Comb->bassline (fluid.pitch~ / fluid.ampslice~)

As mentioned in the thread comparing fluid.pitch~ to fluid.bufpitch~, I mentioned a comb audio to bassline thing I’ve built for my main performance patch for November.

The patch itself isn’t too generally useful, but figured I’d share it anyways.

This is the main bit of the patch pulled out and put into demo form, so it exists a bit out of context.

What it does, is take audio from a comb, runs it into a pitch/onset detection section, to extract “notes” from that, which are then stored in a database to be used by another part of the patch (related to this process actually).

The comb audio is along the lines of this kind of thing:

For the purposes of this patch, I just take the realtime (modulo/transposed) output from the algorithm and send it to an oscillator for sonification.

Sounds cool.

combBass.zip (418.3 KB)

2 Likes

How is something like the transientslicer~ on this material? I tried it myself, but its a bit stale trying things on pre-recorded audio.

First, in retrospect after testing it on my nice monitors at home cycle~ is better for the bass sounds. Original post edited with the new version of the example.

Second, holy shit fluid.transientslice~ is crazy expensive. The helpfile on its own plays back all weird with crunchy/slow/zippery sounds. The CPU doesn’t appear to be high, so not sure what’s going on, but have a look at this:
https://we.tl/t-S9Lp9vPUE3

Is there something wrong with this?

(granted, vector size of 64 is pretty small, but as a point of reference, the whole of my performance patch I had running at 32 I/O without audio zippers until recently, and I’m trying to get it lean again so I can run it at 32 again. just the nearly empty fluid.transientslice~ helpfile sounds like a zipper fest on the same computer)

Alas, it is an expensive algorithm. At vs 32, I think the defaults are pushing it. Take the order down to 20, and the blocksize down to 512. On my machine this hovers around 10-13% with no zippering. YMMV.

1 Like

If that’s the case the defaults should be significantly lower.

I don’t think I’ve ever had a helpfile poo-poo-zipper on me like that, even the epic hirt.convolutionreverb~ helpfile doesn’t go crunchy, it just opens really slowly.

I honestly thought something was wrong with my computer or soundcard.