It could be due to the fact that ... the Fast Track Ultra ... is optimized for 96/24, whereas another audio interface may work better at a lower sample rate.

Dan Lavry has said as much. An interface designer has to pick a sample rate to design around, and might well choose 96KHz as that optimal rate. I would have expected a prosumer-class device like the Fast Track Ultra, being targeted at hobbyists, would have been optimized for 44.1. But who knows what their reasoning was?

And of course you do get lower latency with higher sample rates, whether it sounds better or not. For me, 5ms is fine. It's no different than being 5 feet away from your amp or monitor on stage. I don't play soft synths in real time, as I prefer hardware synths for tracking and then substituting computer-based samples later on. Latency is therefore never an issue.

But mainly, 44.1 makes sense for me because I use a lot of synthesizers. Whether soft or hard, they are all playing 44.1k or 48k samples. There is no benefit to upsampling them, so I'd just be burning up disk space for nothing.

I understand there is a company in Los Angeles that will insert pixie dust into your interface for eight hundred dollars.

bitflipper


I understand there is a company in Los Angeles that will insert pixie dust into your interface for eight hundred dollars.

I will do it for $300.

rabeach


Audio signals that we sample are not signals that are bandlimited and sampled for infinite time therefore per the theorem cannot be perfectly reconstructed using the nyquist frequency.

But again, we are talking about audio for humans to hear. As far as human perception is concerned, the signal is sufficiently band limited. Whatever the filters do not entirely reject is at such a low level that we can not perceive it either at it's original frequency (which also falls outside of our hearing range) or after it aliases back. And the same goes for the errors introduced by the lack of infinite sample points. Any errors are buried deep below the noise floor where we can't hear them.

UnderTow

can we talk wow and flutter and get this over with !!

bitflipper

It could be due to the fact that ... the Fast Track Ultra ... is optimized for 96/24, whereas another audio interface may work better at a lower sample rate.

Dan Lavry has said as much. An interface designer has to pick a sample rate to design around, and might well choose 96KHz as that optimal rate. I would have expected a prosumer-class device like the Fast Track Ultra, being targeted at hobbyists, would have been optimized for 44.1. But who knows what their reasoning was?

Now that you mention it, I remember reading that article by him. There's a PDF of a paper he wrote floating around the forums somewhere iirc.

My latency is quite high when I set it to 44.1. IIRC, it's around 8 ~ 9ms at 128 samples. I checked their site to see if there was an optimal sample rate but I couldn't find anything.

Bub


How does bit depth (dynamic range) and sample rate (frequency range) in the digital domain compare to the analog domain? I've been reading that professional analog tape has a frequency range of 10Hz ~ 30kHz(+) so to be able to emulate analog (which is what we're all striving for isn't it?), wouldn't you need to be recording in the digital domain at at least 88.2/24?

Here's some good reading on Analog Vs. Digital and 96kHz. <- This information is leached from another page that is linked at the bottom. This page has an interesting analog to digital reference chart at the bottom that's not on the page the original information came from.

BTW ... it references 'resolution' several times.

Here here...now we start get closer to the truth... That's why I and so many else Professionals use higher then 44.1 kHz 24bit

Bub

bitflipper

It could be due to the fact that ... the Fast Track Ultra ... is optimized for 96/24, whereas another audio interface may work better at a lower sample rate.

Dan Lavry has said as much. An interface designer has to pick a sample rate to design around, and might well choose 96KHz as that optimal rate. I would have expected a prosumer-class device like the Fast Track Ultra, being targeted at hobbyists, would have been optimized for 44.1. But who knows what their reasoning was?

Now that you mention it, I remember reading that article by him. There's a PDF of a paper he wrote floating around the forums somewhere iirc.

http://www.lavryengineeri...ts/Sampling_Theory.pdf

UnderTow

UnderTow

rabeach


the point is I guess that humans don't perceive it due to the technology they use to acquire it. The value could be undiscovered.

Rubbish. The technology is not limiting us. For instance, we can go way beyond the frequency abilities of our auditory system. If we can't perceive a difference when having more than twice (88.2Khz) the bandwidth, or four times (176.4Khz) or even more than eight times the bandwith (384Khz sampling rates) than we can actually perceive, there is absolutely no reason whatsoever to believe that we would perceive an even bigger bandwidth increase and believing that is just silly. Don't forget that these are off the shelf products available to anyone. It isn't even at the real limits of our technology. (But no one in their right mind would consider going even further into cookoo land than we already have gone with some of these converters).

UnderTow

It is plausable that sampling above the nyquist could provide a more accurate reconstruction of the signal than that offered by sampling at the nyquist frequency. This is plausable because the nyquist theory doesn't offer to provide a perfectly reconstructed signal for the type of signals we are working with. I believe it comes down to sowing and reaping. What is it worth with no perceived value in this day and time to employ higher than nyquist sampling frequencies. Technology evolves and tomorrow all that digital data that is oversampled may have value. To say it canot or does not is rubbish. To say it may or may not is not.

I get the impression you are talking about magic. Some unknown thing that can neither be quantified or perceived. It has no business being introduced into a rational discussion

What do you think the nyquist theory is? Do you think there are any infinitely sampled perfectly bandlimited signals that have been tested to prove that it works.

You have yet to show that a high sample rate can reconstruct it as you put it any better then a lower one.

you are kidding right?

John


Maybe you don't  get it but I have been saying that 44.1 is all you need to reconstruct a frequency band from 20 to 20,000 Hz. That is what the dispute is all about. I have ask you to show where this is wrong. You have not done so so now you ask "are you kidding". You seem to think that one can get "finer" detail with a higher sample rate and that is where you are wrong. Its really a matter of either it can be done or not. It can't be halfway done. That is what you fail to understand. All a higher sample rate gives one is extended band width and nothing more.  It does not increase the accuracy of the frequencies sampled. 

It is not disputed that sampling at a higher frequency will provide a more accurate representation of a signal that is bandlimited  from 20 - 20kHz.  What is disputed is whether one should do it or not. I have no problem with sampling audio at 44.1kHz that's what I do. I have a problem with statements to the effect that there is no reason to sample higher because nyquist says its not needed. Which is an incorrect statement.

double post

It is not disputed that sampling at a higher frequency will provide a more accurate representation of a signal that is bandlimited from 20 - 20kHz.

That is exactly what is being disputed because it isn't true. You can't have read this thread and still believe that.

For reduced discombobulation all round.

I use 24/48