LR-ASD vs PySceneDetect
Both in the video & clipping category. Side-by-side — pick the one that fits your stack tonight.
The 2025 state-of-the-art for 'which face is actually talking.' Fast, tiny, accurate.
- rating
- 4★
- tested
- —
- cost
- free
- install
- sidecar
- stars
- 109
- updated
- 1y ago
You're not building a pipeline yourself. This is a research model, not a product.
Finds every camera cut in your video automatically. Powers smart cropping + transitions.
- rating
- 4★
- tested
- ✓ loya-tested
- cost
- free
- install
- sidecar
- stars
- 4,736
- updated
- 4d ago
You only work with single-camera talking-head footage — scene detection isn't useful there.
why it matters · LR-ASD
LR-ASD is the newest open-source active speaker detection model (Springer IJCV 2025 paper). It tells your video pipeline which person in a multi-face frame is actually talking. Accuracy beats the older TalkNet approach and it's 23 times lighter — fast enough to run on every frame, not just samples. If you're building your own clipping or auto-crop pipeline and accuracy matters more than a pre-built library, this is the one to drop in. MIT, free, Python.
why it matters · PySceneDetect
PySceneDetect scans any video and spits out the timestamp of every hard cut — the moment the camera switches. For multi-cam podcasts, that's the boundary you need so your 9:16 crop follows the active speaker without drifting on stale frames. Used in podcast-clipper crop pipelines alongside face tracking — same library Loya's LYRC export pipeline relies on for scene work. Free, Python, actively maintained (commits this week).