AIAI Actions

Transcribe

Transcribe audio and voice messages to text with timestamps

The Transcribe action converts audio and voice messages into text using OpenAI Whisper, providing both full transcription and word-level timestamps.

How to Use

  1. Receive or send an audio message in a chat
  2. Click the Transcribe button on the audio message
  3. The transcription appears inline with the audio, including timestamps

How It Works

The audio file is sent to OpenAI Whisper which returns a full text transcription with both sentence-level and word-level timestamps. The transcription is stored on the message and displayed in the chat.

Once transcribed, the text is also available to other AI actions — for example, Suggest Reply can reference the transcribed content when generating responses.

Auto-Transcription

Administrators can enable automatic transcription in Settings → Inbox → Auto-transcribe messages. When enabled:

  • Every inbound audio message is automatically transcribed in the background
  • The transcription appears on the message once complete
  • No manual action is needed from operators

Translation

To translate a transcription into another language, use the Translate Transcript action.

Requirements

  • OpenAI onlyOpenAI provider must be connected (Whisper)
  • Not available when using Anthropic as the sole AI provider

On this page