How to play Wav sound samples from memory

Question

For my final year project in college I am working with Wav Files and Python and messing around with them. I would love to be able to play the sound samples from memory rather than write the sound samples out in to a WAV file before I can hear them.

I have been looking for weeks online and have found PyMedia, PySound, PyGame etc and none of them seem to work for me. Every single package gives me errors.

Are there other libraries I am missing that would help me do this ? Or am I just being stupid and can't get the other packages to work.

Exactly what I want to do is along the lines of this:

    #open file and get parameters
    wavfile = Wave.open("file.wav", "r")
    params = wfile.getparams()
    nframes = params[3]

    #get sound samples in a list
    samples = []
    for i in range(nframes):
        samples.append(wfile.readframes(1))

    playsound(samples)

    changedSamples = makeChangeTo(samples)

    playsound(changedSamples)

And I would like to be able to have this in a loop so I can edit and hear the edits while the program is still runnning without having to write the samples to a wav file before being able to hear it as that takes too long.

Any suggestions ? Cheers !

try SSDD or virtual RAM Disks? files, streaming interfaces can hide latencies that would cause playback/record gaps - which are expected with interpreted language with garbage collection like Python — f5r5e5d
You haven't given any information about the errors you're having getting any of these libraries working. pygame certainly can create a Sound object from 'from an OGG audio file or from an uncompressed WAV'. — Alex Taylor

Matthias Matthias · Accepted Answer · 2017-02-22T09:37:39

You should clearly separate those two concerns:

Reading/writing WAV files (or other audio files)
Playing/recording sounds

There are several questions and answers to both topics here on SO.

This is my personal (and of course biased) recommendation:

You should use NumPy to manipulate sounds, it's much easier than handling plain Python buffers. If for some reason you cannot use NumPy, you can still do all this, but it will be a bit more work.

For reading/writing sound files, I recommend the soundfile module (full disclosure: I'm a co-author).

For playing/recording sounds, I recommend the sounddevice module (full disclosure: I'm its main author).

When using those modules, your example would probably become something like this:

import soundfile as sf
import sounddevice as sd

samples, samplerate = sf.read('file.wav')
sd.play(samples, samplerate)
sd.wait()
changed_samples = make_change_to(samples)
sd.play(changed_samples, samplerate)
sd.wait()

If you work in an interactive Python prompt, you probably don't need the sd.wait() calls, you can just wait until the playback has finished. Or, if you get bored of listening to it, you can use:

sd.stop()

If you know that you will use the same sampling rate for some time, you can set it as default:

sd.default.samplerate = 48000

After that, you can drop the samplerate argument when using play():

sd.play(samples)

If you want to store the changed sound to a file, you can use something like this:

sf.write('changed_file.wav', changed_samples, samplerate)

How to play Wav sound samples from memory

1 Answers