Speaker Diarization Backend Programs

This dir contains the backend script 'speakDiar' for the BAS WebServices 
Speaker Diarization Service.
The script was designed to allow several algorithms to be selected by the user, 
but in time we found that most of these were inadequate in terms of quality, and we
decided to allow only one algorithm 'runDiarPyAnnote' designed by Fritz Seebauer.

The 'speakDiar' script is installed on the linux cluster (and the server) together with 
the MAUS development tree, but the Python program runDiarPyAnnote is only installed on the 
server under user 'tomcat#' (in a venv); see for instance 
/vdata/CLARIN/doc/BASWebServices/07_WebservicesBackendProgramsInstallation.txt
for details.

The helper 'runDiarPyAnnote' is a python program that is only installed on the server (linux11)
in a venv stored in the home of user tomcat# (to avoid Python packages conflicts).
It follows that speakDiar can only be executed by user tomcat# on linux11!

See CLARIN doc file BASWebServices/07_WebservicesBackendProgramsInstallation.txt for details about 
the installation of 'runDiarPyAnnote'.

Note: it is a good idea to increase the version of speakDiar every time the version of 
runDiarPyAnnote is increased.

Testing in user 'tomcat#' is a problem since user 'tomcat8' cannot write to /dev/stderr;
therefore we have copies of the scripts *.stdout that write to stdout instead; be sure to 
keep both versions functionally identical!

A typical offline test in user 'tomcat#' would be:
/homes/schiel/MAUS/TOOL/Pipeline/../SpeakDiar/speakDiar.stdout SIGNAL=/homes/schiel/MAUS/TOOL/EXAMPLES/ENGLISH/example-4-general-en.wav OUT=/tmp/Out.par OUTFORMAT=bpf speakMatch= speakNumber=0 minSilenceLength=200 minVoicedLength=200 preference=-2.97 TEXT=/homes/schiel/MAUS/TOOL/EXAMPLES/ENGLISH/example-4-general-en_maus.par
