[sdiy] Taking a Step towards Digital Synthesis?....

[Richard Wentk]

At 14:37 07/01/2004 -0500, Glen wrote:
>At 01:07 PM 1/7/04 , Jay Schwichtenberg wrote:
>
> >In my opinion 24 bit 48 kHz either raw PCM or with loss
> >less compression would be the ultimate.
>
>I wonder why everyone is so focused on bit depth, instead of the sample
>rate? Almost everyone I have spoken to would rather have 24 bits at 44.1
>kHz than 16 bits at 96 kHz. I don't understand this one bit.  ;)

Dynamic range. noise floor, linearity. Those are just three reasons.

It's true that you'll be lucky to get 21 bits of useful information out of 
a prosumer converter. But that's still an 5 extra bits - a whole 30dB's 
worth of detail. This becomes especially obvious for low level signals (of 
which there are plenty on most recordings) where 16 bits really doesn't cut 
it.

>Once again, your focus seems to be on bit depth. Apparently you don't feel
>the much higher sample rate brings anything useful to the 24/96 format?

Not as much of a difference as the extra bit depth. I doubt many people 
could hear the difference at all on typical domestic equipment.

I've seen people make ridiculous claims such as that ATRAC encoded 
compressed audio sounds *better* than CD. For domestic use, the extra 
bandwidth is largely wasted, except on the tiny minority of the population 
who are professionals, audiophiles, or both.

>I believe that the 44.1 kHz sample rate is more of a problem than the
>16-bit depth. I can hear a certain grittiness in the overtones of a 44.1 or
>48 kHz recording that are much reduced (or gone) in a 96 kHz recording. I
>haven't had the pleasure of hearing any higher sample rates, so I can't say
>if there is anything to gain by raising the sample rate even higher.

I suspect this depends very much on hardware. Good quality 44.1 or 48k 
hardware doesn't sound grainy at all. It's also way more expensive than 
most people can afford.

What 96k really does is relax the filtering requirements. It's much, *much* 
harder to build a high quality filter for 44.1kHz than it is for 96k. The 
flipside is that most filters in 44.1k equipment are further from being 
optimal and so have significantly more audible artefacts.

This isn't a direct result of the sampling rate so much as the economics of 
the situation. It's possible to make smooth sounding 44.1k converters, but 
they're complicated and expensive, so few people bother.

Richard