Greetings,

I'm working on a live beat slicer, similar to MLR for the Monome (but in this case, for the Yamaha Tenori-on).

Instead of simply dividing my sample layers by 16, I am trying to do transient analysis of the sample, to determine the 16 best possible positions to start my sample from.

To do this, I am using a variation of saturno's [findhits~] patch, available in this thread:

http://puredata.hurleur.com/viewtopic.php?pid=14201#p14201

Now, this patch works great for defining all possible transient slice points in a given sample. And using a [thresh( message allows me to define a given threshold, with which I can bring down the range of possible slice points considerably.

However, I need to bring the number of possible transient slice points down to only 16, every time. Can anyone offer me some advice on this best way to approach this?

Many thanks!

• Peter

Maybe you could try taking a sample's length and finding the exact positions of a perfect division of 16. Then, analyse it and match the detections that are closest to those divisions. Maybe even have a threshold that says if it's too far out timing-wise, it should be tossed.

Now that I think about it, you could even divide it initially into 32 or 64, depending on the material, and if it's closer to a subdivision that doesn't conform to the division of 16, toss it; otherwise, keep it. This will be an issue if the beat swings really hard, though, but you could possibly at a parameter for that in the pre-detection analysis, too.