From encoder-generated quality scores across the chain to automated switching and failover
Your viewers see the damage first. Your monitoring catches up too late. Your encoder has known about it all along.
That is the core problem that a newly published SVTA specification is designed to solve. "SVTA2128: Transmission of Media Quality Assessment (MQA) Data" defines how encoder-generated quality scores can be carried through the entire streaming delivery chain, from encoder to CDN to player, as a real-time signal rather than a post-event report. SVTA2128 specification
Where it all started
The intellectual origin of this work traces back to Urvashi Pal, Pre-Sales Senior Solutions Engineer at Akamai, who designed and patented the underlying architecture for carrying quality scores across the streaming chain. Filed in October 2023 and granted in December 2025, the patent covers the underlying architecture to enable client-side Video Quality Analysis on live OTT streams. To drive industry-wide adoption, Pal made the technology entirely royalty-free for the media ecosystem, a decision that cleared the path for the SVTA to formalise it as an open standard.
Building on CMCD
Understanding CMSD-MQA requires understanding where it sits in a broader quality signaling stack. Common Media Client Data (CMCD) was an important first step – a standard that allows media players to communicate telemetry upstream to servers and CDNs. It tells you what the client is experiencing. But it says nothing about what the encoder already knows about quality before content even reaches the client.
G&L Systemhaus ran a practitioner-focused workshop on CMCD, together with Akamai, Bitmovin and Touchstream, exploring how it changes error detection, log analysis and CDN integration in practice. That work pointed directly at the question CMSD-MQA is now answering. CMCD workshop on demand
The missing piece: CMSD
Common Media Server Data (CMSD) carries server-side metadata downstream. Its MQA extension defines how encoder-generated quality scores such as VMAF, PSNR and SSIM can be embedded in HTTP response headers and picked up by any downstream component, with no decoding or re-analysis of the content.
The gap this closes is significant. Quality metrics computed at encoding are currently discarded the moment they are generated. Downstream systems rely on bitrate as a rough quality proxy, deploying expensive probes to recreate information that already existed upstream. CMSD-MQA preserves that information and puts it to work.
See it. Score it. Switch it.
Under this three-part title, G&L’s Alexander Leschinsky organised a workshop in December 2025 together with Will Law of Akamai, Adrian Roe of Norsk/id3as and Brenton Ough of Touchstream. CMSD-MQA workshop on demand
See it: quality scores are already being computed at the encoder. CMSD-MQA makes them available to every component downstream.
Score it: those scores are expressed in a standardised format that any CMSD-aware component can read and act on, regardless of which encoder or vendor produced them.
Switch it: with quality data flowing through the chain in real time, components can make decisions based on actual picture quality, from encoding pipeline switching to failover and CDN routing.
The December workshop included a live demo in which just 2% packet loss was enough to trigger an automatic quality-based failover. Jan Ozer distilled the session into a detailed written summary with explanations and screenshots. Read Jan Ozer's write-up
The standard is now published
With SVTA2128 now formally published, the specification moves from draft to a stable reference point for implementation. Early adopters include Norsk, Touchstream and AWS Elemental. For operators evaluating vendors, the practical advice remains: ask when CMSD-MQA support is planned, and consider weighting it in RFPs as a signal of roadmap direction.
From workshop to show floor
Publishing a standard is one thing. Seeing it work across real systems from different vendors is another. At NAB Show 2026 in Las Vegas, the SVTA hosted a live demo at its Innovation Showcase booth titled "CMSD-MQA for Automated Encoding Pipeline Switching", developed by the working group of Akamai, Norsk, Touchstream and G&L Systemhaus. The four companies had previously co-presented the spec at Mile-High Video in Denver. Watch the recorded SVTA demo
Quality as a real-time, first-class signal across the entire delivery chain. The gap is closing.