Python package for signal processing

Question

I am looking for a Python package to perform an efficient Constant Q Transform (ie using an FFT to speed up the process). I found a toolbox named CQ-NSGT/sliCQ Toolbox, but I get the following error:

File "build\bdist.win32\egg\nsgt\__init__.py", line 37, in <module>
File "build\bdist.win32\egg\nsgt\audio.py", line 7, in <module>
File "C:\Python27\lib\site-packages\scikits\audiolab\__init__.py", line 25, in <module>
    from pysndfile import formatinfo, sndfile
File "C:\Python27\lib\site-packages\scikits\audiolab\pysndfile\__init__.py", line 1, in <module>
    from _sndfile import Sndfile, Format, available_file_formats, \
File "numpy.pxd", line 30, in scikits.audiolab.pysndfile._sndfile (scikits\audiolab\pysndfile\_sndfile.c:9632)
ValueError: numpy.dtype does not appear to be the correct type object

There seems to be a problem either with Numpy (which I doubt) or more likely with scikit audiolab. Do you know where the problem comes from?

Answer 1

I use the CQT tools in yaafe: http://perso.telecom-paristech.fr/~essid/tp-yaafe-extension/features.html

Answer 2

Check out PYO, a C-implemented DSP module for python. Has mathematical operations on signals, should include Fourier transforms.

http://code.google.com/p/pyo/

Answer 3

PyMIR has implementations of some audio and music processing routines, such as applying the FFT from NumPy to wav or mp3 data, or applying the CQT (Constant Q transform) or DCT (Discrete Cosine Transform, built using NumPy) to the same kind of audio data:

• https://github.com/jsawruk/pymir

PyMIR is MIT licensed.