@spluta I want to investigate the sine wave to band-limited sawtooth example a bit deeper. Is the Python code and the training dataset available? I can't seem to find it in the RTNeural_Plugin folder.

If I'm not mistaken, Miller uses Fedora. Also, he's about to release Pd-0.56, so it's a good moment to inform him about it. You should probably write to Pd's mailing list which he reads.

Andy Farnell has simulated the sound of water in Pd. Check his book "Designing Sound".

Huge thanks to the team for taking care of this!

For the life of me, I can't figure out how to use the colors message! If I send colors #ff00ff or something similar to pd, I get a bad arguments to colors message.

You're first mentioning police sirens, and at the end of your post you mention that the Arduino will play through speakers. What is the actual sound source you want to use? If it's some digital synthesis that is played through speakers, then you're much better off with Pd than with the Arduino. If you indeed want to control police sirens in some way, I guess the Arduino could do it, but I have no idea how police sirens work (I tend to avoid them, you see ).

Doesn't [tabread4~] apply cubic interpolation? If you store a ramp to a table, you could use that with [line~] to read through it, couldn't you?

You can check the [serial_print] abstraction from here https://github.com/alexdrymonitis/Arduino_Pd
If you print your values in the Arduino code properly (see the help sketch), then in Pd you can use this abstraction with a [send] and get your values with a [receive]. Check the exampes, they should be straight forward.

You can also check the audiolab library (available on deken; Help -> Find externals). It's a collection of vanilla abstractions. Its pitch shifter might be a bit CPU hungry as it is FFT based, but it's pretty good.

Since your source is a .wav file, why don't you store the file to an array and use [tabplay] or [tabread4~] to read it in a loop? I'm pretty sure that a [vline~] with a long list will be way more CPU hungry, let alone the complexity needed for something like this.