Hello

This project is about audio processing.

Dataset comes from Sharif speech dataset.

You can run the demo with these 3 steps:

First install the requirements:

pip install -r requirements.txt

for training, we should extract some features from audio. for this purpose, we used egemaps version 2.

then you should perform the training of the neural network with audio data

python src/codes/trainData.py

This may take a few minutes! (up to 30 minutes)

finally you can test the model:

python src/codes/demo.py

This demo shows the probability for the mood and gender:

moods

"Angry", "Fearful", "Happy", "Sad", "Neutral", "Surprise"

gender

"Male", "Female"

Please add your voices to the src/resource/voices folder