/speech-analytics

Primary LanguagePython

speech-analytics

Documentation under construction...

Transcription and diarization with a remote microphone

Installation steps

conda create --name <name_your_env> python=3.10
conda activate <name_your_env>
conda install portaudio=19.6.0 pysoundfile=0.12.1 ffmpeg=4.3 -c conda-forge
pip install python-dotenv==1.0.1 termcolor==2.4.0 diart==0.7 whisper_timestamped==1.12.20 onnxruntime==1.17.1 pydub==0.25.1 simpleaudio==1.0.4 pyttsx3==2.90

Additional GPU setup (if applicable to your machine) to enable device="cuda". Otherwise, use device="cpu" instead in your decoder code.

pip install torch==1.13.1+cu117 torchaudio==0.13.1+cu117 --index-url https://download.pytorch.org/whl/cu117

Usage

  1. Check .env for the preferred settings.
  2. Run decoder.ipynb on the main server.
  3. Run source/client_mic.py on the client machine with a microphone.
  4. Check the output displayed on the decoder.