Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[Nathan Riddle]

I'm with Tim - I think this is a false equivalency.

Perhaps.

96KHz is a faster sample rate.  If you do the same math you get a higher Nyquist frequency and probably less latency, but other than that, I'm unconvinced.

On a related note, I have two Yamaha 01v96 mixers - the "96" meaning 96KHz capable that I would happily sell for a premium price to anyone who highly values the sample rate, which to my ears sounds significantly worse than my almost 10 year newer lame-o 48KHz-only A&H GLD desks. 

Which sound worse than the SQ or dLive series. Why?

I don't think anyone here is arguing that the 96k is the reason they sound better. They put the 96k in for marketing edit: (so they can make more money) and engineering spent money on algorithms/better components.

[Tim McCulloch]

I don't think anyone here is arguing that the 96k is the reason they sound better. They put the 96k in for marketing edit: (so they can make more money) and engineering spent money on algorithms/better components.

Yeah, and Mackie certainly lent some bullshit mystique to the word "unity" that, to this day, cannot be successfully beaten or educated out of the masses.  They made a bundle with it, though I'd assert that Greg didn't get to spend the profits on R&D.

[Robert Lofgren]

I’m quoting randomly here.

Most mixers/daw’s uses oversampling (2x-4x) for all of its processing to make sure Mr.Nyquist doesn’t introduce aliasing and frequency loss due to the sometimes complex computation. But in the case of daw’s (and I assume in mixers as well) the oversampling is fixed at e.g. 96k so a faster sample rate than e.g. 48k would not speed up any processing. In some cases there could be some additional latency due to SRC though.

There are some cases where higher sample rate in theory would lower the latency. Using a x32 stagebox introduces a 3 samples of a delay compared to the local io. At 96k the latency would be cut in half,  but this is still such a low number that it doesn’t matter in reality.

I call bullshit.  Explain the exact engineering principles involved and verify with empirical evidence.  Extra points for a double blind listening session in which the outcome is greater that 50%.

That more work can be done does not mean a sonically superior outcome.  There is no inherency involved here.

[Tim McCulloch]

I’m quoting randomly here.

Most mixers/daw’s uses oversampling (2x-4x) for all of its processing to make sure Mr.Nyquist doesn’t introduce aliasing and frequency loss due to the sometimes complex computation. But in the case of daw’s (and I assume in mixers as well) the oversampling is fixed at e.g. 96k so a faster sample rate than e.g. 48k would not speed up any processing. In some cases there could be some additional latency due to SRC though.

There are some cases where higher sample rate in theory would lower the latency. Using a x32 stagebox introduces a 3 samples of a delay compared to the local io. At 96k the latency would be cut in half,  but this is still such a low number that it doesn’t matter in reality.

You and I have been in several of these discussions over the years and I think we've mostly agreed and will continue to do so.

My challenge to Scott was to show the homework that better audition is inherent in 96kHz sampling.  Note the word "inherent".  TJ points out that to him, his 01v96 running at 96kHz sounds inferior to his A-H desk at half that rate so anecdotally we can say that for 1 user, subjectively improved audio does *not* depend on 96k.  See where I'm going?

I don't doubt that users find audible differences between consoles, the question is can a double blind panel identify 96 or 48k on the same desk at better than half the time?  Unlikely.  That most folks don't hear a difference doesn't make 96k an invalid choice, only one that can't be defended on the basis of perceived 'sound quality'.

Latency is more than just sample rate, though.  The question is how fast the processing can do its work, assemble it in a buffer and then output the result.  Faster samples don't make for less work, they make for more.  How much less latency you get at 96k can and does make a difference for some artists on IEMs and that's a valid reason to use this sample rate.

If the artist wants everything recorded - sound checks, performances, jamming - I can see 96k as a choice the artist might make for reasons beyond our immediate, live-audio concerns.

It's a great time to be in audio.

[Stephen Kirby]

I'm a little confused about something here.  If you double the sample rate, don't you have to do twice the number of computations on any processing?  Yes, the first byte with computations on it will come out in half the time, (lower latency) but that's providing the processor can keep up.  You don't get better processing, you get less.  Unless you do something to correspondingly up the processing power to keep up with the additional samples.

I got a new interface for my ProTools that runs at 96, but my old computer glitches if I try to do much DSP.  So I keep it at 48.  Seems like the same thing would be happening in a mixing board.

[Mac Kerr]

I got a new interface for my ProTools that runs at 96, but my old computer glitches if I try to do much DSP.  So I keep it at 48.  Seems like the same thing would be happening in a mixing board.

Somewhat, but with ProTools they add features and higher specs and expect you to provide a robust enough computer to run it. With a console the manufacturer provides enough processing power for the console. With some consoles there are varying amounts of processing that you change by adding enough DSP cards for what you want to do.

Mac

[Tim McCulloch]

I'm a little confused about something here.  If you double the sample rate, don't you have to do twice the number of computations on any processing?  Yes, the first byte with computations on it will come out in half the time, (lower latency) but that's providing the processor can keep up.  You don't get better processing, you get less.  Unless you do something to correspondingly up the processing power to keep up with the additional samples.

I got a new interface for my ProTools that runs at 96, but my old computer glitches if I try to do much DSP.  So I keep it at 48.  Seems like the same thing would be happening in a mixing board.

That's what I said in reply #23.

[Stephen Kirby]

Somewhat, but with ProTools they add features and higher specs and expect you to provide a robust enough computer to run it. With a console the manufacturer provides enough processing power for the console. With some consoles there are varying amounts of processing that you change by adding enough DSP cards for what you want to do.

Mac

Understood.  Some folks here seem to think that it's just a firmware update on existing hardware since the I/O claims 96.

[Robert Lofgren]

In theory it is just a firmware update since the hardware is 96k capable.

However, there is not enough computation power to sustain full processing on all ch/busses for 96k and would therefor cripple the usability of the mixer quite severely. I think that uli once said that they could only handle half of the ch/busses/fx if they were to implement 96k. The limitations outweighed any benefits that might been had from using 96k so they eventually decided against it.

Understood.  Some folks here seem to think that it's just a firmware update on existing hardware since the I/O claims 96.

[Luke Geis]

I was under the impression that the 96khz was more for the stage box and certain parts of the audio chain such as the I/O. The stage box adds another 1.3 or 1.8ms of latency to the audio from IP to OP with the X32 and M32. 96khz would halve that extra latency.

[ProSoundWeb Community]