Recognize Audio Emotion.
Eight emotions: neutral, calm, happy, sad, angry, fearful, disgust and surprised.
Create "AudioData", "Output/logs", "Output/mlmodels" folders. Download RAVDESS dataset and put it in "AudioData" folder.
- Replace some Turicreate source files by "Source/Turicreate_Fix/*.py"
- Train model using
turicreate_classifier.py
. - run "iOSSample/AudioEmotion.xcodeproj" on iOS 12(or higher). You can also run
predict_audio_emotion.py
on your Mac/PC.
- run
extract_audio_feature.py
- run
train_audio.py
- run
predict_audio_emotion.py