Techniques for Pitch Follower Smoothing? Guitar and synth trade-offs

[CodePoet]

I'm setting up a preset where I want to hear my guitar amp tone when playing down on the lower strings and have a synth sound when I jump up to high notes on the top couple of strings. I have things set up with the pitch follower on the synth output level and overall it's working fairly well.

The only nit about it comes from moving between the high (synth) and low (guitar) parts. I get a very fast, glitchy synth squeak as I move from the high notes down to the low guitar notes - almost like the pitch follower still has the last high note initially buffered before recognizing the low note again. You can hear it in the example - it's very fast here but is exaggerated in the real scenario with additional delay/reverb on the preset.

I've tried playing it very staccato and making sure I cut off any ringing notes as I move from high to low and also tried some other things with the gates, attack/release on the modifier, etc. but can't seem to get rid of it even keeping the notes totally separate with longer silence between them.

I attached a very stripped-down version of the preset and a sample in case anyone is able to take a listen and a look. Thanks for any assistance!



May need some @Bakerman or @simeon -level support for this one.

 

[Bakerman]

For the synth level modifier try update rate = fast not slow, and set release to 0 ms. Nonzero attack is fine and probably desirable to avoid other blips when playing low.

Also try modifying level with a volume block (volume not level parameter) instead of synth block level. It could be that block levels have a slower maximum rate of change like other examples here: https://forum.fractalaudio.com/threads/faster-bypass-mixer-multiplexer-ramps.147609/

You might be right about pitch follower pausing at the last high note when your level drops below the detection threshold. Watching the modifier dot will indicate this. The 5th to 2nd last chords didn't seem to have the noise so maybe the detector value was already low there.

If you can tolerate a bit of extra latency on the synth sound, one last thing that should help is delaying it slightly (10-40 ms) and modifying volume post-delay.

 

[CodePoet]

@Bakerman - thanks for the analysis. I think I had tried the modifier update rate at all of the values but will try again with the release. I'll try your other ideas.

In regard to the delay, I think I could tolerate some delay if it helped clean it up - but would that just push the same result later? Would there be a way to delay and discard the first x milliseconds of the pitch input via a gate perhaps?

 

[CodePoet]

I moved the modifier on the level to the volume block (volume parameter) with a 40ms delay in front of it - that has helped mitigate it a bit. I tried different delay values - 40ms worked better than 20ms or 50ms. It's still a bit glitchy underneath but this helps to hide it somewhat.

I do notice that after you play up high and stop, the pitch follower modifier stays up high, so it seems if you play a low note next, the synth is triggering using that high pitch follower value before it is ultimately read and brought back down which causes the glitching. (That's my wild guess based on no info anyway ).

 

[Bakerman]

The delay should go before the synth block. I realize now that wasn't clear in the post above. The pitch-volume modifier can be anywhere after the delay.

I think the II still has faster pitch detection (or modifier/parameter response, whatever's causing this problem on the III) for this sort of thing. This is a pitch modifier on synth voice 1 level, 12 ms damping, no delay. In the last few seconds I was freezing the detector on a high note by rolling the volume down then silently back to 10, but I guess it was also stopping high anyway in the first part of the clip.

https://dl.dropboxusercontent.com/s/70sw60u8sw7dnnj/axe2-synthpitchgate.mp3

 

[CodePoet]

Got it - yes, thanks @Bakerman. My best luck so far is with a delay into a volume block to control the volume level into the synth. I'll work it into a full preset and record some samples.

Interesting that you find the II works better than the III in this area - perhaps the III could be tweaked or improved in this area.

Thanks much for the in-depth analysis here @Bakerman - your knowledge on the platform is huge and much appreciated!