Python real time audio mixing

Question

I'm using scapy and pyaudio to play a short length of sine wave every time a packet is sent or received, with the frequency of the sound indicating the sender address.

sniff(prn = makeSound)

In this code, the makeSound function takes a packet, dissects it, and plays a sound based on the sender address.

As far as I know, pyaudio has either blocking or callback mode. In neither of these modes can I play multiple sounds simultaneously.

I need a way to start a note, and have it immediately start being mixed into the audio stream, regardless of how many sine waves are already in the process of being played.

Manual threading seems like overkill for something like this. Is there some sample code showing how I would do this? — is8ac, Aug 22 '15 at 16:26
If threading seems like too much work, then how about Timer() — Rolf of Saxony, Aug 22 '15 at 16:47
What I eventually did was use `os.system('play --no-show-progress --null --channels 1 synth %s sine %f &' % ( 0.1, srcHz))`. The '&' at the end allows the shell to be non blocking. This is of course terribly kludgy, but it works. I will probably upgrade it to in-python threading as per Rolf of Saxony's suggestion. — is8ac, Aug 23 '15 at 15:22
Ha! I tested using sox to blast them onto the command line but decided that you wouldn't accept that as an answer, as whilst being practical it's not very pythonic! :) — Rolf of Saxony, Aug 23 '15 at 15:30

score 0 · Accepted Answer · answered Aug 22 '15 at 16:25

You need to perform the mixing yourself. See the following example-program on how to mix several voices into one single-channel output.

 import struct
 import math
 import pyaudio
 from itertools import count

 pa = pyaudio.PyAudio()

 FORMAT = pyaudio.paFloat32
 CHANNELS = 1
 RATE = 44100

 OUTPUT_BLOCK_TIME = 0.05
 OUTPUT_FRAMES_PER_BLOCK = int(RATE*OUTPUT_BLOCK_TIME)


 def sine_gen():
     time = 0
     format = "%df" % OUTPUT_FRAMES_PER_BLOCK
     voices = []
     voices.append(lambda sampletime: math.sin(sampletime * math.pi * 2 * 440.0))

     for frame in count():
         block = []
         for i in xrange(OUTPUT_FRAMES_PER_BLOCK):
             sampletime = time + (float(i) / OUTPUT_FRAMES_PER_BLOCK) * OUTPUT_BLOCK_TIME
             sample = sum(voice(sampletime) for voice in voices) / len(voices)
             block.append(sample)
         yield struct.pack(format, *block)
         time += OUTPUT_BLOCK_TIME
         if frame == 20:
             voices.append(
                 lambda sampletime: math.sin(sampletime * math.pi * 2 * 880.0)
             )

 stream = pa.open(format=FORMAT,
     channels=CHANNELS, rate=RATE, output=1)

 for i, block in enumerate(sine_gen()):
     stream.write(block)

Python real time audio mixing

1 Answers1