ComfyUI Node
ElevenLabs Pro - Speech to Text
A ComfyUI node in ElevenLabs Pro/Audio with 13 inputs and 3 outputs.
ElevenLabs Pro - Speech to Text
- audio
- text
- language_code
- words_json
◄api_key►
◄modelscribe_v2►
◄language_code►
◄tag_audio_eventsfalse►
◄diarizefalse►
◄num_speakers0►
◄timestamps_granularityword►
◄seed0►
◄diarization_threshold0.50►
◄temperature0.00►
◄keyterms►
◄no_verbatimfalse►
CategoryElevenLabs Pro/Audio
Inputs (13)
| Name | Type | Default | Description |
|---|---|---|---|
| api_key | STRING | — | |
| audio | AUDIO | — | |
| model | COMBO | scribe_v2 | 2 options: scribe_v2, scribe_v1 |
| language_codeopt | STRING | ISO language code (e.g. 'en', 'es'). Empty = auto-detect. | |
| tag_audio_eventsopt | BOOLEAN | false | Annotate sounds like (laughter), (music) in transcript. |
| diarizeopt | BOOLEAN | false | Identify and label different speakers. |
| num_speakersopt | INT | 00–32 | Expected number of speakers. 0 = auto-detect. |
| timestamps_granularityopt | COMBO | word | Timestamp granularity. 'none' returns no timing info (matches API spec). |
| seedopt | INT | 00–2147483647 | Seed for reproducibility. 0 = random. Determinism not guaranteed. |
| diarization_thresholdopt | FLOAT | 0.500–1 | Threshold for speaker diarization. Higher = stricter separation. |
| temperatureopt | FLOAT | 0.000–1 | Sampling temperature. 0 = deterministic. |
| keytermsopt | STRING | Comma-separated key terms to bias recognition toward. | |
| no_verbatimopt | BOOLEAN | false | Remove filler words and stutters (scribe_v2 only). |
Outputs (3)
| Name | Type | Description |
|---|---|---|
| text | STRING | — |
| language_code | STRING | — |
| words_json | STRING | — |