5 ms·
> Isn't it better to show the actual data, not what its analog product is? A stairstep isn't "the actual data". The actual data is a series of impulses. While
by tialaramex 6y ago
> Isn't it better to show the actual data, not what its analog product is?
A stairstep isn't "the actual data". The actual data is a series of impulses. While it is possible that somewhere inside a component of your hardware there's a DAC which is creating zero order hold signals to feed into a reconstruction filter as one cheap way to make the PCM data into analogue audio, Audacity has no way to know that and there's no reason either Audacity or its user should care.
> Frankly, that approximation would be wrong in many cases
You could easily plot a curve that's more accurate than the user's display. You can't see inaccuracies that are smaller than a pixel. There's already suitable code for this calculation (it's called a resampler) inside Audacity.
> Those steps are the data being manipulated.
Again nope, you're manipulating PCM data, an Excel-style view of lists of signed integers would be much closer to "the data being manipulated" in this sense but you presumably don't think that's a good idea.
Your apparent confusion as to what's real here despite having "learned about these things" is exactly why I was pleased to see in the other reply that Audacity now shows lollipop sticks illustrating the actual impulses once you zoom in far enough to see them.
If lollipops are the thing we're showing most users for the next twenty years I have my hopes that conversations like this will become no more necessary than arguing with Flat Earthers.
- ddingus 6y ago"More accurate" is not real Which is it? There was an input signal. Those samples will reproduce it. (Assuming an input matched to the sampling capability under discussion.) That is what the video you linked tells people. I am definitely one of the people, and have shared that exact one many times. Showing people the input waveform makes sense. Overlaying it with the data captured also makes sense. An approximation does not make sense. Can we show them the input signal that generated the data easily? If so, great. Let us do that. What I meant by "real data" is that list of integers. That is what is in the machine. We both know that is not what comes out, nor what came in, nor what people hear. Personally, I have always evaluated that by actually listening, or using my scope, same way that engineer did in the pretty great video you linked.
- tialaramex 6y ago> "More accurate" is not real / Which is it? We'd be trying to draw a curve (a band limited audio signal is a smooth curve, although I guess in the very degenerate case that it's completely silent that curve is also a perfectly flat line), on a display made of rectangular pixels, so, that is always going to be an approximation, but that's really neither here nor there. > Can we show them the input signal that generated the data easily? If so, great. Let us do that. Yes, this is very easy to do, in the video you say you've shared many times it shows this (among many other things) because it isn't hard, unless your idea of a fast computer is a 20MHz 80386 or something, in which case, yeah, that's not easy and maybe isn't worth your time. Or what they have chosen to do now (lollipop graph) is also fine, if you zoom into the lollipop graph on your 44.1kHz recording of a Mariah Carey vocal it's still fine, it only starts to get a bit confusing (making the smooth curve valuable to see what's actually happening) at really high frequencies say, 15-20kHz.
- ddingus 6y agoI like the lollypop graph personally. Well, in that video he is reproducing inputs with only modest frequency complexity. And by approximation, I do not mean pixels. I do mean a reconstruction of the input. That drawn as pixels is fine. Won't the compute burden depend on the frequency domain complexity of the input signal? And if that is approximated, I maintain the existing UI representation is just fine, as it is functional.
- tialaramex 6y ago> Well, in that video he is reproducing inputs with only modest frequency complexity. Monty is making them with a signal generator, so that you can easily see what's going on and (if you have suitable gear) you can reproduce the results. The exact same phenomenon would happen for some hypothetical complicated analogue signal. Notice that Monty also demonstrates a square wave, which is in fact the maximum possible "frequency complexity" because of how audio is defined, even though chances are you're still thinking about it as very simple. > Won't the compute burden depend on the frequency domain complexity of the input signal? No, it's just resampling, the same way you would to go from say 44.1kHz (from a CD) to 48kHz (typical modern fixed rate DAC in a cheap PC), except you're maybe going from 40 samples to 1000 pixels if you've zoomed in that far. It'd use a windowed sinc function.