This project uses the Data set provided my Toronto emotional speech set data.
Step 1: extract the features from the audio files using MFCC, LPC, and Chroma features.
Step 2: train the model using the extracted features.
Step 3: test the model using the extracted features.
Step 4: predict the emotion of the audio file.
Models used: Random Forest, Decision Tree, Naive Bayes, and Neural network MLP
Conclusion
After training the model, the accuracy of the models are:
Random Forest Accuracy: 99.643%
Decision Tree Accuracy: 90.357%
Naive Bayes Accuracy: 80.0%
MLP Accuracy: 99.286%