"Debunking the digital audio myth"
Mickster
Posts: 2,693
Next time some purist comments that my guitar-amp simulator "sounds digital" 
Craig
Mickster
Posts: 2,693
Next time some purist comments that my guitar-amp simulator "sounds digital"
Craig
Comments
He does mention at the end that recording at higher resolution does have benefits during production. This will be for resampling, tone conversion, and mixing reasons.
Also, that example digitiser used in the demo will be using a quality modern post-DAC analogue filter, presumably via a current-mode DAC. The raw DAC output is stepped as drawn in the graph, it just never makes it past the subsequent analogue filter.
The lollipop graph is not at all a real analogue post-DAC signal. It's a theoretical construct of what the ADC sampled numerical values represent. They are trying to keep the presentation as simple as possible I guess but they do seem to be adding extraneous details without proper significance. Hmm, maybe it's a side effect of taking some clips from a bigger lecture.
Early digital audio equipment definitely would've struggled with the stepping effect. It is real if not filtered away in the analogue domain.
So the quality of the DAC's analogue filter is pretty damn critical. Which is where oversampling and sigma-delta DACs come in. This changed the game in making effective analogue filters easier/cheaper to build. Traded some digital wave shaping, as costs for processing dropped, for a simpler analogue filter ... or better audio quality (less stepping effect) with the same quality of filter.
Yeah, so their demo would've used wave shaping tricks to interpolate its DAC. Even though the example ADC operates at 16-bit 44.1 kHz, that doesn't mean its following DAC output was the same. They aren't showing how much tech and historical evolution is backing that demo.
Even when the stairsteps are present from a simple DAC, they just don't matter. The spectral difference between the bandlimited signal and the zero order hold signal is outside the hearing range. At most you'll get distortions in a poorly designed amplifier.
Also, I'm pretty sure that most people probably can't tell the difference between 32kHz and 44.1kHz.
True. But the purists have a stepped picture on the oscilloscope that they can point at then.
"He does mention at the end that recording at higher resolution does have benefits during production. This will be for resampling, tone conversion, and mixing reasons." - I think this is because "live" instruments can produce ultrasonic sounds and when they combine either by reflection of themselves and/or with other instruments it creates beat frequencies that does add a "spacial component" to the overall music experience we feel. If the "spacial component" is captured during the recording, then I agree that you do not need the higher resolution to play the music back.
Ah-ha. This is how I understood it to be and although I'm familiar with the term ZOH, I didn't really understand what it was. My version was "that bit there"
To be covered by the nyquist thing, we are still talking 15KHz, right? Not sure I can hear that but I absolutely know that I don't want to
It's all kinda moot anyway because the end-product is typically compressed to hell and at the final stages, we drop the full-range monitors and mix/master to get the best compromise that sounds acceptable on a wide range of players.
Yeah, I don't think I'll bother
Craig