SyncDev model directory · Profile updated Oct 7, 2026
Speaker diarization from audio: speaker-labeled time segments, exclusive diarization and a separate handoff to transcription.
The specification score summarizes recorded features and limits. It is not a benchmark result.
Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.
Accepts
Produces
pyannote Community-1 is a speaker-diarization pipeline that produces speaker-labeled time segments from audio. It answers who spoke when; transcription is a separate stage.
The pipeline processes mono 16 kHz audio. The documented loader downmixes multichannel input and resamples other sample rates. Access to the checkpoint requires accepting the repository conditions and using a Hugging Face access token.
The output includes ordinary speaker diarization and an exclusive representation that helps align speaker turns with transcription timestamps. Speaker labels are assignments within a recording, not verified real-world identities.
Include overlap, short interjections, noisy rooms and speakers who sound similar. Measure diarization error and inspect the segments a transcription or meeting-notes system would consume. Incorrect speaker attribution can make an otherwise accurate transcript misleading.
The pipeline is released under CC-BY-4.0. Local processing and a hosted pyannote service have different access and operational requirements. Browse speaker diarization models to compare the task separately from speech recognition.
Compare the model on the work you need to complete.
pyannote/speaker-diarization-community-1 is listed for Automatic speech recognition, Speaker Diarization. It accepts audio and produces speaker labeled segments. Use the publisher’s input format and preprocessing for your chosen deployment.
Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.
The weights are downloadable under CC-BY-4.0. Check runtime support and size the complete model and working context for your expected concurrency.
These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.
NVIDIA Research
Nemotron 3 Diarization identifies who spoke when in audio, with streaming or offline operation and support for up to eight speakers.