podcastgen is a podcast generator written for Radio Revolt.
It takes a WAV file as input, and outputs a 16 bit PCM WAV file in which the music has been removed (apart from a few seconds at the beginning and end of each speech section).
The method used is based on the Modified Low Energy Ratio, described in "A Fast and Robust Speech/Music Discrimination Approach" by Wang, Gao and Ying (2003). The goal is to implement the full method described in "Automatic speech/music discrimination in audio files" by Ericsson (2009), see http://www.speech.kth.se/prod/publications/files/3437.pdf.
Syntax: podcastgen <options> <input file>
-h, --help Display this help text
-v Verbose output
-C <float> Set the Low Energy Coefficient
-T <float> Set the Upper Music Threshold
--very-verbose Very(!) verbose output
--no-intro Remove music at start of file
cd build
cmake ..
make
make install