Objective

This article explains how to create a batch transcription configuration for Programmable Voice so that completed calls are transcribed with the language and engine settings you choose, delivered to the destination you specify, and labelled with the right speaker roles.

Product

Programmable Voice

Environment

Twilio Console

Procedure

1. In the Twilio Console, select Communications in the left sidebar.
2. Select Voice, then Transcriptions.
3. Select Create configuration. If you have not yet accepted the AI/ML terms, an AI/ML features dialog appears first — see "How do I accept the AI/ML terms for voice transcriptions?"
4. On the Name configuration step, enter a Transcription configuration name. Optionally add a Description of up to 128 characters. Select Next.
5. On the Set up language & engine settings step, choose your Language settings: Monolingual language code, Multilingual language code, or Twilio managed. Choosing Multilingual language code or Twilio managed fills in Language, Transcription engine, and Speech model for you; choosing Monolingual language code lets you set them yourself. Select Next.
6. On the Destination step, configure at least one destination:
7. Under Transcription webhook callback, enter a value in Add a Webhook URL to have transcription results sent to your own application.
8. Under Conversation configuration integration, select a conversation configuration to send transcripts downstream for further analysis. Select Next.
9. On the Participant mapping step, set Participant 1 (channel 1) and Participant 2 (channel 2) to Human agent, AI agent, or Customer. Select Next.
10. On the Summary step, review your settings and select Create configuration.
11. A confirmation appears reading "Transcription configuration created" with the message "The transcription configuration has been created successfully.", and you are taken to the new configuration's detail page. It also appears in the Transcriptions list.

Additional Information

• At least one destination is required. Attempting to continue past the Destination step with neither a webhook URL nor a conversation configuration shows "You must configure at least one destination method before proceeding."
• Some language, engine, and speech model values are unavailable depending on the Language settings you pick.
• The two participants cannot hold the same role, and an agent role on one channel rules out an agent role on the other. Unavailable options are shown as unselectable.
• To abandon the setup, select the close control, then confirm with Close without saving in the Close without saving? dialog. You are returned to the Transcriptions list and nothing is saved.

This content was generated by AI and reviewed, edited, and verified by a human.