quote

“but think about how it only would have 2 samples per cycle on that wave. ”

That’s right but hearing the difference between a sine and a square wave (2 samples/cycle make a square ok ?) at such a high frequency is almost impossible for me.
Add the fact that the mastering process will usually low-pass the mix at 18-20khz and you’ll understand some people don’t oversample or even record at higher rates than 44.1 (for cd or Mp3 final product)

If you have really awesome A-D and D-A conversion with a word clock, and lots of hard drive space and a fast computer. And you are planning on using PLUGINS at any time. Why not give them more data to derive the sums of their equations from?

I am a Practical audio engineer and also an engineering instructor. I am not a mathematician or an expert at DSP. I do hear the difference, I swear.

I agree that if ALL YOU ARE DOING is recording and playing back, 44.1 or 48k is your best bet….because you are not altering digital much. It is, as the converters saw it coming in, and is …what the converters say it is going out, and we do not hear above 20k.

However, if you plan on using a-lot of digital effects processing inside the computer to your captured audio. The plug-in sees more data, although it is working harder, is more accurately calculating the sum of your changes, I hear this being especially helpful with DIGITAL EQ in the HIGH FREQUENCIES.

In. 44.1 the highest frequency you can theoretically capture is 22.05hz right? Ok l know you can’t hear that really, but think about how it only would have 2 samples per cycle on that wave.

Think about trying to tell a digital EQ to do a high 16k shelf and how the more you try and do “High frequencies” with EQ at 44.1 the thinner it sounds. The amount of plugins you can get away with using (according to my EXPERIENCE AND EARS) with digital mixing is a lot less than you can get away with at 88.2 or 96k.

Your responses please. Those who know both the theory and the practice.

Aside from my studio, I am a guitar player. I have played on both dig and analog gear over the years. I have discovered that for the most part, I can’t stand digital guitar effects. Why? They simply sound sterile. It’s like taking a robot and comparing it to a human. They may make it look the same, but it will never be human.

I can literally walk into a show, and tell by the sound if they are using digital or analog effects. That said… It doesn’t mean that one is “better” than the other. It means that you have to decide what sound you like. The argument earlier to say that digital is better, and basically that people who prefer analog are morons is ludicrous. You really make yourself sound like a moron by being so dogmatic.

Consider other ideas, but ultimately do what sounds best for you and your project.

Mark, I appreciate hearing from a pro on the matter.

Are you saying the anti-alias filter has an effect at frequencies below Nyquist? I thought oversampling made that a moot point.

@Mark: >The idea of ‘accuracy’ is the flawed concept

Maybe we’re ultimately saying the same thing.

But all that’s saying is that the harmonics of the combined sounds produce frequencies greater than 22KHz. And we already know that ultrasonic frequencies can’t be represented by a 44.1KHz sample.

@geist: > I know however, that at least mathematically this is true.

Without question.

@Jon Davis: > The difference is when you zoom in on a real sound wave derived from a mix.

Remember, though, that the waveform shown in your DAW is not the same as the sound that comes out of your speaker.

@Jon Davis: > you *can* hear the “jagged” difference.

Can you, though? Have you done blind A/B tests to confirm this?

The main benefit you get from higher sample rates relates to the steepness of the anti-alias filter curve and having more ‘frequency space’ to use a smoother filter.

Even if you band limited the input to your A/D to 20kHZ, and not a frequency over that got through, and then recorded the bandlimited signal at 96kHz, the above benefit of a higher sample rate still applies.

It has nothing to do with frequencies being present in your music above 20kHz, nor does it have anything to do with the ability of engineers to hear above 20kHz, or having more samples to “capture sound more accurately.”

The idea of ‘accuracy’ is the flawed concept, not the idea of higher sample rates. There’s a big difference.

It’s not that the average human ear is sensitive to frequencies above ~20kHz but more and issue of the wave form that is created when you combine two audible frequencies. Say for example a cymbal is putting out a 12kHz and 15kHz tone. The resultant complex wave form will, at certains points in time, have a curve that cannot be written inside of a 44.1kHz sample. The slope is too steep. But when you hear it, you still will hear the two separate frequencies. Therefore you need higher sampling rates to be able to write these complex waves correctly. This is basically the same problem that Neil describes when he talks about the triangle waves.

Now I’ve never really done a proper test to find out what the answer to this question is. But I believe it is a question of human hearing, and not equipment. Can people *perceive* the difference. I make recordings for people to listen to, not machines to analyze. Some people I know claim to hear it, some do not.

I know however, that at least mathematically this is true. You lose resolution in your complex waves carrying multiple very high frequencies.

Oh, and just to chip in on the analog vs. digital thing. Analog in this day and age should be treated as an effect.