AKAI "SynthStation" looks fun

[Kazrog]

TheWinterSnow - did you not see the oscilloscope reads in the article I linked to? I agree with everything in your post (in theory), but in practice, none of that changes the simple fact that once you get into the khz range, digital PCM doesn't accurately describe square waves at all. The DAC becomes completely irrelevant when the data it's being fed isn't even close to accurate.

I agree that analog components are imperfect - but they are still capable of producing 100x more accurate square waves at high frequencies than any digital audio system that I'm aware of. The problem begins in how the data itself is stored - the resolution of the audio files themselves must increase exponentially from present standards to solve the problem adequately.

Does this affect most of the music we're doing? To some small degree, absolutely, but the convenience of digital audio and the promise of future progress and the eventual superiority of digital audio over analog keep me enthused to the extent that I've become not only a producer, but also a developer.

 

[TheWinterSnow]

TheWinterSnow - did you not see the oscilloscope reads in the article I linked to? I agree with everything in your post (in theory), but in practice, none of that changes the simple fact that once you get into the khz range, digital PCM doesn't accurately describe square waves at all. The DAC becomes completely irrelevant when the data it's being fed isn't even close to accurate.

This effects both analog inputs being recorded and digitally created square waves. An analog square wave generator will look the same after it has been recorded to a digital medium, as the problem comes from converting both forwards and back, actually making recording an analog sample WORSE because it goes through the process twice.

The latest link you posted the guy even said, the DA conversion was smoothed out, as there is no way in hell that a square wave could become a sine wave without a filter.

I made some digital square waves and looked at them through my DAW. Now what I saw would be very typical in the circuits that I was building merely a few weeks ago. As the square wave begins to reach SRf/2 the summing ability is limited to a doubled rise and fall time, and from his rise and fall time calculations would reach the point of of being a triangle wave at around SRf/2(.75) which turns out to be 16.5KHz. When I ran through the various square waves the point in which the wave became a full triangle wave was about 18KHz, and partially became triangle due to sampling lag (which I will explain in a second) at 12KHz which holds the equation SRf/2(.75)(1/sqr2) where the frequency 12KHz has become a null in the frequency response due to harmonic distortion, its still there, the harmonics of the triangle wave have essentially cancelled them out from a harmonic standpoint, causing minor gauss effect.

Sampling lag is the biggest issue, not an increased slew rate. Because the summing amps are what are taking the digital logic and converting is to audio, they are responsible for most of the slew rate, however since audio frequencies do no line up with a new sample being taken and will not be given to the DA until after the new sample has taken place. This applies to all frequencies but gets worse at higher frequencies noticeable at SRf/2(.75)(1/sqr2) and is the largest reason for audible distortion in a square signal.

for your argument to hold extremely true, the only way an analog synth would be better would be that you took an analog synth and plugged it directly into an audio amplifier and a speaker. Digital or not, the signal will be digitized and this effects all audio in the digital world period, we all know this, that is why we always talk about how analog is better. That is why the high end of digital recording is soft and distorted. Notice how I said the point of which a 44.1KHz sample rate begins the ill effect of causing bad harmonics distortion? 12KHz, the same spot we low pass guitars, there is a reason that we always low passed there, everything beyond that on guitars are noise, and now you know why its nothing but noise. So in reality this effects everything, not just square waves.

So what you say do holds some water, this will be effective for all square waves that are not PURE analog, so if you have an analog synth on a CD or DVD for that matter, the distortion will apply and even more so for the analog synth because it does the process twice.

 

[Kazrog]

for your argument to hold extremely true, the only way an analog synth would be better would be that you took an analog synth and plugged it directly into an audio amplifier and a speaker.

To be clear, when discussing the analog synth example, I am talking about a 100% pure analog signal path. Scenarios such as synth to amp, or synth to tape, synth to tape to vinyl, etc.

We still haven't beaten tape in terms of sample rate. Once we do, we'll be much better off.

 

[TheWinterSnow]

192 does have the sampling rate that would be basically undetectable but we haven't standardized it as the main sample rate of digital audio. We are still using the same technology that was used 20 years ago because you can store more information. Hell most everyone today (the average listener) is listening to subpar 16/44.1 with all the .mp3 on their mp3 players, you know, to get more songs on them.

 

[Kazrog]

192 does have the sampling rate that would be basically undetectable but we haven't standardized it as the main sample rate of digital audio. We are still using the same technology that was used 20 years ago because you can store more information. Hell most everyone today (the average listener) is listening to subpar 16/44.1 with all the .mp3 on their mp3 players, you know, to get more songs on them.

+1, totally agreed. I think 192 is probably "good enough." Probably...

 

[drew_drummer]

Personally I think 44.1khz is "good enough" and that 192 is most definitely overkill.

On my bands first EP, I recorded at 88.2khz - figured that'd cover nyquist good enough. But when it came to using amp sims on the DI'd guitars and using plugins... I very quickly maxed out my Core 2 Duo at the time.

For the album I went with 44.1khz, because I couldn't be arsed with messing about with bouncing tracks to save CPU. I think the results were great, and I don't regret tracking and mixing in 44.1khz.

If I had a phat rack of preamps, eq's, compressors, and other effects, and everything was going to be audio... I'd record at 88.2khz and dither down to 44.1khz at the end of the mix.