On Tuesday, 3 May 2022 at 15:40:45 UTC, Ola Fosheim Grøstad wrote:
> On Tuesday, 3 May 2022 at 14:59:12 UTC, claptrap wrote:
> Ok, but in DSP I think many ideas are obvious if you know the field, but getting the right mix, the right hacks, the right tweaks, getting it to run fast and making it sound good takes a lot of effort (or can happen as an accident :-). I certainly don't doubt that there are many years of highly skilled effort that has gone into the product as it is today. But that is solid engineering, not a moment of "wowsers!" :-D
That's pretty much my experience. The actual math / "engineering" part is fairly straightforward if you're decent at math. But making it sound good is a bit more art than science i reckon. I guess at the end of the day because its being used to make art and that is a much more subjective realm.
> > See to me that's less impressive, I mean I reckon people were doing FM synthesis with analog hardware already. So it was more likely just a refinement, or exploration, it's actually technically pretty simple.
It is difficult to find any individual discovery that is obviously impressive, and I guess putting a sin() into another sin() may seem intuitive, given people already used LFOs. I think the work he put into making it musically useful and expressive creating new types of bell-like sounds is why people emphasis his contribution. I find this wiki-quote a bit funny: «This was Stanford's most lucrative patent at one time, eclipsing many in electronics, computer science, and biotechnology.»
It's just that building blocks in an FM synthesiser are quite simple, at least conceptually, I reckon I could knock one up in about 30 minutes, just the audio part anyway. Even the math is pretty straight forward, what sidebands you'll get etc... I think maybe it seems complicated to the end user cause it's not very user friendly to make presets. But it's actually pretty simple, and was probably already being done on analog gear, I mean I imagine VCOs existed with linear frequency control back then?
AutoTune, i reckon days maybe? Plus a lot of research and months of time experimenting trying to make it not sound like crap?
> > I mean real time pitch tracking and artifact free pitch shifting are orders of magnitude harder problems than FM synthesis.
Many people worked on that though? It is very much the work of a community… In general most things in audio build on something else. Like, the concept of vocoders is in some way ingenious, but it was invented for speech in telecom by Bell labs in 1930s.
That's engineering though isn't it, the higher you get up complexity wise, the more you're building on work done by other people. It doesn't mean we should only be impressed by people who lay foundations.
Permalink
Reply
