/
Agent - Speech Features

Agent - Speech Features

 

Overview

Agent includes a pair of optional features that let you use speech instead of typing:

  • Speech-to-Text – dictate requests using your microphone.

  • Speech-to-Speech (Voice Chat) – conduct a spoken conversation with Agent.

Speech-to-Text

Permissions & browser support

  • Your browser must allow microphone access. If you previously blocked Agent from accessing the microphone, update the site permissions.

  • Speech-to-Text is supported in Chrome, Safari, and Edge.

Quick Start

  1. Click the microphone icon in the bottom-right of the chat input.

image-20260108-134708.png
image-20260108-134740.png

  1. Allow microphone access if prompted.

  2. Speak your request.

  3. Click the microphone again (or pause) when you’re done.

Behaviour

Speech recognition stops automatically after 5 seconds of silence or immediately if you click the microphone button again.

Limitations

  • Language support: Supports English and Norwegian. Change the language in Settings.

  • Browser compatibility: Speech-to-Text is not supported in Firefox.

Speech-to-Speech (Voice chat )

Voice Chat is available only in the Azure Marketplace and in SaaS-based Agent deployments.  It is not supported in Agent instances integrated with Index.

Important: Voice Chat does not support image files as context. Any selected images will be ignored.

Voice Chat lets you ask Agent questions verbally and receive spoken responses. It uses a dedicated Speech-to-Speech model, separate from the model selected for text chat.

Quick start

  1. To begin Voice Chat, click the button on the bottom-left of the chat input box.

image-20260108-134958.png

  1. Allow microphone access if prompted.

  2. Speak your question

  3. Agent will generate a spoken response.

  4. Transcriptions of your input and Agent’s output will appear in the chat window.

Conversation flow

  • When listening for input, agent will wait until you pause.

  • While a response is being played, further audio input is ignored. After playback ends, Agent will resume listening.

  • The input control allows you to pause response playback and/or to mute microphone input:

image-20251015-124328.png
image-20251015-124350.png

 

Availability & requirements

Voice Chat is available only in the Azure Marketplace and in SaaS-based Agent deployments.  It is not supported in Agent instances integrated with Index.

Limitations

Voice Chat does not support image files as context. Any selected images will be ignored.

Troubleshooting

  • Microphone icon is missing or disabled – Check that your browser and Agent version are compatible with the speech features.

  • No microphone prompt – Microphone access may be blocked. Update site permissions in your browser to allow microphone use.

  • Voice Chat button not visible – Voice Chat is only available in the Azure Marketplace and SaaS versions of Agent.

  • Agent doesn’t respond when I speak – Make sure Agent isn’t currently playing back a response; it only listens when playback is stopped or paused.