So I’ve talked about chaining onset detectors before but that was more about using them to analyze different things (novelty → amplitude), which is something I still want to revisit for buf-based slicing, but I was thinking about having parallel onset detectors for real-time stuff.
Since @tremblap shared his magical™ settings in the thread about
fluid.ampslice~ I’ve been using that for fantastic results. It’s super fast/tight and because the Sensory Percussion pickup has such a sharp attack, the retriggering can be really short, letting me play buzz rolls etc…
So this works perfectly as it is … when I hit the surface of the drum. Granted, that’s most of the time, but occasionally I may have a bell in my hand, or be holding things above the drum, so in those cases the tight thresholds/settings on the SP pickup don’t really work well, or consistently.
That leads me to think about having parallel onset detection where the SP pickup is the first line of defence and then locks out anything else from happening, assuming (perhaps incorrectly?) that if I hit the head of the drum, that pickup would always pickup the signal and onset first, so it could lockout a DPA that’s a couple inches further away.
So my overall thinking is then having the DPA go through a second onset detector with settings tuned to be as tight as I can manage for an air-based mic (basically my “old settings”). I can then rely on the SP pickup for when I play on the head, and the DPA for when I play off the head.
I’m a bit unsure as to how to approach this architecturally in terms of which can lock out which, and what durations. For example, the
@minslicelength for my SP pickup is really tight at 1024, which conceivably stop blocking the DPA from picking up the resonance of the initial sound. So I don’t think I can just use the lockout from the contact pickup to lock out the air mic, and similar problems happen in the other direction.
The other issue is potentially how to lock things out since each lockout is cross pollinating. I can do some kind of signal-based lockout, or just a
sah or something, but not 100% sure on that.
Either way, wanted to post this on here to see if anyone has explored multiple parallel onset detectors on the same source? (e.g. contact mic + air mic) And if so, how’d you go about doing it?