To run this, you need to:
pip3 install -r requirements.txt
It is already trained, you can use test.py to test your voice.
You can:
- Tweak the model parameters ( or the whole model ) in
ser.py. - Add more data to
datafolder in condition that the audio samples are converted to 16000Hz sample rate and mono channel,convert_wavs.pydoes that. - Editing the emotions specified in
utils.pyinAVAILABLE_EMOTIONSconstant.
When you modified anything, you can run ser.py to retrain the model.
If you want to only use this efficiently, definitely check Emotion Recognition using Speech repository which ease this process a lot.