Skip to content

Live Captions#

OpenVidu Meet includes a built-in Live Captions feature that turns speech into text in real-time. This is a powerful tool for making your meetings more accessible to hearing-impaired participants, helping participants in noisy environments, and assisting non-native speakers.

How to Enable Live Captions in OpenVidu Meet#

Local Meet Deployment Limitation

Live Captions are not available in local Meet deployments. You must use either the OpenVidu Local deployment or an OpenVidu production deployment to enable this feature.

1. Connect to your OpenVidu deployment#

SSH into an OpenVidu Node and navigate to your OpenVidu deployment directory.

Depending on your OpenVidu deployment type:

If you are using OpenVidu Local (Development), simply navigate to the configuration folder of the project:

# For OpenVidu Local COMMUNITY
cd openvidu-local-deployment/community

# For OpenVidu Local PRO
cd openvidu-local-deployment/pro

If you are using OpenVidu Single Node, SSH into the only OpenVidu node and navigate to:

cd /opt/openvidu/config

If you are using OpenVidu Elastic, SSH into the only Master Node and navigate to:

cd /opt/openvidu/config/cluster/media_node

If you are using OpenVidu High Availability, SSH into any of your Master Nodes (doesn't matter which one) and navigate to:

cd /opt/openvidu/config/cluster/media_node

2. Enable the Speech Processing Agent#

Modify file agent-speech-processing.yaml to enable the Live Captions Service with processing: manual:

docker_image: docker.io/openvidu/agent-speech-processing-vosk:3.8.0

enabled: true # (1)!

live_captions:
    processing: manual # (2)!
  1. Set enabled to true to activate the Speech Processing Agent.
  2. Set processing to manual; participants will activate captions on demand via a toolbar button.

Info

By default, the Speech Processing Agent uses a local Vosk model for speech-to-text transcription.

For a more advanced setup, consider using a cloud-based provider. See Cloud providers for more information.

Default language is English

The Speech Processing Agent uses English for speech-to-text transcription by default. To use a different language, you must configure a different Vosk model. See Vosk models configuration for details on changing the language model.

3. Enable Captions in OpenVidu Meet configuration#

Edit the meet.env file and ensure the following configuration variable is set:

MEET_CAPTIONS_ENABLED=true

4. Restart OpenVidu#

Apply your changes by restarting OpenVidu. This ensures the system recognizes the new live captioning capabilities.

Depending on your OpenVidu deployment type:

Run where docker-compose.yaml is located:

docker compose restart

Run this command in your node:

sudo systemctl restart openvidu

Run this command in your Master Node:

sudo systemctl restart openvidu

Run this command in one of your Master Nodes:

sudo systemctl restart openvidu

5. Enable/Disable Captions for specific Rooms#

Captions are enabled by default when a room is created, whether through the UI or the REST API . This behavior can be overridden to enable or disable captions on a per-room basis from the Room Features step of the room configuration wizard, using the Captions toggle.

Using Live Captions in a Meeting#

Once live captions are enabled for a room, any participant can turn them on during the meeting by clicking the captions button in the toolbar. Captions then appear instantly at the bottom of the screen as participants speak, with no additional configuration required. The interface is designed to be easy to read without blocking the video feed.