26.08.19 : speakDiar 0.1 : dummy script to test BAS WebServices interface
11.10.19 : speakDiar 0.5 : first alpha version using LIUM with pre-trained models
02.11.19 : speakDiar 0.6 : fixed annotConv output for emuDB and exb
03.02.20 : speakDiar 0.7 : added options TEXT, speakNumber, speakMatch, SDType
28.06.20 : speakDiar 0.9 : added SD method 'DiarXvector' that points to 'runDiarXvector',
           added option 'maxSpeakDiar', added passing VAD tier from input BPF to 'DiarXvector',
           set default SDType=DiarXvector.
03.07.20 : speakDiar 1.0 : added option 'vadMergeGap' for SD method DiarXvector
... here we lost track for some versions; basically we did a complete redesign based 
    on the pyannote libary, added a speaker clustering and did a lot of testing; finally we
    found that this new development is so much better than the existing two SD methodes, that
    we decided to bin them
28.02.21 : speakDiar 2.2 : new version of runDiarPyAnnote 1.0 (first beta) that restricts number of threads to 10 (was 40)
25.11.21 : speakDiar 2.3 : runDiarPyAnnote is not longer installed system-wide but only into a venv stored in the 
           home of tomcat8 on linux11; speakDiar now calls simply 'runDiarPyAnnote' which is a bash script found in /usr/local/bin
           that activates the venv, executes the runDiarPyAnnote in the venv and deactivates the venv.
           It follows that speakDiar can only be executed on linux11 with user tomcat8!
18.12.21 : speakDiar 2.4 : bug fix: the alignemnt of word segments to speaker segments was faulty causing mislalignments of words
18.02.22 : speakDiar 2.5 : bug fix: the last segment exceeds the length of the signal file by one sample
