Detection uses two settings per command: Threshold and Predictions in a row. Specify these values according to the needs of your project. A command is detected when its probability stays above the threshold for that many consecutive predictions.
The settings table lists each of your commands together with two system entries:
- OTHER — Audio that does not match any command
- SILENCE — Background or no speech
To adjust detection:
- If the model triggers when you did not say the command, increase the threshold or the number of predictions in a row.
- If the model does not trigger when you say the command, decrease the threshold or the number of predictions in a row.