Speaker Diarization: Who Spoke When in Audio & Video
Pricing
Pay per usage
Speaker diarization with sherpa-onnx: detect how many speakers there are, when each one talks, and optionally attach speaker labels to an existing transcript. Works on any audio or video.
Diarize speakers, or return supported capabilities.
https:// URL, or kv:/ from an earlier run (e.g. the audio16k.wav produced by video-audio-toolkit). Any format works — it is converted to 16 kHz mono internally.
https:// URL or kv:/ of a speech-to-text-whisper transcript.json. When set, the actor also returns the same words/segments with a speaker label attached.
{ num_speakers, cluster_threshold, min_duration_on, min_duration_off, threads }. Set num_speakers when you know the exact count; otherwise clustering uses cluster_threshold (default 0.5). Sharding: start_sec, duration_sec, overlap_sec — diarize only that window; turns are offset back to the original timeline, items past duration_sec flagged _overlap. Speaker IDs are only consistent within one run/shard.
{ signed_upload_url } to PUT diarization.json straight into your own storage.
on_success = keep only result artifacts in the run's key-value store. always = also drop them after the signed upload. off = keep everything for debugging.
![Europages [$6💰] B2B Company Data + Emails & VAT IDs avatar](https://images.apifyusercontent.com/wz09jINlk8Vak8_v_-pJg3T5XihmtQuLw4MkXX2tPyw/rs:fill:76:76/cb:1/aHR0cHM6Ly9hcGlmeS1pbWFnZS11cGxvYWRzLXByb2QuczMudXMtZWFzdC0xLmFtYXpvbmF3cy5jb20vbmduOHZSRVNUa3NleVdFNnAtYWN0b3ItZUtUWWlhVU9tbzFrZEQ0N0QtcHpFVENCNVpzWi1ldXJvcGFnZXMucG5n.webp){kind=avatar}
