The GStreamer team has released version 1.28.1, the first bugfix update in the stable 1.28 series. Notable additions include a Whisper-based speech-to-text transcription element, which expands GStreamer’s audio processing and transcription capabilities. The new debugseimetainserter plugin supports testing SEI metadata insertion in video streams.
This update corrects several video playback and rendering issues. It resolves scaling and resizing issues with UIView on EAGL and Vulkan backends, improves reverse playback and gap handling, and fixes field and order changes in avviddec’s mixed interlace mode.
CUDA and OpenGL interoperability in cudaupload and cudadownload has been improved, and decodebin3 now handles collection switching more effectively.
WebRTC components receive further enhancements. Updates include improved H.264 profile and level negotiation in webrtcsink, support for custom signaling headers, new request-type pads in webrtcsrc for encoded downstream data, and role-specific signaling in livekit WebRTC integrations.
Moreover, the v4l2 plugin now supports a stateful AV1 decoder, and vtdec and vtenc include multiple fixes for Apple’s VideoToolbox framework. The applemedia stack receives build improvements and patches for tvOS support, including cross-compilation updates. A new subproject also provides the LunarG MoltenVK SDK for development environments.
Additional improvements include enhanced wavpack handling for format changes and non-S32 samples, improved Continuity Counter management in tsdemux, updates to speechmatics for configurable audio event detection, and several fixes in livesync to improve reverse playback and latency handling.
For more information, see the announcement. Binaries for Android, iOS, macOS, and Windows are expected to be available soon.
