How andreaunali/signbridge works
SignBridge turns either typed English or recorded speech into an animated ASL gloss sequence. The runtime workflow is input selection, optional Whisper transcription, spaCy-based gloss conversion (including auxiliary filtering, lemmatization, and fingerspelling), pose lookup, then animation and video output. The shipped Pose directory is the precomputed sign dataset; README describes its one-time preparation, not a runtime stage. Animation internals are separated where the excerpts show distinct responsibilities; unsampled dataset-preparation implementation is not mapped.
Main entry points: main.py CLI workflow · animation_lib.py Animation director · INPUT_AUDIO Audio files · Pose Pose JSON dataset
Components 15
Input and language
- CLI workflow
main.py - Audio files
INPUT_AUDIO - Whisper transcription
- Typed text
main.py - Gloss conversion
main.py - spaCy NLP
Sign pose data
- Pose JSON dataset
Pose
Animation and output
- Animation director
animation_lib.py - Pose loader
animation_lib.py - Pose interpolation
animation_lib.py - Sign transitions
animation_lib.py - Avatar renderer
animation_lib.py - MP4 output
animation_lib.py
Other components
- User
- OpenCV video I/O
Connections 18
- User → CLI workflow: chooses input
main.py - User → Audio files: provides audio
- Audio files → CLI workflow: selects file
main.py - CLI workflow → Whisper transcription: transcribes with
main.py - User → Typed text: enters sentence
- Typed text → CLI workflow: reads input
main.py - CLI workflow → Gloss conversion: converts text
main.py - Gloss conversion → spaCy NLP: parses tokens
main.py - Gloss conversion → Pose JSON dataset: checks glosses
main.py - CLI workflow → Animation director: renders sequence
main.py - Animation director → Pose loader: loads poses
animation_lib.py - Pose loader → Pose JSON dataset: reads JSON
animation_lib.py - Pose loader → Pose interpolation: interpolates frames
animation_lib.py - Animation director → Sign transitions: creates bridges
animation_lib.py - Animation director → Avatar renderer: renders frames
animation_lib.py - Avatar renderer → OpenCV video I/O: draws with
animation_lib.py - Animation director → MP4 output: writes video
animation_lib.py - MP4 output → OpenCV video I/O: uses writer
animation_lib.py