Distinctive benefits of 24bit recording?

Reply #28
Sure there is a good reason to record at 16 bits: 32 channels for an hour and a half. 

A one hour stereo recording at 16/44 fits on a CD. A 24 bit recording does not. Not even on 80 minute CD-Rs.

For many practical scenarios, the storage cost at recording is unlimited and free. For distribution you would do whatever to fit within the distribution limits but that is another discussion.
Quote
If there was some real benefit to recording at 24 bits, then the argument that it doesn't cost that much to do it has some merit. But there is no audible benefit.

Recording everything at 24 bits is like saying "I have lots of money so I will always pay twice list price for everything that I buy". If it floats your boat, by all means do it!

I dont follow you there. Is it not true that even cheap "24-bit" A/D-converters have a performance surpassing that of theoretically perfect 16-bit ones? This "headroom" could be used for:
1. Sloppy input gain settings (or reduced likelihood of clipping)
2. Creative non-linear audio effects

While still having an end-to-end resolution of >=16 bits (if microphones, recording venue etc is really, really good). My "prosumer" soundcard measure better than a 16-bit soundcard could ever do in RMAA loopback tests. The last disks I bought were 2 TB. I never think about filesize when recording audio...

-k

Distinctive benefits of 24bit recording?

Reply #29
Sure there is a good reason to record at 16 bits: 32 channels for an hour and a half. 

A one hour stereo recording at 16/44 fits on a CD. A 24 bit recording does not. Not even on 80 minute CD-Rs.


For many practical scenarios, the storage cost at recording is unlimited and free. For distribution you would do whatever to fit within the distribution limits but that is another discussion.


While the storage costs are minimal, the real world costs are often not free. For example, I started out making 24 track recordings of a live event that ran about 75 minutes. Due to the software I was using, 24 bit data required the use of 32 bit words, so the cost of the doubled resolution for data storage was 2X, which according to your statement should still be free, right?  What ate me up was the cost in terms of my time of doing things like backing the data up to portable media and carrying it off site for processing, and then restoring it and editing it. The extra data words also nearly doubled the time it took to render my 24 tracks down into a stereo mix-down.  I'd pay for it all with a smile if there were an audible benefit, but seeing none...

Quote
Quote
If there was some real benefit to recording at 24 bits, then the argument that it doesn't cost that much to do it has some merit. But there is no audible benefit.

Recording everything at 24 bits is like saying "I have lots of money so I will always pay twice list price for everything that I buy". If it floats your boat, by all means do it!


I don't follow you there. Is it not true that even cheap "24-bit" A/D-converters have a performance surpassing that of theoretically perfect 16-bit ones? This "headroom" could be used for:
1. Sloppy input gain settings (or reduced likelihood of clipping)
2. Creative non-linear audio effects


Normal 16 bit recording offers you at least 10-15 dB of real world headroom.  So here's the deal - if you are sloppy and need > 16 bits, you pay the prices in your time that I mentioned above. If you are careful, all that time differential stays in your life.  It turns out to be adequate motivation for me to be a little careful.

Quote
While still having an end-to-end resolution of >=16 bits (if microphones, recording venue etc is really, really good).


I know of no such equipment or venue. Especially the venue part.  You could have a room that was NC20, but that all goes out the door when you bring in the music ans.

Quote
My "prosumer" soundcard measure better than a 16-bit soundcard could ever do in RMAA loopback tests. The last disks I bought were 2 TB. I never think about filesize when recording audio...


Like I said, when you actually start doing this sort of thing for a day or two in a row, or do it for several hours once a week, the priorities change from a discussion of hardware capabilities to a discussion of your time.




Distinctive benefits of 24bit recording?

Reply #30
While the storage costs are minimal, the real world costs are often not free. For example, I started out making 24 track recordings of a live event that ran about 75 minutes. Due to the software I was using, 24 bit data required the use of 32 bit words, so the cost of the doubled resolution for data storage was 2X, which according to your statement should still be free, right?  What ate me up was the cost in terms of my time of doing things like backing the data up to portable media and carrying it off site for processing, and then restoring it and editing it. The extra data words also nearly doubled the time it took to render my 24 tracks down into a stereo mix-down.  I'd pay for it all with a smile if there were an audible benefit, but seeing none...

24 tracks of live recording is probably a corner case. If you are in a studio, having a few disk units costs little time or lugging around. Most other people doing on-site live recording probably get by using less than 24 tracks.
Quote
Normal 16 bit recording offers you at least 10-15 dB of real world headroom.  So here's the deal - if you are sloppy and need > 16 bits, you pay the prices in your time that I mentioned above. If you are careful, all that time differential stays in your life.  It turns out to be adequate motivation for me to be a little careful.

My experience with live music is that it is hard to predict levels, you are in a stressed situation, and there are umpteen things that deserve your focus. Not having to worry about A/D input level should make that just a little easier.
Quote
I know of no such equipment or venue. Especially the venue part.  You could have a room that was NC20, but that all goes out the door when you bring in the music ans.

But does the spectral properties of acoustic room noise match those of ADC noise? If not, it is going to be hard to define a hard limit at which point the A-weighted noisefloor of your ADC is guaranteed to be perceptually insignificant compared to the A-weighted acoustic noise-floor in every part of the spectrum.

Quote
Quote
My "prosumer" soundcard measure better than a 16-bit soundcard could ever do in RMAA loopback tests. The last disks I bought were 2 TB. I never think about filesize when recording audio...


Like I said, when you actually start doing this sort of thing for a day or two in a row, or do it for several hours once a week, the priorities change from a discussion of hardware capabilities to a discussion of your time.

You may well have valid reasons for prioritizing as you do. I just dont think that they are necessarily valid for (many) others than yourself.

Why do certain server hardware have error correction on their memory, when bit errors are _really_ uncommon, and it adds to the cost? The reason may be that some applications are very critical, and a moderate cost increase is no obstacle. If I were to record a once-in-a-lifetime concert, and 24 bits gave me even a slightly increased probability of less audible noise, less audible clipping or a more focused recording engineer able to fix issues on-the-fly, it might be worth it.

-k

Distinctive benefits of 24bit recording?

Reply #31
Hi there,
I was just browsing the forum idly and decided to comment on this discussion - even registered just for this! :-)

I just wanted to comment that, though almost everyone responding to the OP has referred to dynamic range as the chief benefit of 24-bit recording, it seems to have gone uncommented (except perhaps by the OP, who talks of "high-resolution") that 24-bit doesn't just determine the upper and lower boundaries of the recorded volume, but also the granularity, that is the number of the steps, in-between.  Hence a 24-bit recording will more realistically reflect not only the difference between loud and soft, but also the subtle changes in volume that occur within that range.  For that reason alone, I suggest that 24-bit recordings are to be recommended over 16-bit recordings.

I'm always confused by the approach to digital recordings - analogue sound has infinite values, and yet people seem determined to produce high-quality recordings by registering as few of those values as possible :-s

Endo.
How many characters am I allowed here anyway?  Not many :-(

Distinctive benefits of 24bit recording?

Reply #32
I'm always confused by the approach to digital recordings - analogue sound has infinite values, and yet people seem determined to produce high-quality recordings by registering as few of those values as possible :-s


Ignoring for a minute the apparent misunderstanding on the relationship between bitdepth, and the "bounds" and "granularity" (hint, they are the same) this is a non-issue.

But lets, for a minute, run with the misunderstanding and play with the numbers:

16 bits = 65536 values, 65536 steps.
I rather suspect you'll never be able to tell the difference between 45 and 45.05 dB, much less between 45 and 45.002 dB.
I rather suspect that's plenty "realistic" for ya.








Can you?



Not to mention the fact that the "real world", the analog world, is granular too.  And it ain't much finer than 16 bits!. 

When there is +- 0.01 random volts of noise on your wire, what is the difference between an analog 1.234v signal and an analog 1.235v signal?  Hmm, we just might just have bounds and granularity in an analog system just the same!
Creature of habit.

Distinctive benefits of 24bit recording?

Reply #33
@Endovidic:

Hi and welcome to HA.

In response to your reply, it is a matter of dynamic range and SNR.  Many of these treasured analog recordings have a SNR and dynamic range less than that which can be delivered by 16-bits.  If you want to deal with them at a higher resolution, knock your socks off, just don't make any claims that they actually sound better on this forum unless you can back them up with objective evidence.
http://www.hydrogenaudio.org/forums/index....showtopic=16295

This is one of the cardinal rules of this forum, specifically #8:
http://www.hydrogenaudio.org/forums/index.php?showtopic=3974

While it might not be the case elsewhere, we have an expectation that you honor the agreement to read and abide by all of the rules that you made when registering.


Distinctive benefits of 24bit recording?

Reply #35
24-bit doesn't just determine the upper and lower boundaries of the recorded volume, but also the granularity, that is the number of the steps, in-between.  Hence a 24-bit recording will more realistically reflect not only the difference between loud and soft, but also the subtle changes in volume that occur within that range.  For that reason alone, I suggest that 24-bit recordings are to be recommended over 16-bit recordings.

The "granularity" of a 16-bit recording manifests as quantization distortion. When proper dither is applied, quantization distortion is converted to quantization noise. It really is noise. It doesn't make things sound "granular" or any other colorful adjective; It just raises the noise floor of the recording. Given this, I hope you can appreciate that bit depth and dynamic range are directly related and there is no separate "granularity" concept.

Welcome to Hydrogen Audio!

Distinctive benefits of 24bit recording?

Reply #36
I just wanted to comment that, though almost everyone responding to the OP has referred to dynamic range as the chief benefit of 24-bit recording, it seems to have gone uncommented (except perhaps by the OP, who talks of "high-resolution") that 24-bit doesn't just determine the upper and lower boundaries of the recorded volume, but also the granularity, that is the number of the steps, in-between.
As Soap said, these are the same thing. A lower granularity means a higher minimal (non-zero) amplitude as a proportion of full-scale, i.e. a lower dynamic range.

Quote
I'm always confused by the approach to digital recordings - analogue sound has infinite values, and yet people seem determined to produce high-quality recordings by registering as few of those values as possible :-s
But with a much greater confidence of reproducibility because digital media are not affected by stochasticity at all points in the signal chain, as is analog! And as has been said, good luck determining a difference between something as run-of-the-mill as 16-bit and its supposedly superior analogue source, even if the inherent vagaries of the former could somehow be eliminated. (You might also want to keep to yourself disingenuous quips such as the assertion “people seem determined to produce high-quality recordings by registering as few of those values as possible”.)

Distinctive benefits of 24bit recording?

Reply #37
Quote
analogue sound has infinite values

Yes, but that does not mean it has infinite information in it. The maximum amount of information that a signal can hold is determined by the amount of noise and the frequency bandwith.

And the output of a da converter is not steped at all, so has infinite values as well.

Distinctive benefits of 24bit recording?

Reply #38
I just wanted to comment that, though almost everyone responding to the OP has referred to dynamic range as the chief benefit of 24-bit recording, it seems to have gone uncommented (except perhaps by the OP, who talks of "high-resolution") that 24-bit doesn't just determine the upper and lower boundaries of the recorded volume, but also the granularity, that is the number of the steps, in-between.


There is an intimiate relationship between effective step size and noise floor. For example, lets say that we have 24 bit data, which implies 16 million steps between 0 and FS.  However, since this is the real world, let us say that there is a noise floor that is 8 steps high. If I add a signal, the signal has the 8 step high noise riding on it. If I try to estimate the value of the signal at any point, there is always an 8 step ambiguity in my estimate because I don't know the exact value of the 8 step high noise signal at any point in time.

Resolution is based on the idea that I know the exact value of the signal at any time, but in reality there is always this 8 step high noise that keeps me from knowing the exact value of the signal. So what is the resolution - 24 bits because my data channel can handle 24 bits, or 21 bits because of this 8 step ambiguity (3 bits) that is always there?

The answer is that the effective resolution is only 21 bits because there are 3 bits of noise which are not part of the original perfect signal.

Thus the effective resolution of a signal is limited by how much noise is in it, which is incidentally part of what is known as Shannon's information theory.

Quote
Hence a 24-bit recording will more realistically reflect not only the difference between loud and soft, but also the subtle changes in volume that occur within that range.  For that reason alone, I suggest that 24-bit recordings are to be recommended over 16-bit recordings.


Simply not true. The noise floor not only relfects the difference between the loud and soft but it also depends on how noise makes all parts of the recording ambigious. Furthermore, the noise in real world recoding venues typically puts the nise floor no more than 80 dB down, which means that 16 bits has at least 10 dB  headroom. More typical numbers put the noise 70 dB or less down. Now there is 20 dB worth of headroom.

Quote
I'm always confused by the approach to digital recordings - analogue sound has infinite values,


That simply isn't true. When you are talking about analog signals, you have to include the ambiguity caused by noise and distortion.  This is like the difference between data and information. If your data is noisy or imprecise, it has a lot less information than its bit patterns might suggest.

Just because you measure a signal and find a million different values doesn't mean that all million values each have a unique meaning. The million values only relate to your measuring equipment, not what you are measuring.



Distinctive benefits of 24bit recording?

Reply #39
Hey,
Sorry for the delay in responding to follow-ups. Unfortunately I don't have time to address everything that's been posted, but I can say this: I don't think rule #8 can apply when dealing with a priori knowledge---only with a posteriori. And I don't recall making any subjective statements.  You are welcome, however, to poke holes in my reasoning and comprehension of digital matters. I'll make sure I fully understand what all the responses add up to before posting further on this thread.

Cheers 
How many characters am I allowed here anyway?  Not many :-(


Distinctive benefits of 24bit recording?

Reply #41
There is absolutely no reason that you cannot perform a double-blind test on a 24-bit recording and a 16-bit version of that same recording.

Hi,
Interesting that you should respond already, I was about to clarify my position relative to your comment:
Quote
just don't make any claims that they actually sound better on this forum unless you can back them up with objective evidence.

I didn't.  What I wrote was:
Quote
a 24-bit recording will more realistically reflect not only the difference between loud and soft, but also the subtle changes in volume that occur within that range. For that reason alone, I suggest that 24-bit recordings are to be recommended over 16-bit recordings.

At no point do I refer to human perception, therefore, I state again, rule #8 does not apply.

Cheers
How many characters am I allowed here anyway?  Not many :-(


Distinctive benefits of 24bit recording?

Reply #43
24-bit doesn't just determine the upper and lower boundaries of the recorded volume, but also the granularity, that is the number of the steps, in-between.  Hence a 24-bit recording will more realistically reflect not only the difference between loud and soft, but also the subtle changes in volume that occur within that range.  For that reason alone, I suggest that 24-bit recordings are to be recommended over 16-bit recordings.

The "granularity" of a 16-bit recording manifests as quantization distortion. When proper dither is applied, quantization distortion is converted to quantization noise. It really is noise. It doesn't make things sound "granular" or any other colorful adjective; It just raises the noise floor of the recording. Given this, I hope you can appreciate that bit depth and dynamic range are directly related and there is no separate "granularity" concept.

Welcome to Hydrogen Audio!

I don't want to deviate too much from topic, but the term "noise" in this context has always confused me. When someone says that "distortion is converted to noise" my mind goes "wait. Distortion *is* noise." (when speaking of audio)  What does this noise sound like? Buzzing? Hissing? Why would you want more of it in the recording? Isn't the noise floor the line between what i would call "the music" and what ever "noise" my equipment is making, as in a cassette tape has a higher noise floor (hissing) than a CD?

Music lover and recovering high end audiophile

Distinctive benefits of 24bit recording?

Reply #44
Imagine a simple analogue sine wave. Sample it at some finite bit depth. Each sample is a measure of the sine wave amplitude at the moment you took the sample. Since a limited number of amplitude measures can be expressed by the bit depth you are using to sample, there is a error in the expressed value of each sample. That error is some plus or minus amount around the "actual" amplitude.

The magnitude of each error is always less than the amplitude difference of the next discrete value (plus or minus) that can be expressed by that bit depth. Since the sine wave is regular, the errors quantities will vary in a regular way around the "actual" values. This is quantization distortion and it has an unmusical sound of its own.

If you first scatter random low level noise around the sine wave, you will hear this, more or less, as faint white noise. This is indeed extra noise but it is benign, like not very loud tape hiss. It does not interfere with the quality of the music greatly. Now each sample value will still be in error by the same magnitude, but because each measure includes both the regular sine wave plus or minus the random noise, the error is randomized and no longer has that unpleasant unmusical quality. Thus the quantization error remains, but the distortion is eliminated. We add the noise to achieve this result because we cannot get rid of the error.

This noise is never very loud, and thus never exceedingly objectionable, although we would like to not have to have it. It is just better than the alternative. But, we can actually not have it, to a large extent, by using proper noise shaping. Most of it ends up at frequencies few of us can hear.

Distinctive benefits of 24bit recording?

Reply #45
I don't want to deviate too much from topic, but the term "noise" in this context has always confused me. When someone says that "distortion is converted to noise" my mind goes "wait. Distortion *is* noise." (when speaking of audio)  What does this noise sound like? Buzzing? Hissing? Why would you want more of it in the recording? Isn't the noise floor the line between what i would call "the music" and what ever "noise" my equipment is making, as in a cassette tape has a higher noise floor (hissing) than a CD?


Generally, when we say distortion, we mean something like "extra stuff that correlates to the actual signal somehow", whereas noise means "extra stuff that isn't correlated to the signal". So distortion is dependent on the signal and noise is random.

By adding dither, you randomize the quantization error/distortion and thus effectively decouple it from the signal. So you end up hearing "clean signal plus a little noise" instead of "distorted signal". And based on a number of factors, the noise will hopefully be less audible and/or objectionable than the distortion (if it was audible in the first place).

Distinctive benefits of 24bit recording?

Reply #46
Thank you Andy, that was a very clear explanation!
The magnitude of each error is always less than the amplitude difference of the next discrete value (plus or minus) that can be expressed by that bit depth. Since the sine wave is regular, the errors quantities will vary in a regular way around the "actual" values. This is quantization distortion and it has an unmusical sound of its own.

Doesn't that mean that higher bit rates would allow for more steps in between the highest and lowest values, and thus have smaller margins of error? More precision? Wouldn't that lead to less of a need for low level noise to be applied? Doesn't that mean that >16-bit is preferable during the engineering even if you can't actually hear it afterwards, or is it only better in theory?

[quote name='drewfx']

Thanks Drew. I get the difference now. Seems obvious when i think about it.  You know, i must have figured that out years ago. I remember when Metallica's Master of Puppies came out and my father complained about it being "noise." I said something like "it's distortion, not noise."  Turns out I was right.

Bearcat
Music lover and recovering high end audiophile

Distinctive benefits of 24bit recording?

Reply #47
24 bit has quantization errors only 1/256 the size for 16 bit. Higher bit depth (especially 32 and 64 bit floating point format) is frequently used for processing because of this (and certain other reasons). Actually hearing the difference would require (at the least) a large number of operations  (the errors are cumulative). If you elect to use a 24 bit delivery format, dither isn't gong to matter. If you use a 16 bit delivery format, dither isn't going to really matter either, but you not very likely to find any professional work that doesn't use it.

Distinctive benefits of 24bit recording?

Reply #48
Thank you. It makes sense.

By the way, i meant "Master of Puppets" not "Master of Puppies"
Music lover and recovering high end audiophile

 

Distinctive benefits of 24bit recording?

Reply #49
Doesn't that mean that higher bit rates would allow for more steps in between the highest and lowest values, and thus have smaller margins of error?


You seem to be confusing sample size in bits with number of samples per second.  Either one can results in higher bit rates.    Putting more bits in each sample allows for more steps between the highest and lowest values and decreases error and increases the potential for precision.

Quote
Wouldn't that lead to less of a need for low level noise to be applied?


Yes, the size of any dither signal that might be applied is less for samples with more bits in them.


Quote
Doesn't that mean that >16-bit is preferable during the engineering even if you can't actually hear it afterwards, or is it only better in theory?


If you can't hear it, then by definition the change is only better in theory.

So to review. hig sample rates and putting more bits into each sample is theoretically better. But, if the sample rate is as high as  44 KHz, and the samples have as many as 16 bits, you are already very, very, deep into diminishing returns, from a sound quality standpoint.