Talkscriber

Talkscriber provides real-time speech-to-text over a streaming WebSocket connection. In Vapi, Talkscriber uses the Whisper model to turn the caller’s audio into transcripts for the assistant.

Speech-to-text

Set transcriber.provider to talkscriber. Vapi currently supports the whisper model and 100 transcription languages.

$
curl -X PATCH "https://api.vapi.ai/assistant/ASSISTANT_ID" \
>
  -H "Authorization: Bearer $VAPI_API_KEY" \
>
  -H "Content-Type: application/json" \
>
  -d '{
>
    "transcriber": {
>
      "provider": "talkscriber",
>
      "model": "whisper",
>
      "language": "en"
>
    }
>
  }'

For additional API configuration options, review the TalkscriberTranscriber fields in the Create Assistant API reference.

Talkscriber is available through Vapi’s managed integration, so you do not need to add a separate Talkscriber API key.

Configuration

Choose a language

Set transcriber.language to the language spoken by the caller. Providing the expected language gives Talkscriber a specific language target; if you omit it, Vapi uses English.

Supported languages

The following languages are included in Talkscriber’s supported language list and accepted by Vapi.

When to use Talkscriber

Talkscriber is a useful choice when your assistant needs:

• Real-time streaming transcription during live calls.
• Whisper-based transcription with an explicit language target.
• Coverage for languages that may not appear in a smaller provider-specific language set.

The language setting is a language hint, not a translation setting. The transcript remains in the language spoken by the caller.

Related