16-bit/48khz VS. 24-bit/96khz

[Robert D]

Plugins don't lower track count in a session, CPU power lowers the amount of plugins you can have available at any time. The CPU has little or nothing to do with how fast a larger file (as in 96k) is read off a drive as compared to smaller files (44.1k). Hard drive I/O speed is the primary determinant of the actual number of tracks you can playback or record at once.

Of course, you could be talking about the 96k processing inside the plugins themselves, which could increase CPU load. In practice, however, I think the benefits outweigh the cost. In my experience this does not cut my CPU power for plug ins in half at 96k compared to 44.1k. YMMV

If your not recording at 96k strictly because your afraid of the CPU load IMHO your missing out.

That first sentance must have been written by Scott McClullan. 24/96 is moving into being somewhat practical, but in the real world, with the huge amounts of data it consumes across multiple projects and backups of those projects, and the processing resources it really does use, no matter how you parse the language, it's a very noticable weight to carry on the way to burning a 16/44.1 CD. If my memory (the grey matter kind) serves me, recent discussions on this showed that the majority opinion and practice was to stick with 24/44.1 for most bound for CD projects.

Cheers,
RD

 

[cultureofgreed]

That first sentance must have been written by Scott McClullan.

HAHA what an insult! Is that fact that I am working on my PhD in Poly Sci that obvious?

24/96 is moving into being somewhat practical, but in the real world, with the huge amounts of data it consumes across multiple projects and backups of those projects, and the processing resources it really does use, no matter how you parse the language, it's a very noticable weight to carry on the way to burning a 16/44.1 CD. If my memory (the grey matter kind) serves me, recent discussions on this showed that the majority opinion and practice was to stick with 24/44.1 for most bound for CD projects.

In my opinion, if the majority thinks that 24/44.1k is more practical then 24/96k the majority needs to shell out the extra hundred dollars for a 300gig drive instead of a 40 gig drive. If the majority is throwing compressors and eqs on every channel in their session and their 24/96k sessions are being bogged down then the majority should invest in decent mic preamps, microphones, and converters so they don't have to rely so heavily on plug ins. If your sound source is good and captured correctly then there is no need for 24 channels, or more, of compressors, eqs, and limiters. Heck, the best soft compressors and eqs are not native anyways. Most of the native stuff sounds like crap. As for backup, a DVDR is less then $50 now and the media is dirt cheap also.

I honestly can't understand the logic in 24/44.1k except perhaps that there may be a loss during truncation down to 16/44.1k from 24/96k.

I dunno, sound is subjective anyways. Maybe I am wrong.

 

[Farview]

If the majority is throwing compressors and eqs on every channel in their session and their 24/96k sessions are being bogged down then the majority should invest in decent mic preamps, microphones, and converters so they don't have to rely so heavily on plug ins.

Depending on the style of music you record, this may be true. However, when you do a lot of the more synthetic genres of music it isn't just a matter of mic technique. When you have no control over the instruments that your clients bring in or the outcome they expect, it takes a little more doing. Or even if you frequently mix things that other people have recorded.

When you are living in a small world and only do your own projects, or at least similar projects, you can have these black and white ideals. A lot of us can't.

BTW, you don't truncate sample rate, that happens to bit depth if you don't dither. Sample rate conversion can be a bigger evil than just recording at the lower rate in the first place.

 

[apl]

Ken:

In a nutshell it's the frequency cutoff point of you're top end The theory states that in order to prevent aliasing at high frequencies you have sample at twice the rate of your desired frequency hence to get 22.5khz you have to sample at 44.1khz.

In theory it's pretty simple, but in practice it's a little more complicated due to the nature of analog low-pass filters. This is why higher quality converters cost more.

Brane, you're very nearly correct. The Nyquist sampling theorem states that if you sample at some frequency, you can perfectly reconstruct a signal up to one half of the sampling frequency. However, if there is information at frequencies above the 1/2 sampling frequency, it will trick the algorithm and show up as f(new)=f(sampling)-(f(sampling)-f(old)). Thus, you need an antialiasing filter to prevent those artifacts. Since you can't make a filter with infinite slope, in general, the anti-aliasing filtering scheme gives you frequency response to about 80% of 1/2 of the sampling frequency. So, for CDs, you're probably good to about 18kHz.

 

[Robert D]

Sample rate conversion can be a bigger evil than just recording at the lower rate in the first place.

This is the crux of the bisket. If it was the case that 24/96 blows away 24/44.1, AND that quality gain survived the SRC process and dither intact, then of course it would be worth the extra overhead to use it. It just doesn't really pan out that way in the end. I think I'll revisit this with my own test again using my Lynx converters though. Last time I tried was with Delta converters.

Culture - I'm glad my Scott M reference wasn't lost on you. That's too funny how much you got it! No offense.

-RD

 

[cultureofgreed]

Depending on the style of music you record, this may be true. However, when you do a lot of the more synthetic genres of music it isn't just a matter of mic technique. When you have no control over the instruments that your clients bring in or the outcome they expect, it takes a little more doing. Or even if you frequently mix things that other people have recorded.

When you are living in a small world and only do your own projects, or at least similar projects, you can have these black and white ideals. A lot of us can't.

BTW, you don't truncate sample rate, that happens to bit depth if you don't dither. Sample rate conversion can be a bigger evil than just recording at the lower rate in the first place.

I meant truncate as in: To shorten by or as if by cutting off. I always did get the common use of that word confused =) Sample rate conversion is the big evil of 96k from what I hear.

If we are talking about a commercial studio then I must be forced to agree with you that you would want to recieve 24/44.1K. It probably saves you money in aspirin. Especially if you being brought in tracks to mix that were previously recorded and sound terrible, the damage has already been done in that case.

 

[chessrock]

If you track at 88.2, the damage from going down to 44.1 is negligable ... at least from what I've read and been taught (I could have heard wrong, obviously, as findings on this stuff seems to continually evolve).

That said, it still isn't very practical at this time, as RobertD so eloquently pointed out. I do, however, believe that once 64-bit processing becomes the standard, a whole new door could be open for the higher sampling rates. Stay tuned.

I'm still surprised to see how many of you were skipping class the day they covered "oversampling." Was Guitar Center having a sale and no one told me about it?
.