Best Format for recording

[philsmith]

Great discussion, great answers! Just to add a bit, the Nyquist formula pegs the upper frequency limit to about 1.5k times the sample size. So a 16-bit sample has a frequency limit of 24khz. Which is beyond the range of human hearing. So, technically speaking, you really don't need more than a 16-bit sample size. CD and mp3 are both 16-bit. Having said that, I use 24-bit, not just to be hypocritical, but because I have the disk space and what the heck.

The thing about the sample rate is that modern DACs are really good at interpolating. Sample rate would be the most important factor, if it weren't for this fact. It's like how a digital picture is much clearer the more pixels it has. A PCM encoding is much clearer the more samples it has. But if you have the right interpolation, you can blow up a picture and maintain clarity. The same holds true for PCM. There's twice as much information in a 96k sample rate than a 48, but modern DACs interpolate the times between, usually by averaging, and come up with pretty much the same sound.

What I'm saying is that 16-bit, 44.1khz is really adequate. But if you have the cpu and disk space, might as well go for the highest that works for you. Because there's no harm in having a higher quality than you really need.

[Casey H]

Great discussion, great answers! Just to add a bit, the Nyquist formula pegs the upper frequency limit to about 1.5k times the sample size. So a 16-bit sample has a frequency limit of 24khz. Which is beyond the range of human hearing. So, technically speaking, you really don't need more than a 16-bit sample size. CD and mp3 are both 16-bit. Having said that, I use 24-bit, not just to be hypocritical, but because I have the disk space and what the heck.

The thing about the sample rate is that modern DACs are really good at interpolating. Sample rate would be the most important factor, if it weren't for this fact. It's like how a digital picture is much clearer the more pixels it has. A PCM encoding is much clearer the more samples it has. But if you have the right interpolation, you can blow up a picture and maintain clarity. The same holds true for PCM. There's twice as much information in a 96k sample rate than a 48, but modern DACs interpolate the times between, usually by averaging, and come up with pretty much the same sound.

What I'm saying is that 16-bit, 44.1khz is really adequate. But if you have the cpu and disk space, might as well go for the highest that works for you. Because there's no harm in having a higher quality than you really need.

Yes but we know many music libraries ask for 48K 24b or 48K 16b. Given that, you may as well record at 48K/24, mix down that, and then down convert for those who want other formats. So far 48K/24 is the max we get asked for.

[philsmith]

Good point. But I know of one, at least, which is now asking for mp3s@320 kbps. I have feeling that's the future.

[Casey H]

Good point. But I know of one, at least, which is now asking for mp3s@320 kbps. I have feeling that's the future.

Many ask for 320K mp3 either alone or along with the hi res files. But once you export your highest res wav or aif, you then create everything "less" as needed. Side note: I only create mp3 files from a 44.1K hi res because some have issues with 48K mp3s. So I'll convert the 48K/24 to 44.1K/16 for making mp3s and to have for other use. There are a number of good programs for format conversion, I love "Switch" from NCH Audio. I think they still have a free version though the paid version offers more including tagging.

I probably said this before but the process I like is:
(1) 48K/24 main export
(2) 48K/16, 44.1K/16 conversions from main (I rarely or never see 44.1K/24 used but if needed, can do)
(3) 320K, 192K, 128K conversions from 44.1K file

YMMV. FWIW. There is always more than one way to skin a cat.

Casey