Personal Blog

Reading and Writing WAV files with compression in Python using Pywav

Testing voice applications involve handling RTP streams. Most of the IVR applications use PCMA/PCMU codecs to stream voice over RTP. PyWav is a Python library which helps you to convert those RAW streams with PCMU/PCMA encodings into a WAV file. The library is similar to PyWave, but the limitation with PyWave is, it will not support compression.

What is PyWav?

When building a SIP test tool, I have to convert RTP audio streams with PCMA/PCMU encodings to WAV files for validation. PyWav is the result of that activity. Using PyWav you can write RAW streams into a playable WAV file supporting PCM/PCMA/PCMU compressions. Similarly, you can even read a WAV and get the RAW data from that WAV file to send it over RTP stream.

Reading a WAV file

Below sample code shows how to read a WAV file:

wave_read = pywav.WavRead("<filename to read>")
# print parameters like number of channels, sample rate, bits per sample, audio format etc
# Audio format 1 = PCM (without compression)
# Audio format 6 = PCMA (with A-law compression)
# Audio format 7 = PCMU (with mu-law compression)

Following are the list of parameters that are extracted from the WAV file:

  • Number of channels
  • Sample rate
  • Byte rate
  • Bytes per sample
  • Bits per sample
  • Sample length
  • Data (raw data)
  • Audio format
  • Extra params

To know more about WAV files and formats refer this article.

Writing a WAV file

You can write the RAW data captured from the RTP stream into a WAV file using the following sample:

# first parameter is the file name to write the wave data
# second parameter is the number of channels, the value can be 1 (mono) or 2 (stereo)
# third parameter is the sample rate, 8000 samples per second
# fourth paramaer is the bits per sample, 8 bits per sample
# fifth parameter is the audio format, 1 means PCM with no compression.
wave_write = pywav.WavWrite("<filename to write>", 1, 8000, 8, 1)
# raw_data is the byte array. Write can be done only once for now.
# Incremental write will be implemented later
wave_write.write(<raw data>)
# close the file stream and save the file


You can install this as a pip package using pip3 install pywav. The source code of this project is available on Github.

Background photo created by kjpargeter -

comments powered by Disqus