Explain bit depth and frequency response to me

[BDawg]

I understand the basics of bit depth and how it shapes the sound curve to produce a more realistic sound. I also understand how having a larger than 44.1 kHz frequency response can improve dynamics (sort of).

But, what is the point of having a 24/96 (or greater) output resolution if everything you listen to is 16/44 CDs or 20/48 DVDs? People swear that a 24/96 sound card makes the mp3s and CDs they have sound that much better. Why would it? Doesn't crap in == crap out?

 

[PrinceXizor]

If you are outputing a 16/44k signal, in theory, a 24/96k card will not give you any benefits. In theory.

However, even in the digital realm, errors occur and the higher capacity of the audio card will help cover the errors thus leading to increased signal quality.

Also, there are likely other features of the cards (not directly related to bit depth or frequency) that positively affect sound quality.

1. D-A convertors - Since these convertors are designed to accomodate 24-bit 96Khz signals, they are usually more accurate at reproducing an analog waveform from the digital bits than what may be found on a "cheaper" card that only supports 16/44Khz operation.

2. Clock jitter - this is one item you hear very little about sometimes. Basically, clock jitter is how accurate is the space between samples. In other words, if you are taking 96K of samples per second, how equal are all the spaces between such samples. The D-A conversion algorithms assume perfectly equal spacing, of course. So, the greater the timing difference due to clock jittter, the greater the possibilty of the algorithms piecing together the digital samples in a slightly distorted way. Again, 96K is a higher frequency than 44K, so the clocks on these cards are usually higher quality, so, when playing back a 44K signal, you still benefit from the improved jitter.

These are a few things I could think of. Perhaps others can comment as well.

P-X

 

[BDawg]

Out of 67 views, nobody else knows anything about this subject?

 

[onix]

From a theoretical perspective, the Nyquist criterion requires that we sample at a rate of only 2X times our desired bandwidth - that would mean 44.1kHz is overkill, since the very extreme limit of human hearing is 20kHz. However to sample at exactly 40kHz requires prefect low-pass filters, i.e. flat-band until 20kHz, and zero-pass above), no noise, a perfect clock, etc.

But in an imperfect world we have to make some accomodations - and some of the issues raised by PrinceXizor are valid.

The 16-bit resolution of CD audio cannot be improved upon either, but how well we can even reproduce the dynamic range (i.e. bit depth) depends again on practical issues such as noise shaping (in one-bit & multibit systems) and filtering.

Ususally the additional headroom is benefitial and allows for looser tolerance in design of filtering and DSP algorithms. The labels used by manufacturers are just hype; what you should care about are the quailty of the components used and their ability to reproduce what has been recorded. The latter (I repeat) cannot be improved upon.

 

[f95toli]

Another, more practial, reason is that most (if not all) new audio D/A are designed to work with SACD and DVD-A (even very cheap DVD-players can often handle one or both "superaudio" formats nowadays)
.
So even if a system is designed to work only with 16 bit 44 kHz signals it might use a D/A which handles all signals as if they were 24 bit; this does not matter for sound reproduction but has the advantage that the marketing deparment can put 24 bit/96 kHz (or even 192kHz) on the box.