Sinesum Message without creating zipper noises / clicks & Pops

Reply to Sinesum Message without creating zipper noises / clicks & Pops on Sun, 14 Jun 2020 01:06:00 GMT

Roomy — Sun, 14 Jun 2020 01:06:00 GMT

Hello All!

I am trying to add harmonics to a table using the sinsum message function with fair results. Unfortunately though, when the message is triggered (every time a harmonic is added/changed) Pd makes an audible mess. I thought it was some kind of table reading error where its caught on a DC offset. But the noise is just pd having a hiccup from the function of the message.

Any idea how I can recreate the sinesum message without it using [; Sinesum(??

Video of my patch. You can see from the sliders on the right I am adding harmonics (1,2,3,4...) to the sinesum and creating some nasty noise. Theoretically, there shouldn't be any noise as an [osc~] obj is the only thing audible.
Untitled.mov

Reply to Sinesum Message without creating zipper noises / clicks & Pops on Sun, 14 Jun 2020 06:45:11 GMT

whale-av — Sun, 14 Jun 2020 06:45:11 GMT

@Roomy The easiest way is to duck the audio as the graph is redrawn.
See your other thread..........https://forum.pdpatchrepo.info/topic/12892/how-can-i-trigger-a-message-when-any-array-value-is-changed/4 for a solution.
David.

Reply to Sinesum Message without creating zipper noises / clicks & Pops on Sun, 14 Jun 2020 10:12:53 GMT

jameslo — Sun, 14 Jun 2020 10:12:53 GMT

@Roomy It looks and sounds like PD is falling behind in audio processing because you're asking it to fill the entire table with 12 harmonics for every single update coming from any slider. PD would have to complete that between processing audio blocks. I'm wondering if it still happens when you only write the first harmonic? If the noise goes away, then since you only move 1 slider at a time, maybe it's possible to just update the amplitude of that one harmonic by adding the difference, positive or negative, from the last value? You wouldn't be able to use sinesum in that case--you'd have to manually step through the table and add the sinusoidal difference to each sample.

Reply to Sinesum Message without creating zipper noises / clicks & Pops on Sun, 14 Jun 2020 22:53:03 GMT

ddw_music — Sun, 14 Jun 2020 22:53:03 GMT

It looks and sounds like PD is falling behind in audio processing because you're asking it to fill the entire table with 12 harmonics for every single update coming from any slider.

You might also try to reduce the number of updates. Sliders are (probably) firing dozens of messages per second. In Max I'd have used [speedlim] to lower the rate of messages. Maybe cyclone has that...? I didn't look. Or hack something up with a delay and a spigot.

hjh

Reply to Sinesum Message without creating zipper noises / clicks & Pops on Mon, 15 Jun 2020 18:58:20 GMT

jyg — Mon, 15 Jun 2020 18:58:20 GMT

You can check my sinesum abstraction :

(from my pdr lib, inside inst_tabosc.pd ) https://github.com/jyg/pdr.
The sinesum is computed in signal domain (right window) using ifft~ object, every time a message changes the harmonic content (left window) in $0-harnonics table