Skip to main content

Using Live Voice Transcription in elvex

Learn how to use live voice transcription in elvex to speak your messages and have them transcribed in real time.

Typing long messages or dictating complex instructions can slow you down. elvex supports live voice transcription, which lets you speak your messages and have them transcribed in real time — so you can interact with your agents hands-free or more naturally.

You might use live transcription when:

  • You want to dictate a long, detailed message without typing

  • You're working hands-free or on the go

  • You find it faster to speak your thoughts than to type them

What is live transcription?

Live transcription converts your spoken audio into text directly in the elvex composer as you speak. You can review and edit the transcribed text before sending it, giving you the speed of speech with the precision of text.

Transcription providers

elvex supports multiple transcription methods depending on your setup:

  • Browser (default) — elvex uses your browser's built-in speech recognition out of the box. No additional configuration is required.

  • OpenAI — if your account has an OpenAI provider connected, elvex can use it for transcription. Text appears in segments, added after each natural pause in your speech.

  • Mistral/Voxtral — if your account has a Mistral/Voxtral provider connected, transcription streams word-by-word as you speak.

All methods produce accurate results; the difference is in how text appears in real time and whether a provider needs to be configured by your admin.

How to use live transcription

  1. Open a conversation in elvex

  2. Click the microphone icon in the composer toolbar to start transcription

  3. Speak your message — your words will appear in the composer as you talk

  4. When you're done, click the microphone icon again to stop

  5. Review the transcribed text and click Send when ready

For admins: configuring the transcription provider

By default, elvex uses the browser's built-in speech recognition (shown as Disabled in settings). If you'd like to use OpenAI or Mistral/Voxtral for transcription instead, you can update this in your admin settings.

  1. Go to Admin Settings > Agents > Audio transcription provider

  2. Select your preferred transcription provider from the dropdown

  3. Save your changes — all users on your account will now use the selected provider

Did this answer your question?