Skip to content
Recognize Audio Emotion.
Branch: master
Clone or download
yulingtianxia 修改编译问题
update Swift 5
Latest commit 08587d6 Apr 17, 2019
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
.idea
Source
iOSSample
.gitignore
LICENSE
README.md

README.md

AudioEmotion

Recognize Audio Emotion.

Eight emotions: neutral, calm, happy, sad, angry, fearful, disgust and surprised.

Article

音频情绪识别

Usage

Create "AudioData", "Output/logs", "Output/mlmodels" folders. Download RAVDESS dataset and put it in "AudioData" folder.

Turicreate

  1. Replace some Turicreate source files by "Source/Turicreate_Fix/*.py"
  2. Train model using turicreate_classifier.py.
  3. run "iOSSample/AudioEmotion.xcodeproj" on iOS 12(or higher). You can also run predict_audio_emotion.py on your Mac/PC.

Keras

  1. run extract_audio_feature.py
  2. run train_audio.py
  3. run predict_audio_emotion.py
You can’t perform that action at this time.