Numpy-based NIST SPH audio-file reader
Numpy-based NIST SPH audio-file reader. This is for use with NIST SPH audio-files, the most likely use being extracting the TEDLIUM_release2 audio into formats that standard tools can easily process.
Note that this library doesn’t require any external tools such as vox or gstreamer. It just loads the data into a numpy array and then lets you dump it back out to wave files.
Note that the library does not support files with embedded-shorten-* encodings, only the base ulaw encoding. You will need to convert such files with:
sph2pipe file.sph file-raw.sph
to allow them to be loaded.
from sphfile import SPHFile sph =SPHFile( 'TEDLIUM_release2/test/sph/JamesCameron_2010.sph' ) # Note that the following loads the whole file into ram print( sph.format ) # write out a wav file with content from 111.29 to 123.57 seconds sph.write_wav( 'test.wav', 111.29, 123.57 )
MIT License (c) 2017 Mike C. Fletcher
1.0.3 – Allow for other header keys during header format parsing
1.0.2 – Use signed integers for 2 and 4-byte sample_n_bytes
1.0.1 – Fix to allow for files that have non-sample-multiple bytes in the data section
1.0.0 – Initial release
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.