If your final output will be VHS or DVD PCM audio, then stay with 48 kHz so that you don't have to transcode. If you are going to DVD AC3 then it doesn't matter since it will be transcoded anyway. The basic idea is that you want to avoid transcoding whenever you can, as each pass degrades tha audio somewhat. Good luck,
-Don

For voice-overs, I'd stick at 16 bit 48k. Were it anything with a lot of dynamic range, or a wide frequency spectra, I'd record at 24/48 or 24/96. Yes, it gets dithered down to a 16bit 48k signal at some point, but better the resolution and information was there at some point, so it's a much smaller loss. Upsampling a sound does not degrade the audio in the same manner that downsampling does.

Ah, SPOT, now I've got it too - thanks

DELETED

A symphony orchestra is an example of both wide dynamic range (ie soft to loud passages) and wide frequency range (bass to the highest overtones and "air"). Recorded voice is an example of narrow dynamic and frequency range. Other material may fall somewhat in-between; pop music (the way it's currently recorded) often has a wide frequency spectra but the final product is compressed and normalized so heavily that the dynamic range is smaller.

Kevin.

I've read that the pros use 24-bit recording simply because the dynamic range is much wider. It's not that 16-bit isn't wide (it's huge--s/n is 96db below clipping). The issue is that if you ever hit the 0db mark the artifacts will be horrible, so you make sure the peaks are substantially below 0db. But then you are using up some of that 96db range.

Better to record onto a medium that has enormous dynamic range, and then you can normalize to 0db back in 16-bit mode.

As for sampling at greater than 48K, there are those who swear that they can hear the difference, and those who swear they can't. My gut feel is that the guys in the latter camp are the honest ones.

It's not so much the recording resolution. Even 12/44.1 is pretty good and that's all you get on a CD. You'll most likely hit the limitations of the analogue part of the recording chain before the limitation of digital resolution.

However doing anything in the digital domain at that resolution has a big impact. If you reduce the level by 1dB then the last bit is lost. Say you rendered the track out with the level 1dB down and then bought that back in and bought the level back up 1dB. Your real resolution is now only 11 bits. Things would rapidly start to sound narsty. If you upsampled the original 12 bit track to 16 or 24 bit you're still going to have the same thing happen but loosing on bit in 24 is way less noticeable than loosing 1 in 12.

I only mention this because I've been looking around for a reasonably priced field recorder and was put off a ot of them by the lower bit depth. In reality though 12 bit if used intelligently is more than adequate.

Ummm, CD is 16/44.

MJ

You're right!

Stupid me, I was thinking of the alternate way of recording DV audio, what Sony consider 'standerd' just to confuse a lot of people.
Still the same priciples apply.