Speech to Text
The new Speech to Text service allows transcribing in real time the audio tracks of an OpenVidu Session. It makes very easy to include captions in your video calls, no matter what language your users speak!
This feature, available in OpenVidu Pro and Enterprise editions, is designed to support different transcription engines in the background. For now we have released Speech to Text with Azure support, but others are coming very soon (including a self-hosted open source alternative for no extra cost). You can read the Speech to Text documentation here.
Support for HTTP IP cameras
IP cameras API is now compatible with cameras publishing to
https URIs (these are generally in
.mjpeg format). These are without a doubt the most common IP cameras URIs, so this support opens up new possibilities of use.
OpenVidu Components features
- In case of a node crash in your OpenVidu Deployment, OpenVidu Components emits a new event
onNodeCrashed. This allows the application to implement fault tolerance in case of a node crash. See Fault Tolerance.
- Update to Angular 14.
OpenVidu Call features
- Speech to Text integration. Enable live captions in your OpenVidu Call sessions.
- Added Basic Auth to securize the access to the application. Now by default there is a login page requesting for credentials before accessing a session (this can be disabled). New env vars:
- Adds support for fault tolerance and automatic reconnection to the session without user intervention in case of a node crash. See Fault Tolerance.
- Minor UI improvements.
- OpenVidu Call : users with no webcam or microphone were shown a blank screen when loading OpenVidu Call landing page. OpenVidu Call now allows connecting users with no input devices available.
- IP cameras : if option
onlyPlayWithSubscriberswas enabled, there was a bug preventing the camera feed to automatically reactivate once a new Subscriber requested the camera's Stream. Hence, the camera's feed appeared frozen and the only way to fix it was unpublishing and publishing the IP camera again. Now this is fixed.
- Recording : COMPOSED recordings now use some new internal configurations that should help reducing the launch time of the recording module.
- S3 recording : when using non-AWS S3 buckets, there could be access denied errors if the S3 endpoint was a domain name and not an IP. Now configuration property
OPENVIDU_PRO_AWS_S3_WITH_PATH_STYLE_ACCESSallows forcing path style requests in the internal S3 client, which gets rid of this problem.
- openvidu-browser : calling
Publisher.replaceTrackwas triggering a wrong
StreamPropertyChangedevent, which resulted in a wrong modification of
videoActiveStream properties. Now new
StreamPropertyChangedEvent.changedPropertyvalues have been added:
audioTrack. You can use these new values to know when a remote user has called
replaceTrackon its Publisher object.