I\'m getting a little lost in the references to 24 bit samples. Weren\'t 32 bits, two or three years ago, supposed to be the next step? CoolEdit wanted to create and save things in 32 bits. Then CoolEdit was bought out. Now, the Plugsounds sets have a 32 bit playback mode. (I have no idea what this means for the Plugsound VSTI\'s, except that you can click on a button that says 32 bits--were they recorded at 32 bits?)
I\'m not trying to undermine the new 24 bit libraries. I\'m just trying to understand, from my limited perspective, what\'s happened that suddenly 24 bits is a new standard if 32 bits is possible. The 32 bit sounds I hear are good.
And for that matter, why stop at 32 bits? Is it just that no one has created a hardware or software recorder that can capture at a higher rate?
Bandwidth (potential polyphony) vs. quality. 32 bit samples will double the disk throughput required to play each sample over 16-bit samples. 24-bit only increases the load by 50%.
In many cases, there\'s not even a very good reason to go beyond 16-bit samples. If you\'re not using the dynamic range, there\'s really no advantage. 16-bit will represent many instruments very well...perhaps even most. 24-bit will take you from silence to the threshold of pain with full quality. 32-bit audio will take you from silence to physical harm. No microphones exist which would even capture the range (there are really no mic/preamp/electronic combos available to even capture a true 24-bit range).
Most people agree that 24 bits is plenty enough, as Bruce remarked.
The issue is not so much the bit rate of the samples themselves but the _replay_ rate. Having 32 bits means that when you mix multiple 16 or 24 bit sample streams together you have more headroom to play with and less rounding-up is involved. For instance Steinberg\'s ASIO protocol already use 32 bits internally (even when using 8, 12, 16 or 24 bit samples).
Another issue is that the current OSs and the CPUs they run on have an internal 32 bit architecture, so it makes sense to use that native power when processing data.
No doubt when 64 bit processors and OSs become more prevalent the software vendors will want to take advantage of the extra width to process sound even more effectively and use it as \"selling\" factor.
While the fact that streaming real-world 32-bit audio requires twice the bandwidth than streaming 16-bit material is false, it is true that it need more bandwidth, to get a result which is probably not needed for most applications.