Audio Analyzer for vup
This repo is for bmmaudioanalyzer project.
A thrift based server is provoded by the project.
First, spleeter to perform separation of audio files to 2 stems. Ex. Vocals (singing voice) / accompaniment separation (2 stems).
Second, inaSpeechSegmenter splits audio signals into homogeneous zones of speech, music and noise.
install python=3.7.2
install cudatoolkit=10.0.130
install cudatoolkit-dev=10.0
install cudnn=7.6.0
install ffmpeg
pip3 install keras==2.3.1 librosa==0.7.2 numba==0.48.0 matplotlib==3.2.0
pip3 install thrift ffmpeg-python sidekit pyannote pandas
pip3 install tensorflow-gpu==1.15.2
After all dependencies are successfully installed, you should put some audio files in the audios folder.
use follow:
cd main/python
sh ./run_server.sh
sh ./run_client.sh