Ceeblue has rolled out to production the latest version of the Media Fabric. This release introduces three fundamental advances:
- Deterministic Streaming
- Unified Media and Metadata Handling
- Significantly Improved Resilience under Real-World Conditions
Deterministic Streaming
At its core, segment generation is now deterministic across origins, meaning identical media time always produces identical segment boundaries and identifiers. This directly improves cache efficiency, enables seamless origin failover and horizontal scaling, and ensures alignment across WebRTS, LL-HLS, HESP, and CMAF outputs.
Unified Media and Metadata Handling
In parallel, the system now treats media and data as a single entity through a unified timeline that merges audio, video, and timed metadata such as ID3, SCTE-35, and subtitles. This ensures accurate synchronization, eliminates drift between tracks, and enables consistent behavior across ingest, processing, and delivery.
Building on this, WebRTS introduces per-segment composability, allowing segments to be dynamically constructed via URI-based track selection. This enables instant track switching, multi-camera angle changes, and flexible media-plus-metadata combinations, all while maintaining timeline integrity and independent segment playback.
Improved Resilience
The release also delivers substantial improvements in live playback robustness, particularly in imperfect conditions such as missing segments, sparse tracks, delayed inputs, or discontinuities. Enhanced track selection, safer drop logic, and better startup behavior reduce playback failures and visual artifacts. Across transport layers, TS, SRT, RTP, RTMP, and WebRTC have been hardened, improving synchronization accuracy, timestamp handling, and long-running stability.
Finally, metadata support is now a first-class feature across all workflows, including direct streaming interfaces like HTTP-POST. Timing precision has been significantly improved, especially for SCTE-35 signaling, RTP clock alignment, and RTMP timestamp rollover scenarios. This includes initial support for subtitles on selected input (HTTP-POST/mkv) and output (WebRTS) protocols.
The overall result is a shift toward a timeline-centric architecture with deterministic outputs, composable delivery, and resilient live performance.
To learn more about these new capabilities or to set up a test account to try them out, go ahead and contact us.
