Do you know all effects types in detail or do you outsource some?

[mystran]

Feasible for some one person on average.

The average person does not know how to write code and probably doesn't even want to learn. The average person also probably tries their best to avoid math to the point where trying to teach them DSP is a futile effort. When you're looking at a niche field that requires specialized knowledge thinking about "one person on average" is about as pointless as it gets.

To get somewhere, you need to aim at the top and not the average.

[soundmodel]

Well here's one reading list by Valhalla:

https://valhalladsp.com/2021/09/22/gett ... undations/

The Schroeder papers are about 8 years after a Ph.D.

However, judging from how old the references are, maybe Valhalla thinks most about reverbs was written already 20+ years ago.

[whyterabbyt]

Well here's one reading list by Valhalla:

https://valhalladsp.com/2021/09/22/[b]g ... erb-design[/b]-part-2-the-foundations/

The Schroeder papers are about 8 years after a Ph.D.

However, judging from how old the references are, maybe Valhalla thinks most about reverbs was written already 20+ years ago.

Or maybe he specifically picked out material which 'can be considered the foundations of all algorithmic digital reverbs that followed' like he very specifically says?

[soundmodel]

Well here's one reading list by Valhalla:

https://valhalladsp.com/2021/09/22/gett ... undations/

The Schroeder papers are about 8 years after a Ph.D.

However, judging from how old the references are, maybe Valhalla thinks most about reverbs was written already 20+ years ago.

Or maybe he specifically picked out material which 'can be considered the foundations of all algorithmic digital reverbs that followed' like he very specifically says?

Well, if they are indeed foundations, then maybe the issue of development becomes less complex.

However, I am not satisfied with the idea that delay networks are all there is to e.g. reverbs.

Maybe we're also confused about marketing here, because due to marketing we cannot exactly know how authentic or complex something is, or whether someone e.g. tries to sell "analog models" that have no circuit models at all.

Or when we talk about 10s of reverb models, then do we have that much difference or more like 10 presets?

[whyterabbyt]

However, I am not satisfied with the idea that delay networks are all there is to e.g. reverbs.

Bully for you.

Though of course you may want to consider the irony of your stance given your core questioning about levels of expertise etc in this thread.

[soundmodel]

But there is more to reverbs than delay networks. Ray-tracing for example, some sort of wave guides, ...

I sort of naively feel like just ray-traced reverbs could be, what, 10 years and 50+ papers? E.g. in comparison of how much literature there's for lightning in graphics.

Again, how much is reasonable from one developer?

[BertKoor]

The secret to fine reverbs has been cracked decades ago. Don't study academic papers, just read audio news. It was twenty years ago today:
https://www.soundonsound.com/reviews/yamaha-spx2000
And a decade before that:
https://www.soundonsound.com/reviews/digitech-tsr6
And there is competition:
https://www.soundonsound.com/reviews/art-quadrafx

Again, how much is reasonable from one developer?

About two thousand hours of work per year is reasonable. What is accomplished in that time varies wildly. Sometimes one delivers 5x as much as the average, sometimes it's only a fifth.

[dmbaer]

About two thousand hours of work per year is reasonable. What is accomplished in that time varies wildly. Sometimes one delivers 5x as much as the average, sometimes it's only a fifth.

Yep. And, at least in corporate IT, the 5x guy will make maybe 20% more money doing it.

[whyterabbyt]

But there is more to reverbs than delay networks. Ray-tracing for example, some sort of wave guides, ...

waveguides are basically delays
raytracing is basically used for calculating delay times and frequency response


the notion that time-based effects shouldnt be mostly implmented via time-based algorithms doesnt make sense. real-world reverberation is echoes.

[soundmodel]

Yes, but I don't think the concept of sound travel is similar to repeating a sample. Based on this e.g. algorithmic and convolutional reverbs are substantially different.

Then there's the: does it sound good?

I sort of think that there's a reason why Valhalla focuses only on reverbs and delays.

[BertKoor]

Physics (and the simulation of) does not care about what you think...

[soundmodel]

What I meant is that it sounds unlikely that one can do state-of-the-art reverbs while doing something else.

[whyterabbyt]

Yes, but I don't think the concept of sound travel is similar to repeating a sample.

Doesnt really matter. The physics of audo reverberation in a space is time-displaced reflections of the original audio signal, modified by the surfaces they reflect odd.
The first one of these things is delay, the second is filtering.


The notion of 'sample' is inherently flawed in such a context. The source signal is being effected, not played back.

[whyterabbyt]

What I meant is that it sounds unlikely that one can do state-of-the-art reverbs while doing something else.

No, it doesnt sound like that at all.

[syntonica]

You just do one effect at a time. Start with a Delay Line and soon, you'll have flangers, delays, choruses, reverbs, etc. If you're already a math nerd who likes to program, it's not necessarily that hard for a single person to learn how to program all the effects.

That said, you have people like Sean at Valhalla that eats, sleeps and breathe reverb algos, and has done so for years. That's why his plugins are the cream of the crop. It's the generalist vs. specialist dichotomy.

Regarding outsourcing, many do, but it's usually the big houses with good financial backing that do so. While anybody can program, you have to pay for all that knowledge of you want good results.

Sorry for the meandering, random thoughts.