A Python script for Windows that classifies voice commands and runs shell commands.
Confusion matrix of the trained model
The model was trained on four voice commands:
| Voice command | Shell command |
|---|---|
| python | start /B python -m idlelib |
| notepad | start /B notepad |
| terminal | start /B wt |
| documents | start /B explorer "%userprofile%\Documents" |
The model was trained on my own voice data, so it might not perform well with your voice. You can use training/sample_recorder.py and training/trainer.py to record your own samples and retrain the model.
To use the launcher run main.py then press ⊞ Win+G to issue a voice command.
Further information is available in Slovene language in the report.