speech

Triggered when speech-to-text transcription is completed for a call recording. Transcription is processed asynchronously after the call ends.

When This Event Fires

• A call recording has been transcribed by the speech recognition system
• The transcription result is stored and ready for retrieval

Payload

Field	Type	Description
call_id	integer	Internal call ID
speech_state	string	Transcription state (e.g. "speech_stored")
payload	array	Transcription segments (see below)

Transcription Segment Fields

Each item in the payload array represents a segment of the transcription:

Field	Type	Description
text	string	Transcribed text for this segment
start	float	Start time offset
end	float	End time offset
start_sec	float	Start time in seconds from call beginning
channel	string	Audio channel — "left" (caller) or "right" (receiver)

Example Payload

{
  "call_id": 12345,
  "speech_state": "speech_stored",
  "payload": [
    {
      "text": "Hello, I'd like to inquire about your services.",
      "start": 0.5,
      "end": 3.2,
      "start_sec": 0.5,
      "channel": "left"
    },
    {
      "text": "Of course! How can I help you today?",
      "start": 3.5,
      "end": 5.8,
      "start_sec": 3.5,
      "channel": "right"
    },
    {
      "text": "I'm looking for a business phone solution.",
      "start": 6.0,
      "end": 8.5,
      "start_sec": 6.0,
      "channel": "left"
    }
  ]
}

Channel Mapping

The channel field indicates who is speaking: "left" is typically the caller (external party) and "right" is the receiver (your team member).

Related Events

• incoming_call_end / outgoing_call_end — the call that was transcribed
• summary — AI summary generated from this transcription