Realtime Commit Strategies · Speech To Text · elevenlabs/skills · Skills Docs

Realtime Commit Strategies · Speech To Text · elevenlabs/skills · Skills Docs

Type	Description
Partial	Interim “best guess” results that update frequently as audio is processed. Use for live feedback (showing text as the user speaks), but don’t save these - they may change.
Committed	Final, stable results after a commit occurs. Use these as the source of truth for your application - they won’t change.
Committed with Timestamps	Same as committed, but includes word-level timing data for subtitles, karaoke, or lip-sync.

async with client.speech_to_text.realtime.connect(
    model_id="scribe_v2_realtime",
) as connection:
    # Send audio
    await connection.send({
        "audio_base_64": audio_base_64,
        "sample_rate": 16000,
    })

    # Commit when ready (e.g., pause in speech, end of sentence)
    await connection.commit()

const connection = await client.speechToText.realtime.connect({
  modelId: "scribe_v2_realtime",
});

// Send audio
connection.send({
  audioBase64: audioBase64,
  sampleRate: 16000,
});

// Commit when ready
connection.commit();

import { useScribe, CommitStrategy } from "@elevenlabs/react";

const scribe = useScribe({
  modelId: "scribe_v2_realtime",
  commitStrategy: CommitStrategy.VAD,
  // Optional VAD tuning:
  vadSilenceThresholdSecs: 1.5,    // Silence duration before commit
  vadThreshold: 0.4,               // Speech detection sensitivity (0-1)
  minSpeechDurationMs: 100,        // Minimum speech length required
  minSilenceDurationMs: 100,       // Minimum silence length required
});

const connection = await client.speechToText.realtime.connect({
  modelId: "scribe_v2_realtime",
  vad: {
    silenceThresholdSecs: 1.5,    // Silence duration before commit
    threshold: 0.4,               // Speech detection sensitivity (0-1)
    minSpeechDurationMs: 100,     // Minimum speech length required
    minSilenceDurationMs: 100,    // Minimum silence length required
  },
});

Parameter	Description	Default
`silenceThresholdSecs`	Seconds of silence before auto-commit	1.5
`threshold`	Speech detection sensitivity (lower = more sensitive)	0.4
`minSpeechDurationMs`	Ignore speech shorter than this	100
`minSilenceDurationMs`	Ignore silence shorter than this	100