diarizeapp/plda-speaker-clustering

PLDA Speaker Clustering & Whitening Transforms

Canonical repository for PLDA Speaker Clustering & Whitening Transforms maintained by the diarizeapp ecosystem.

Overview

NumPy array archives (.npz) containing Kaldi/VBx PLDA covariance matrices (Phi, Gamma, Sigma, mu) and x-vector whitening projections used for probabilistic speaker clustering.

Hardware Acceleration & Execution Providers

• Hexagon HTP NPU (Snapdragon X Elite): N/A — Statistical parameter matrices evaluated directly in pure Rust math routines; non-graph asset.
• DirectML / CUDA: Fully supported via standard ONNX Runtime.
• CPU (x86_64 / ARM NEON): Universal SIMD fallback.

Provenance & Integrity

Maintained by diarizeapp to guarantee persistent asset availability and prevent 404 dependency breakages.