polyvoice 0.6.9

Speaker diarization for Rust — who spoke when. ONNX-powered: Silero VAD, WeSpeaker embeddings, Pyannote segmentation, K-means/AHC clustering, overlap detection.
Documentation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
# TODO — src/ahc

## Current

## Next

- [ ] Add property test: clustering is deterministic for identical inputs.
- [ ] Benchmark AHC vs scikit-learn reference on small synthetic data.

## Known Gaps

- No benchmark for AHC runtime scaling with embedding count.
- Elbow heuristic is not configurable from the public API.

## Deferred

- [ ] Explore single-linkage vs complete-linkage variants.