Skip to navigation
Calling

live_transcribe

View as MarkdownOpen in Claude

Start live transcription of the call. The transcription will be sent to the specified webhook URL. The transcribe calls guide walks through a first call, the webhook events, and AI summaries.

For whole-call transcription delivered after the call ends, see transcribe.

Properties

live_transcribe
objectRequired

An object that accepts the following properties.

live_transcribe.action
string | objectRequired

The action to perform. See actions below.

Actions

The action property controls the transcription session. Use start to begin transcribing with configuration options, stop to end an active session, or summarize to request an on-demand AI summary mid-session.

action.start
objectRequired

Start a live transcription session.

start.webhook
string

The URL to receive transcription events via HTTP POST. When live_events is enabled, each utterance is sent as soon as it’s transcribed. When the session ends, the whole conversation is sent, with a summary when ai_summary is enabled. See Webhooks for the events. Authentication can also be set in the URL in the format of username:password@url.

start.lang
stringRequired

The language to transcribe. See supported voices & languages.

start.live_events
booleanDefaults to false

Whether to enable live events.

start.ai_summary
booleanDefaults to false

Whether to enable automatic AI summarization. When enabled, an AI-generated summary of the conversation will be sent to your webhook when the transcription session ends.

start.speech_timeout
integerDefaults to 60000

The timeout for speech recognition in milliseconds. Minimum value: 1500.

start.vad_silence_ms
integerDefaults to 700 | 500

Voice activity detection silence time in milliseconds. Default depends on the speech engine: 700 for Deepgram, 500 for Google. Omitting the value or setting it to 0 uses the engine default.

start.vad_thresh
integerDefaults to 400

Voice activity detection threshold. Range: 0 to 1800. Setting it to 0 uses the default of 400 rather than a zero threshold.

start.debug_level
integerDefaults to 0

Debug level for logging.

start.direction
string[]Required

The direction of the call that should be transcribed. Possible values: remote-caller, local-caller. Also accepts a SWML variable reference for the whole array or for each item; the resolved values must be one of the possible values.

start.speech_engine
stringDefaults to deepgram

The speech recognition engine to use. Possible values: deepgram, google. Also accepts a SWML variable reference that resolves to one of these values.

start.ai_summary_prompt
string

The AI prompt that instructs how to summarize the conversation when ai_summary is enabled. This prompt is sent to an AI model to guide how it generates the summary.

Example: “Summarize the key points and action items from this conversation.”

start.hints
string[]

Words or phrases that help speech recognition pick out specific terms, such as product names or industry jargon. Each hint is a plain string; empty strings are ignored. Also accepts a SWML variable reference that resolves to an array of hints.

Examples

version: 1.0.0
sections:
main:
- answer: {}
- live_transcribe:
action:
start:
webhook: 'https://example.com/webhook'
lang: en
live_events: true
direction:
- remote-caller
- local-caller
speech_engine: deepgram

Webhooks

Every event arrives as an HTTP POST to the webhook URL. Read channel_data.content_disposition to tell them apart: