The project is made for educational purposes, as the homework of the course deep learning for audio processing.
checkpoints/ # Folder with model checkpoints
models/
├─ kws_baseline_model.pth # A baseline-based streaming model
├─ kws_final_model.pth # The final compressed streaming model
main.ipynb # A notebook with experiments and a report
stream.py # Example of streaming on Windows
stream_mac.py # The same streaming, but for Mac
speed up rate | compression rate | streaming |
---|---|---|
10.65 | 10.28 | + |
You need to run the code for streaming, it has already configured the loading of the desired model and hyperparameters are configured:
python stream.py
You may have to configure the input device for Windows or Linux, for Mac you can run the script below:
python stream_mac.py