The audiogen package provides time domain audio processing tools using Python generators.

This makes some types of audio sample generation and processing pretty easy:

# mix 440 Hz and 445 Hz tones to get 5 Hz beating
beats = audiogen.mixer(
        (audiogen.tone(440), audiogen.tone(445)),
        [(constant(1), constant(1)),]
)

The actual samples won't be generated or stored in memory until they're actually consumed – for instance, when they're being written out to disk in a wave file:

with open("output.wav", "wb") as f:
    audiogen.sampler.write_wav(f, beats)

Generators' at-consumption-time computation also allows creating infinitely long output, e.g. to stream to speakers rather than a file on disk:

audiogen.sampler.write_wav(sys.stdout, audiogen.tone(440))

Or just:

audiogen.sampler.play(audiogen.tone(440))

You can also use standard generator tools, e.g. the itertools module, to handle audio data:

    beep_silence = itertools.chain(audiogen.beep(), audiogen.silence(0.5))
    infinite_beeps = itertools.cycle(beep_silence)

audiogen.sampler.write_wav(sys.stdout, infinite_beeps)

The easiest way to play directly to a soundcard output is to use the audiogen.sampler.play function, which will play your samples using PyAudio:

import audiogen
import itertools
import sys

audiogen.sampler.play(
    itertools.cycle(itertools.chain(audiogen.beep(), audiogen.silence(0.5)))
)

Alternatively, you could write your wave data to stdout, e.g. myaudio.py:

import audiogen
import itertools
import sys

audiogen.sampler.write_wav(
    sys.stdout,
    itertools.cycle(itertools.chain(audiogen.beep(), audiogen.silence(0.5)))
)

Then pipe to a command line audio player like Sox:

python myaudio.py | play -t wav -

Install with:

$ pip install audiogen
$ pip install --allow-external PyAudio --allow-unverified PyAudio PyAudio

PyAudio is optional. If it's not installed, playing audio via the soundcard with audiogen.sampler.play() will not be available, but generating Wave files – including for piping to an external player, like sox – will work just fine.

Note that to install PyAudio on Mac OS X, you'll need to first install portaudio:

$ brew install portaudio

Projects built with audiogen:

• koch, a Koch method Morse code trainer and audio generator
• afsk, a ham radio APRS/Bell-202 audio frequency shift keying encoder

Get the source and report any bugs on Github:

https://github.com/casebeer/audiogen

• 0.1.2 - Add band pass and band stop IIR filters; custom beep lengths from @jhobbs
• 0.1.1 - Fix multiplexing bug causing increase in pitch when using mixer() to produce more than one output channel
• 0.1.0 - Breaking changes: new arguments to tone(), play() blocks by default