New params for beatmatching, updates for bandpass, crush, coarse

[bgold-cosmos]

I’ve added the new parameters “matchcps” and “stretchTo” that simplify beatmatching or automatically stretching samples to specified lengths with Tidal. These override the “speed “ and “accelerate” parameters.

I’ve also added some extra functionality to the bandpass filter and the quantizers “coarse” and “crush”. Using a negative frequency in “bandf” will switch the bandpass to a bandnotch filter. Using a negative value for “crush” will use a nonlinear quantization scheme that drops the effective noise floor a bit and handles quiet samples at low bit depths a bit better. And using a negative value for “coarse” will use different “resampling” method that has a bit less aliasing and effectively low-passes the sound.

Finally, to try and future-proof Dirt a little, I also made a slight change to the way the server handles messages with too many parameters. Instead of logging a generic error and playing no sound, it will now log a shorter error message but play the sound if the rest of the parameters are valid.

This updated Dirt should be fully backward compatible with prior versions of Tidal.

[yaxu]

Hey Ben Sorry have been travelling and haven't caught up with the forum.. Looks great, wondering though about whether stretchTo and matchcps could be turned into a single parameter, e.g. unit. speed "1" |+| unit "cycle" for matchcps, speed "1" |+| unit "sec" for seconds and speed "1" |+| unit "rate" for the current behaviour. We should also try to get Tidal to pass through the current cps for us as a hidden parameter, so that speed can be expressed as a ratio of the current cps Also if you think your filters are better than why not just replace the existing ones rather than using negative numbers?

[bgold-cosmos]

OK, I hadn't thought of having a modifier parameter like "unit". I think that would be elegant - passing cps as a hidden parameter might be useful for other situations too. I'm not entirely sure how to get Tidal to stamp the OSC messages with it, but I can take a look and see if I can figure it out.

I don't really think the new crush and coarse are "better"; often I prefer the sound of the old ones. But after a while I've found that for some samples they just don't work very well, and I thought an alternative algorithm might make them more useful or appealing to people.

[yaxu]

Hey Ben

Ok I've pushed a change so you should be able to switch cpsStamp to True in Dirt.hs and start receiving the current cps with every message, is that helpful?

[bgold-cosmos]

Very cool - I was pretty stumped looking at Stream.hs but you've done the hard stuff for me. I've successfully tested communication of cps from Tidal to Dirt, so I should be able to hack the rest up in Dirt soon.

[kindohm]

Makefile was added to .gitignore, that's why it isn't included.

[bgold-cosmos]

OK, here's the new version with "unit". You can look at the latest commit for detailed comments, but in short now you can do something like

cps 0.75
d1 $ sound "[breaks125, breaks125/2, breaks165, rave:6/4]" |+| unit "c"

and Wham! automatic beat-matching. It's a lot of fun.