Today, the GStreamer project announced the release of GStreamer 1.26.5, marking the fifth maintenance update in the GStreamer 1.26 series, a leading open-source multimedia framework that is free and cross-platform.
This version introduces support for JPEG XS within the videorate and imagefreeze elements, enhances the cea708mux by allowing the discarding of specific services on inputs, and improves the awstranslate element to support brevity. Additionally, the awstranscriber2 now has an option to partition speakers in the transcription output.
GStreamer 1.26.5 also upgrades the cea608overlay and cea708overlay to accept GPU memory buffers, contingent on downstream support for overlay composition. Improvements were made to the Speechmatics speech-to-text engine, specifically exposing the mask-profanities property, and updates were applied to wasapi2 audio device monitoring.
Moreover, this update enhances WebRTC capabilities by adding a WHEP client signaller and the whepclientsrc element on top of webrtcsrc. Many improvements and fixes were made to various elements like threadshare, rtpbin2, and gst-device-monitor-1.0.
Several bugs from previous releases were addressed, including those related to caps negotiation in audioconvert, decodebin3 issues with ONVIF metadata streams, and memory leaks in V4L2 during dynamic resolution changes. Other fixes cover the VA encoder, Vulkan integration, and various build improvements aimed at ensuring reliability and stability.
For further details on the changes brought by GStreamer 1.26.5, users can check the release notes. The source tarball for GStreamer 1.26.5 is now available for download from the official website for those wishing to compile it from source, or it can be installed from most distributions’ stable repositories.