In real-time!
Start training by playing the desired type of music while pressing and holding down the corresponding button. Collect at least 150 examples of each as well as some examples of room ambience while pressing and holding down "Noise". When you're done hit the "Train" button.
Once the model is trained, begin testing by hitting the "Listen" button. The app will then start listening to incoming audio and report the confidence values for each emotion in the table below.
If you're not happy with how the model is performing, you can go back and add some more examples, then retrain it.
| Emotion | Confidence |
|---|---|
| Happy | 0.0 |
| Sad | 0.0 |
| Angry | 0.0 |
Here you can save your trained model for later use. This will generate two files, one for the topology and one for the weights of the model.
Load the topology and weights separately then hit the "Listen" button to pick up from where you left off.