Frame or Block processing ?

[Matt_SF]

Hey guys, please excuse my lack of DSP knowledge. The question is in the title : which processing type do I need, and for which purpose ?

An example :
I'm building a delay modulation, and usually I use a contrainer.fix8_block (or 16, 32, ..., 256) for precise modulation, to avoid clicks & cracks. But I can also use a container.frame2_block which does the same (soundwise), but is just a bit more CPU intensive.

[Christoph Hart]

Yes pretty much like you said. The container type defines the modulation rate so you can decide the CPU cost vs. sound quality on a per case basis (but obviously keep in mind that the cpu will go down when you export it).
If you use a feedback node you also should know that the block size is the minimal feedback delay time. That means if you‘re having a delay with a feedback loop in a standard container and you set the delay time to zero, you will get a feedback delay that is as big as the audio buffer size, so for this use case putting it in a 32 or 64 sample block will make it more consistent.

[Matt_SF]

@Christoph-Hart Ah, great. Thanks Christoph for clarification !

[Dan Korneff]

https://www.mathworks.com/help/dsp/ug/sample-and-frame-based-concepts.html

[Matt_SF]

@Dan-Korneff Thank you Dan! A bit of theory never hurts :)