I learned programming twenty years ago by clicking View Source on random websites.

Time was when you could just read the code on the average website! That's quite rare now.

I used sfxr in some game jams last year, except I shipped the library with the game and had it generate sounds at runtime. (I don't think it's smaller/faster than the mp3s would have been, but it was very satisfying!)

I also used Software Automatic Mouth (originally for the C64) for TTS:

https://discordier.github.io/sam/

https://fenomas.github.io/wafxr/

It can export wav files, but its main purpose is to generate code snippets for playing the sound effect through webaudio, underneath using a much more general library. But people familiar with audio synthesis might prefer just using the underlying dependency directly - it basically turns arbitrary config into webaudio. This playground shows how it works:

https://fenomas.github.io/wasgen/

I'm not sure what this webpage is, jsxfr is the name of the JavaScript library on nodejs which is a different url to this page.

But yeah once you start generating things procedurally you can do things like generate 10 sounds with subtle variations to make things less repetitive.