K Ring Technologies Ltd

Lua Audio

So a library for Lua to do some kind of audio. I don’t mean multimedia loading as this could be simple os.execute(...) call wrappers, with perhaps "...&", but a bit of that is still an idea to add later. It’s more sort of chip-tune kind of stuff. So os.execute("<c-program> <arg> ... | pw-play --channels=1 -&") for a more exciting audio generation experience. As each note would then become threaded, which for small sound experimentation would be fine.

The sample rate of 48000 does seem high for chip-tunes, but is a feature of pw-play and not and sample generator program. The program just has to emit signed 16 bit samples by default for a length and then close the output stream. This would then make the thing portable across platforms as standard C code, with minor scripting changes for other languages and “Windoze” in particular.

It then just becomes setting some standard argument conventions and a heuristic based on computer history is useful here.

  1. Length is seconds. A classic it can be controlled 1st step.
  2. Volume between 0 and 1 for the next level of control of stimulation.
  3. Frequency with 0 being a 440Hz and +1/-1 going up and down one semitone.
  4. Filter also a frequency using the same note scaling.
  5. Vol.drift in 6dB per 100% length.
  6. Freq.drift in semitones per 100% length.
  7. Filt.drift also in semitones per 100% length.
  8. Mod.volume well parameter 8 kind of starts as parameter 2 of a modulator.

This then makes for an easy as many as needed with a 1 second beep for no parameters, and all modulators starting off with parameter clone, but zero volume. So the 10th parameter just applies some frequency matched PM modulation, with cloned parameters from the carrier oscillation.

It’s sounds like an idea. Well, more than an idea. The C code is written. The test.raw file has been made. So upto 19 parameters, and some sensible max length of 16 seconds. Next up is a Lua wrapper.

I decided on just a single pole filter with exponential FM, so that tremolo would be easy. It does however cause a detune based on the modulation. This could be stopped by using phase modulation, but spectrally that would be brighter. I also think the frequency dependence to modulation, while not to the filter cut off is more interesting, and more natural instrument baffle sounding. Also sightly easier, and what’s to stop audio-too?

1
2
3
4
5

     _            _                  _
  __| | ___  _ __(_)___   _ ____   _(_)_ __ ___
 / _` |/ _ \| '__| / __| | '_ \ \ / / | '_ ` _ \
| (_| | (_) | |  | \__ \_| | | \ V /| | | | | | |
 \__,_|\___/|_|  |_|___(_)_| |_|\_/ |_|_| |_| |_|

Tags