Added server-side SAD support

Added _/server/scripts/setup_SAD.sh_ script which configures the SAD model files
Updated /server/kaldigrpc/config.py to include SAD model configuration classes
Added /server/kaldigrpc/segment.py to implement segmentation

Segmentation doesn't support online processing currently. It is possible that this feature will be used at the end of the pipeline on the whole wav.

Setup SAD model files cd server/scripts ./setup_SAD.sh
Build server (Optional example with Chime6 ASR model ) (./setup_chime6.sh) make build-server kaldi_model=/home/george/kaldi-grpc-server/server/chime6_model/ image_tag=kaldigrpc:chime6-latest
Run server by mounting your local directory which contains the .wav to be segmented docker run -p 1234:1234 -ti -v <local dir>:/tmp kaldigrpc:chime6-latest --port=1234 --max-workers=3
In another terminal enter container's bash docker ps docker exec -it <container-id> bash
Run segmentation process cd kaldigrpc python segment.py --model-dir=../SAD_model/ --wav=/tmp/<your wav file>

georgepar / kaldi-grpc-server