Karplus-Strong Synthesis in Pure Data Vanilla (Pd Full Tutorial) | Simon Hutchinson

[@Sazzlytle]

yoooo, thats to cool to see

[@amattenet]

on the high values it stops changing pitch when you go below 1ms of time, could it be that the delread doesn’t accept values > 1ms?

[@SimonHutchinson]

Hmmm. I'm a little out of my depth, but I think you're right. It says the default "block" size for [delread~] and [delwrite~] is 64 samples. Since I'm working at 48KHz sample rate, one "block" is then 1.3333 ms... and that's pretty much exactly where the pitch stops changing. Can anyone out there confirm?

[@lgnbrrtt]

You can get around this by using the [block~] object with an argument of 1, which gives the window a block size of 1 and lets you use smaller time values below 1 ms and get frequencies up to the nyquist. Important note however: [dac~] does not work when audio is reblocked, so you have to use [block~ 1] inside a subpatch along with the synthesis, and then send it outside. One minor correction: if I understand correctly the default block size for *PD* is 64 samples, not for the objects.Tthe delay objects compute audio at whatever the window block size is. Hope this helps!

[@SimonHutchinson]

Thank you! 🙏 Very helpful! Yes. Thanks for the correction: the block size is for the *window* not the *object.*

[@_DRMR_]

Why not use `[1000 $1(` and run straight into the left inlet of the divide? :)

[@SimonHutchinson]

That's a great idea too! 👍 I think my favorite is still the [expr], though, but this suggestion is way better than what I did in the video.

[@_DRMR_]

@@SimonHutchinson I never use expr since I want everything to be compatible with Heavy (and we haven't totally figured out how to implement it ..)

[@SimonHutchinson]

@@_DRMR_ Ah! I didn't know that. That's an excellent tip! Thanks!