Overly ambitious concept: synth-sustain

[earthtonesaudio]

I have this idea of a "synthesized sustain" sort of effect, where you have an internal oscillator which is constantly sweeping through its range, mix it with an sidechained audio signal, and when the sum of the two reaches a certain level, the oscillator locks onto that frequency and is then injected into the main audio path.

My hope is that this would act like a resonant filter that automatically synchronizes with (some of, or at least one of) the notes you play, or maybe like an automatic version (no fancy footswitching required) of the Boss Distortion Feedbacker.

[ashcat_lt]

How long does it sustain?

The effect you're looking for sounds a lot like PianoVerb.  Not sure how they do it.  Some sort of tuned delay line mumbo...

[Taylor]

an automatic version (no fancy footswitching required) of the Boss Distortion Feedbacker.

I would approach this differently: why not take the DF circuit (or more likely just buy the Behringer clone because it's a huge circuit to build/expensive collectible to buy). Then use a pulse wave LFO to do the sustain switching automatically. I assume it sustains when you hold the switch, which is just a logic ON, probably. So use a pulse wave with wide pulse width, set to a period of a few seconds, so that when it flops to max voltage, your switch is held on by the positive logic state.

Somebody told me he was building something kind of like this using 2 of these circuits and my tap tempo tremolo board. Never heard back about it, but I assume he's doing it with something similar to what I describe above. This would let you select the sustain period by changing the LFO speed.

[Gurner]

The hardest part of this project is the "locking on" of the internal oscillator to the incoming guitar signal. The first part of a plucked guitar note is too rich in harmonics to extract a fundamental with ease ...& by the time the harmonics have dropped out a bit (ie further along the timeline), you've got very little time to extract the frequency (or have the oscillator lock onto it), before the note decays beyond the point of no return.

IMHO this one is more suited to some DSP ....it'd take some rather intricate & very clever circuitry to do in analogue.

[earthtonesaudio]

ashcat_lt - That PianoVerb sounds closest to what I'm picturing in my head.  But with mine the frequency would not be split into discrete steps, so the sustained note wouldn't have to be in tune with the local oscillator.

Taylor - That would be an interesting effect in its own right, but not exactly what I am looking for.

Gurner - I agree about the difficulty of extracting a fundamental from the guitar.  The Boss Feedbacker uses fundamental detection, and still doesn't really track that well.  Personally I'd rather lock onto *any* harmonic of the input waveform, rather than only the fundamental.  I think it might sound more interesting.


I think figuring out the dynamics is the hardest part.  Right now I'm thinking the "lock oscillator" command would depend on the input waveform + oscillator waveform reaching some threshold, but the "release oscillator" command would start when only the input enveloped dropped below some level, then have an additional (user adjustable) time constant on top of that.  So you could set it to closely follow your playing dynamics, or stretch it out so it keeps playing for a while after you stop.  Another thing you could do is set an input threshold to override (or not) the oscillator release time, so you could play quietly over the sustaining note, or play loudly to stop that note and obtain a new one.

[spectraljulian]

I'd think what you'd want to do is build a pitch to CV converter feeding a VCO followed by a VCA.  Then have the reverse guitar envelope modulate the VCA. 

Easier said than done though. 

[Paul Marossy]

My "Guitar Synth In A Box" idea from about 7 years ago was to take the guitar signal and route it through a freq to volt converter chip into a VCO, seperate the high and low frequencies and then blend those two signals. I don't remember anymore exactly how far I got with the idea, but I do remember that it had a problem with tracking. It basically didn't like anything below about the 12th fret or so. I kind of gave up on the idea.

http://www.diyguitarist.com/PDF_Files/GSIAB-BlockDiagram.pdf

[earthtonesaudio]

Thanks for the info, Paul.  I'm having a hard enough time getting a "block diagram" of this, let alone actually building something with parts.  A lot of stuff in here is in the "easier said than done" category. 

[spectraljulian]

From what I've read on here, analog Pitch/CV is really hard to pull off DIY. 

I'm wondering if something like Arduino could do Pitch to CV.  Or a DSP. 

[earthtonesaudio]

Most of the attempts at this sort of thing seem to use fundamental extraction, but I think locking onto any prominent input frequency (not just the fundamental) would yield just as interesting an effect.

[Paul Marossy]

Thanks for the info, Paul.  I'm having a hard enough time getting a "block diagram" of this, let alone actually building something with parts.  A lot of stuff in here is in the "easier said than done" category.  

Your OP reminded me of my idea I had a while ago. I think in some cases, block diagrams really help to simply a concept. Yes, it is much easier said than done. But if you can pull it off, it might be pretty darn cool.  

I just never had the time to take my idea further, I was too busy trying to support my new family at the time. Now I have lots of time and no money. Or motivation for that matter.  

Most of the attempts at this sort of thing seem to use fundamental extraction, but I think locking onto any prominent input frequency (not just the fundamental) would yield just as interesting an effect.

Yeah, that could be interesting. It would give it a certain "predictable unpredictability".  

[WangoFett]

Most of the attempts at this sort of thing seem to use fundamental extraction, but I think locking onto any prominent input frequency (not just the fundamental) would yield just as interesting an effect.

Yeah most of the attempts at fundamental extraction involve locking onto any prominent input frequency right?  They rely on filtering to try and make the fundamental more prominent, but even with some filtering they can get confused by the 2nd harmonic (a la familiar MXR blue box circuit), unless there is more sophisticated logic involved (a la Boss OC-2 fundamental extractor).
You can make a 'fundamental' tracker track even more pseudo-random by messing with the filtering e.g. cutting lows instead of highs boosting highs confusing the tracker even more.