audio source (10.4gb) = http://www.repository.voxforge1.org/downloads/SpeechCorpus/Trunk/Audio/Main/16kHz_16bit/
Total repository size = 33mb (approx)
SVM classifier model saved with 97% test accuracy to 'finalised_model.sav' and Neural Network classifier model saved to 'neural_network.h5' using this dataset. This has been created for distributing and testing on different languages,accents and on different people to compare different models.
- librosa
- matplotlib
- pyaudio
- sklearn
- pandas
- scipy
- numpy
- wave
- keras
You can now also intall and use pitch as library function. *pip install pitch* .
### Steps 1. Download the full "ML_final" folder only, at a specific path (preferred C:\\ ) 2. Open 'svm.py' or 'neural_network.py' and set the path variable to your downloaded folder (path="C:\\\\ML_final\\\\") 3. Run the 'svm.py' file and test your results for rbf kernal SVM model or try 'neural_network.py' for neural network model
Atulya Kumar
Viren Baria
Bhargav Desai
Sanjeet Krishna
Parth Mehta
Licensed under the MIT license