Head-Fi.org › Forums › Equipment Forums › Sound Science › Bit depth
New Posts  All Forums:Forum Nav:

Bit depth

post #1 of 42
Thread Starter 

I know what it is technically.

But I never listened to 32-bit audio, and never had a sound card that can play that kind of audio.

When I got my Xonar ST I was under the impression I was getting the best I could for my headphones, I guess I was wrong?

post #2 of 42
Quote:
Originally Posted by naike View Post

I know what it is technically.

But I never listened to 32-bit audio, and never had a sound card that can play that kind of audio.

When I got my Xonar ST I was under the impression I was getting the best I could for my headphones, I guess I was wrong?


Thanks for sharingbiggrin.gif... For the best sound i'd get a pair of HD650, a WA6 or Burson 160D, and cable should be providedbiggrin.gifHave fun!

 

post #3 of 42
Quote:
Originally Posted by cifani090 View Post




Thanks for sharingbiggrin.gif... For the best sound i'd get a pair of HD650, a WA6 or Burson 160D, and cable should be providedbiggrin.gifHave fun!

 


Wait, what? Have you now decided to automate your useless posts with bots or something?

To the OP, I have great difficulty distinguishing from 16, 24, and 32 bit audio files. Do you even have any music in 32 bit? If not, it really won't matter at all. Almost no music is released in 24 or 32 bit.
post #4 of 42
Quote:
Originally Posted by revolink24 View Post



Quote:
Originally Posted by cifani090 View Post



Thanks for sharingbiggrin.gif... For the best sound i'd get a pair of HD650, a WA6 or Burson 160D, and cable should be providedbiggrin.gifHave fun!

 




Wait, what? Have you now decided to automate your useless posts with bots or something?

To the OP, I have great difficulty distinguishing from 16, 24, and 32 bit audio files. Do you even have any music in 32 bit? If not, it really won't matter at all. Almost no music is released in 24 or 32 bit.


lol, no it wasn't an open ended question, so i just put that and helped him out the best of my ability.

 

post #5 of 42
Thread Starter 

No I can't listen to 32-bit audio, my card only plays 24-bit audio, I was just wondering because I'd consider the Xonar ST a high(er) end card and it doesn't support 32-bit audio.

post #6 of 42
There is a lot of pro gear (much higher end than your Xonar) that still only does 16 bit audio.

Bit depth only has to do with the DAC chip itself, and is not directly related to the quality of the DAC.
post #7 of 42

Technically, 16 bit audio provides a 96 dB dynamic range, meaning the softest details you can perceive are 96 dB softer than the loudest ones.

 

A quiet room has 30 dB of ambient noise, meaning that softest sounds get drowned into the background noise.  30 + 96 =126 dB which is louder than a gunshot 1m away.

 

Since one can't even use the full dynamic range of 16 bit in a normal environment, I don't see the use usefulness of 24 bit.

post #8 of 42
Quote:
Originally Posted by khaos974 View Post

Technically, 16 bit audio provides a 96 dB dynamic range, meaning the softest details you can perceive are 96 dB softer than the loudest ones.

 

A quiet room has 30 dB of ambient noise, meaning that softest sounds get drowned into the background noise.  30 + 96 =126 dB which is louder than a gunshot 1m away.

 

Since one can't even use the full dynamic range of 16 bit in a normal environment, I don't see the use usefulness of 24 bit.


Thanks for crunching the numbers for me, I almost did it myself but then I became lazy. tongue.gif

I agree wholeheartedly.
post #9 of 42

Think about it this way, you have numbers between 0 and 65536 and you assign a volume to each one and for simplicity they have to be evenly spaced. We could say that for every step up the volume increases by a big ammount (say the equivalent of fingers snapping) and then we would be able to record very loud noises and very quiet noises (think about how loud 65536 times a finger snap would be) OR we could instead say that one step is corresponds to a small amount (lets say a fly breathing) then we would not be able to record a very loud noise (65536 times a fly breathing is still not all that loud) but we could differentiate between noises that were very close together in loudness.

 

Thats in theory at least. What they might do is pick a sound level that they think is so small that people can't differentiate two sounds that far apart and make that the step size (this is of course relative though because of amplification so there would have to be some reference voltage/speaker). But 16-bits might not be enough for them to record a loud enough sound, so they could cheat and make the step bigger (making the difference between two sounds big enough that we could hear it, would probably sound like white noise) or they could use more bits (24, 32...). Or they could say that zero isn't actually zero but is the quietest sound possible which could be much louder than true zero. This would shift the whole range up so then you could record a sound as loud as you want, but you would have a minimum volume. Either way the range would be fixed by the bit-width and step size. I think what happens is that DbSPL increases by an agreed upon amount (20uPa according to Wikipedia) and so each step in digital audio is that big. The limit of dynamic range is based on that step size.

 

But because this is going to be amplified the step size can change. Amplification multiplies everything by a fixed amount, so if the difference at some reference voltage was 20uPa if we amplify by a factor of 10, the difference is now 200uPa which might be more audible. So very quiet tracks that you have to turn up might have more noise. So the solution would be record those passages louder, but that would limit your dynamic range much more than 96Db since you're no longer using the stuff that would need to be amplified. I really don't know how all this works out with real recordings but it is at least conceivable that because of amplification the dynamic range could be small enough that 24-bit was warranted. 32-bit however I highly doubt, remember that binary numbers increase by a factor of 2 for every extra bit, so a 32-bit number isn't twice as big as a 16-bit, its 2^16 times as big which means it can refer to 2^32 = 4,294,967,295 different values. If 16-bit was so close that you can barely tell, then even 17-bit would be plenty (except that computer people don't like things that arn't powers of two or at least multiples of 8).

 

In other words: The quality of your soundcard is much more determined by the quality of the chips and the algorithms they use and the sheilding they employ etc... than by the bit-depth. Don't worry about it. The sound quality is much more important. Not to mention that no one sells 32-bit audio to consumers its only used for recording because you can use the extra bits for things like round-off error or other little tricks to make it sound better.

 

 

Anyone with real DSP experience wanna chime in and correct anything? I'm guessing on some of this stuff so a real expert would not be unwelcome. I wanna learn more about this stuff too.

 

Edited by VioletConqueror - 4/29/11 at 11:52pm
post #10 of 42

In addition you can calculate back the number of used bits when you have the dynamic range of your system..

post #11 of 42

^^ Does that mean you record with far more bits than you could possibly need and then normalize down to make it fit in whatever bit-depth you need? I wish I enough time to take some DSP classes (I'm a Computer Engineering student right now) but I really don't have enough time. This stuff is really cool though.

post #12 of 42
24 bit is often used for recording because it is definitely "safe." Recording on 16 bit could potentially lead to clipping before compression, but with 24 bit, it can be compressed to avoid all clipping. I don't see the point, however, of 24 bit in consumer playback.

My DAC supports 32/192. I can't tell the difference.
post #13 of 42

Let's work out some numbers, and let's consider the max volume recorded volume on a CD at 0 dB, which means that the softest detail is at -96 dB.

 

Let's consider The Planets (Holst) conducted by Dutoit at the Montreal symphonic orchestra. It is a CD with VERY large dynamics.

The loudest instant is at - 1 dB, the recoding has an average of -30 dB, that's a very soft recording, very akin to a torture test.

You want to listen at a loud level, actually louder the real life performance, let's say 90 dB, the peaks are at 120 dB (~equivalent of a gunshot at 1 m)

 

The passage were one would notice a lack of detail are the softest ones, for example, the beginning of the Saturn passage, averaging at -47 dB with a minimum at -60 dB.

Which means that it's the softest detail can still be 96-60=36 dB softer than the softest second of music.

 

Let's see how this plays into our scenario, 120-60=60, at this exact second, the music plays at 60 dB, with the smallest details drowned at 6 dB softer in the room's noisefloor, that's no even taking into account that when playing a signal at 60 dB, the main signal somewhat maskes the smaller details (ie. it's easier to hear a mosquito flying in a silent room than in a rock concert).

 

Now, some of you may say that 36 dB below the main signal is not enough, this is where dither+noise shaping comes in, basically it's an acoustical trick that brings the subjective dynamics of a 16 bit file up to 120 dB (you can look up the details, it's quite complicated).

Now we get our softest signals at 0 dB, 30 dB below the noise floor of the room and 60 dB below out main signal.

 

All problems are solved, 16 bit playback is indeed enough.

 

Finally, why do I call this a torture test? Because a "normal" CD that has not succumbed to the loudness wars is usually mastered much louder, maybe at -12 to -15 dB, with its softest passages at -30 dB and not at -60 dB like in the Planets CD.

PS: Graphic example of how a 1 bit (black and white) can subjectively have a greater bit depth (ie. more shades of gray) if you look at it form a certain distance.

Michelangelo%27s_David_-_Floyd-Steinberg.png


Edited by khaos974 - 4/30/11 at 4:52am
post #14 of 42

If the jump from 16 to 24-bit depth mainly has to do with the dynamic range of sound, and isn't likely to be perceived due to the dynamic range already allowed by 16-bit, why is the jump from 8 to 16-bit sound so instantly noticeable, even on very poor speakers?

 

I remember listening to a bunch of 8- and 16-bit clips back when I first got a sound card on my PC.

post #15 of 42

I think naike was getting at the fact that a lot of DAWs and sound applications can internally process the sound at higher bit-depths than what hardware supports (32-bit for Foobar, 48-bit for Pro Tools, etc.). Even in professional studios, most hardware sound cards and DACs only support up to 24-bit.

 

As other posters have commented, 16-bit has a ton of dynamic range if used properly, so what is the benefit of higher bit-rates? Dynamic accuracy. Less interpolation. Even if we're talking about rock music that only uses the top 10% of available dynamic range, that 10% represents 6,553.6 values in a 16-bit system and 1,677,721.6 values in a 24-bit system. This is the same benefit seen with an increased sampling rate. Not only does an increased sampling rate result in higher representable frequencies, it results in more accuracy in the audible range due to a decreased need for interpolation because of the additional samples.


Edited by infinitesymphony - 4/30/11 at 8:49am
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: Sound Science
Head-Fi.org › Forums › Equipment Forums › Sound Science › Bit depth