DSP Robotics Support

There are cases where it would be nice to interpret MIDI myself and do my own voice scheduling. There is practically no way to build a voices object myself unless I can convert what I really want to do into MIDI that the system will already interpret correctly. This would mean my only sane way to do my own voice scheduling is to skip the Voices step and generate a series of Float arrays and use a Float Array to Poly on each. I'd prefer to use the Voices type that seems to be standardized on, but being able to output Poly IOs from Ruby (Poly and PolyBoolean) would also make this easier.

My primary goal is to make synthesizers that work with the extended precision of a Haken Continuum and correctly handle the Voice per Channel data, but I suspect there will be other uses.

(Also note I think ruby IO of Busses would be very handy too.)

There's no chance to fulfill your request. Streams work on a per-sample basis, while Ruby is much slower and relies on a framebuffer, with the size of that buffer dependent on the audio output driver (e.g. ~2 MB for DirectSound, variable sizes with ASIO)