TESTING Pulsar Modular MDN EQ and...

Sinus Well · Feb 29, 2024

xorome said: ↑

PD and BertomEQA send blocks of a single sample of non-silence followed by thousands of 0s (silence) to the plugin. For BEQA, I get about 8000 samples of perfect silence for every one sample of non-silence. I haven't bothered testing PD, but it's going to be similar.

If a plugin (or host - BC PatchWork in this case) implements bypass-on-silence it may prematurely stop processing samples (because silence is wasted CPU - or so a developer might think.)

From BlueCat's webpage:
- No CPU load on idle: processing shuts down when fed with silence (optimal CPU usage).

Hmm!

You can't just cut off filters from processing or you leave them in a broken state (exactly like those ripples in the magnitude response pictures).

Filters have a tail, just like reverbs. If it's a FIR filter, the tail goes in front (delay compensated hopefully) and may be up to a couple hundred or thousand samples long. If it's an IIR filter, the tail goes to the end and may be even longer.

There are ways for plugins to tell the host "hey listen, you must continue feeding me data - even if it's just silence - for X many samples".

BUT not all plugins do this nor do all hosts respect that number.

In other words. What you may be seeing in PD/BEQA is how the plugin (and/or host) reacts to a single 1 followed by hundreds or thousands of 0s. While this is interesting, it's also not showing what it does to 99.99999999999999999999999% of material you throw at it.

I suggested noise because it's a pretty good way to make sure bypass-on-silence can't trigger, no other reason. You can try with any other signal that isn't silence too.

If there's a bypass-on-silence toggle in BC PW or MDN - even better.

I have neither MDN nor BC PatchWork and can't verify that that's what's happening. But the magnitude ripples look an awful lot like throwing 1s at filters that were cut off prematurely in the previous block.
Click to expand...

Agree with everything that has been said. I just don't see noise as a very suitable input material in this case because it has a high level fluctuation, which makes it quite difficult to measure subtle changes in the frequency response.

However... whatever ones preference is, the real point is, why is BC Patchwork in the measurement chain?
Impulse series - plugin - analyzer. Everything else is unnecessary and is just a potential interferer.
Then you will also get reasonable measurements.

Similar Threads - TESTING Pulsar Modular	Forum	Date
testing out schwabe digital gold clip	Software	Feb 6, 2024
on testing m4l jawns	Live	Oct 1, 2023
Testing ChatGPT for potential sound design applications	how to make "that" sound	Feb 22, 2023
Soundtrack MIDI file for testing orchestral libraries?	Film / Video Game Scoring	Sep 14, 2020
Microhone Testing Program is there one	Mixing and Mastering	May 12, 2020

TESTING Pulsar Modular MDN EQ and...

Sinus Well Audiosexual

PROFESSIONAL AUDIO LOVERS

TESTING Pulsar Modular MDN EQ and...

Sinus Well Audiosexual

Useful Searches