I am developing subtitling software in C++ and I use pocketsphinx. For
speaker diarization / segmentation, CMU Sphinx seems to recommend LIUM,
which indeed works great, but it comes in a java binary. Thus I need to
require that all users have JRE installed on their machine (or ship it
with my app), which I'd rather not.
The alternative seems to be ALIZE-LIA/RAL, which does have a C API. I do
understand some French, but the documentation is not exhaustive and I
have not yet figured out how to use it for speaker diarization.
Does anyone have experience with ALIZE speaker diarization; how did it
work for you? Is it worth the effort to understand the API?
Or do you have tips on any other software?
Grateful for any tips,
Emil
If you would like to refer to this comment somewhere else in this project, copy and paste the following link:
The alternative seems to be ALIZE-LIA/RAL, which does have a C API. I do
understand some French, but the documentation is not exhaustive and I
have not yet figured out how to use it for speaker diarization.
Alize is for speaker identification only, it doesn't have support for diarization.
I am developing subtitling software in C++ and I use pocketsphinx. For
speaker diarization / segmentation, CMU Sphinx seems to recommend LIUM,
which indeed works great, but it comes in a java binary. Thus I need to
require that all users have JRE installed on their machine (or ship it
with my app), which I'd rather not.
The alternative seems to be ALIZE-LIA/RAL, which does have a C API. I do
understand some French, but the documentation is not exhaustive and I
have not yet figured out how to use it for speaker diarization.
Does anyone have experience with ALIZE speaker diarization; how did it
work for you? Is it worth the effort to understand the API?
Or do you have tips on any other software?
Grateful for any tips,
Emil
Alize is for speaker identification only, it doesn't have support for diarization.
There is
http://shout-toolkit.sourceforge.net/classShout__Cluster.html
I'm not aware of other popular implementations.
Ok, I was wrong, they recently introduced spkseg which should do diarization:
http://alize.univ-avignon.fr/svn/LIA_RAL/trunk/LIA_SpkSeg/
Hello,
Alize speaker verification tool kit is the best compared to HTK and Sphinx(modified for speaker verification).