Speech-Emotion-Recognition

Pytorch implementation of "3-D Convolutional Recurrent Neural Networks With Attention Model for Speech Emotion Recognition".

I follow the original tensorflow code and change the tensorflow parts to pytorch ones. Please reference the original github for more details.

Dependency:

• pytorch
• python_speech_features
• wave
• pickle
• sklearn

Demo

After download the IEMOCAP dataset:

python zscore.py
python ExtractMel.py
python model.py

---

or you can download the processed file, IEMOCAP.pkl

python model.py

---

The best valid_UA of this code is about 0.6619.

Reference

https://github.com/xuanjihe/speech-emotion-recognition