This project was migrated from https://code.google.com/p/py-audio on May 17, 2015
Keywords: Voice, Codec, Speex, Python, RtAudio, Resample, Device
Members: voipresearcher, theintencity
Links: Support, Download binary modules for Python 2.6 x86_64 on Mac OS X 10.6.4, Oct 2011, 114 KB, download count 315
License: GNU Lesser GPL
Others: starred by 8 users
This project builds the Python bindings for audio device interface, speex codecs library and text-to-speech library for real-time voice communication. In particular it provides Python wrapper modules for the following projects:
- The RtAudio Project 4.0.8
- The Speex Project 1.2rc1
- The Flite Project 1.4
Instructions to build on Unix-based platforms including Mac OS X and Linux using Python 2.6.
- Checkout the latest version of this py-audio project from git.
$ git clone https://github.com/theintencity/py-audio.git
$ cd py-audio
- Download source, uncompress and build rtaudio
$ tar -zxvf rtaudio-4.0.8.tar.gz
$ cd rtaudio-4.0.8
$ ./configure
$ make
This will create the static and dynamic libraries in rtaudio/. We are interested only in static library librtaudio.a.
On Linux, if it complains about not found file, please change SHAREDLIB to SHARED at two places in its Makefile, and do make again.
- Download source, uncompress and build speex.
$ tar -zxvf speex-1.2rc1.tar.gz
$ cd speex-1.2rc1
$ ./configure
$ make
This will create the static and dynamic libraries in libspeex/.libs/. We are interested only in static libraries libspeex.a and libspeexdsp.a.
On Linux, if it complains about missing PIC option, change your configure command to ./configure CFLAGS=-fPIC above and do configure and make again.
- Download source, uncompress and build flite.
$ bunzip2 flite-1.4-release.tar.bz2
$ tar -xvf flite-1.4-release.tar
$ cd flite-1.4-release
$ ./configure
$ make
This will create the static libraries in build/i386-darwin9.6.1/ or similar.
- Create a new
setup.pyfile in py-audio directory. On Mac OS X use the following content
from distutils.core import setup, Extension
module1 = Extension('audiodev', sources = ['audiodev.cpp'],
include_dirs = ['rtaudio-4.0.8'],
extra_link_args = ['rtaudio-4.0.8/librtaudio.a', '-framework', 'CoreAudio'])
module2 = Extension('audiospeex', sources = ['audiospeex.cpp'],
include_dirs = ['speex-1.2rc1/include'],
extra_link_args = ['speex-1.2rc1/libspeex/.libs/libspeex.a',
'speex-1.2rc1/libspeex/.libs/libspeexdsp.a'])
import os
libdir = 'flite-1.4-release/build/%s-%s%s/lib'%(os.uname()[-1],
os.uname()[0].lower(), os.uname()[2])
module3 = Extension('audiotts', sources = ['audiotts.cpp'],
include_dirs = ['flite-1.4-release/include'],
extra_link_args = ['%s/lib%s.a'%(libdir, x) for x in (
'flite_cmu_us_kal', 'flite_cmu_us_awb', 'flite_cmu_us_rms',
'flite_cmu_us_slt', 'flite_usenglish', 'flite_cmulex', 'flite')])
setup (name = 'PackageName', version = '1.1',
description = 'audio device and codecs module',
ext_modules = [module1, module2, module3])
On Linux, use the following content
from distutils.core import setup, Extension
module1 = Extension('audiodev', sources = ['audiodev.cpp'],
include_dirs = ['rtaudio-4.0.8'],
libraries = ['pthread', 'asound'], extra_link_args = ['rtaudio-4.0.8/librtaudio.a'])
module2 = Extension('audiospeex', sources = ['audiospeex.cpp'],
include_dirs = ['speex-1.2rc1/include'],
library_dirs = ['speex-1.2rc1/libspeex/.libs'],
libraries = ['speex', 'speexdsp'], extra_link_args = ['-fPIC'])
import os
libdir = 'flite-1.4-release/build/%s-%s%s'%(os.uname()[-1],
os.uname()[0].lower(), os.uname()[2])
module3 = Extension('audiotts', sources = ['audiotts.cpp'],
include_dirs = ['flite-1.4-release/include'],
library_dirs = [libdir],
libraries = ['flite_cmu_us_kal', 'flite_cmu_us_awb', 'flite_cmu_us_rms',
'flite_cmu_us_slt', 'flite_usenglish', 'flite_cmulex', 'flite'],
extra_link_args = ['-fPIC'])
setup (name = 'PackageName', version = '1.0',
description = 'audio device and codecs module',
ext_modules = [module1, module2, module3])
On Linux, If a different sound device was used, then change the library from asound to that.
- Compile the Python bindings. On Mac OS X use the following command after replacing the arch flag to i386 or ppc if applicable instead of x86_64.
$ ARCHFLAGS="-arch x86_64" python setup.py -v build
On Linux use the following command.
$ python setup.py -v build
If you have multiple versions of Python installed, you may want to replace python to python2.5 or python2.6 as applicable in the above command. Note that python2.5 does not support x86_64 on Mac OS X.
- Copy the generated modules to the current directory.
$ cp build/lib.*/audio*.so .
This will copy audiospeex.so, audiodev.so and audiotts.so to current py-audio directory.
An example file named test.py is available to allow you to test in loopback mode. Once you start the application, it opens the audio device, and for every captured frame it performs resampling and encoding/decoding, and finally plays back to the audio device.
$ python test.py
<ctrl-C to terminate>
An example file named tts.py is available to allow you to test text-to-speech feature. You can start it by supplying the text on command line.
$ python tts.py hello, how are you?
How to use this in SIP-RTMP gateway?
You only need to build audiospeex.so for transcoding. Place this module in your PYTHONPATH or in rtmplite directory so that it is accessible to siprtmp.py application. Then start siprtmp.py as shown on that project page with -d option, and it should detect the audiospeex module to enable transcoding.
The API is straightforward. You can use the built-in help command in Python to know the details. Moreover the example test.py illustrates how it works.
>>> import audiodev, audiospeex, audiotts
>>> help(audiodev)
>>> help(audiospeex)
>>> help(audiotts)