SimpleDiarizer
Diarization in five lines of code: a minimal pipeline wiring Silero VAD to SpeechBrain speaker embeddings (x-vector or ECAPA) with spectral or agglomerative clustering — no configuration required.
About
SimpleDiarizer (cvqluu/simple_diarizer, PyPI: simple-diarizer) is exactly what the name promises: the shortest path from an audio file to diarized segments. It chains three well-known open components — Silero VAD for speech detection, SpeechBrain's x-vector or ECAPA-TDNN for speaker embeddings, spectral clustering (from wq2012/SpectralCluster) or AHC to group them — behind a five-line API: pick an embedding model and a clustering method, call diarize(), get segments, optionally plot the diarized waveform. An experimental ESPnet ASR hook can add transcription. It exists for one audience: people who need 'good enough' who-spoke-when without learning a diarization framework. Honest expectations matter here: a third-party (VoxScribe) test measured ~20% DER on CALLHOME for this pipeline versus ~8% for pyannote community-1 — roughly two and a half times the error. That's the price of simplicity. The license is GPL-3.0, worth noting if you're shipping it inside a commercial product. Pick it for prototypes, teaching, and quick analyses; graduate to pyannote or Sortformer when accuracy starts to matter.
Specs
Context window
Pricing
Free quota
API access
Weights
Released
Benchmarks
- ~20% DER on CALLHOME vs ~8% for pyannote community-1 (third-party VoxScribe benchmark, unaudited)\n- No author-published benchmarks
Provider-reported unless a source says otherwise. Benchmarks are marketing-adjacent by nature — treat them as directional.
Use cases
Trust score — 5/10
Scored 5/10 (TIER metrics, metrics-v1). Evidence: real PyPI package (0.0.9) with documented pipeline and a demo; third-party benchmark measured ~20% CALLHOME DER. Strengths: genuinely the simplest diarization on-ramp, honest composition of reputable components (Silero + SpeechBrain). Docked for: (1) accuracy far below modern alternatives (~20% vs ~8% DER); (2) small single-author project with sporadic maintenance; (3) GPL-3.0 license constrains commercial embedding; (4) no peer-reviewed backing. License: GPL-3.0.
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 5.0/10
- Maintenance · recency, releases, issue load
- 3.0/10
- Openness · license + docs
- 7.0/10
- Manual review · sanity read + red-flag check
- 5.0/10
- · Simplest diarization on-ramp: Silero VAD + SpeechBrain x-vector/ECAPA + spectral/AHC clustering.
- · Third-party test: ~20% DER on CALLHOME vs ~8% for pyannote community-1.
- · Sporadic single-author maintenance; no author-published benchmarks.