Viseme Extraction

Extract mouth shapes from audio for lip-sync animation.

Usage

1. Learn visemes (one-time setup)

python viseme_learner.py

Add video files to data/ folder, then run this to create representative face images for each viseme.

2. Extract visemes from audio

python batch_viseme_extractor.py folder_with_wav_files

Creates JSON files with viseme timing data for Unity animation.

Output

JSON with keyframes for each viseme (1-14):

{
  "keyframes": [
    {"time": 0.0, "viseme": 14, "duration": 0.1},
    {"time": 0.1, "viseme": 6, "duration": 0.2}
  ]
}