RESEARCH — DELIVERY ALGORITHM
ATX
Audio Transport
ATX carries a full immersive mix on six channels — a fixed bandwidth, whatever the number of objects in the master. One predictable stream to provision and cache, and full flexibility at the far end: the same feed renders to a speaker layout, a soundbar or headphones with head-tracking.
THE PROBLEM
Immersive doesn't fit the pipe
Object-based immersive audio was designed for the studio: dozens of objects, each with its own channel and metadata. That is the right shape for mixing — and the wrong one for moving. A full master can weigh sixteen, thirty, a hundred channels, and every link between the mix room and the listener — SDI paths, IP streams, storage, multi-room buses — was sized for a handful.
For a streaming or broadcast platform the problem is not only size — it is variability. A catalogue where every title carries a different channel count is a catalogue that cannot be provisioned: bitrate ladders, CDN caching, decoder capabilities and player buffers all want one known figure, not a range that depends on how each mix was authored.
The usual answer is to give up early: render a fixed downmix and ship that. Every endpoint then inherits the same compromise — the stereo listener gets a fold-down, the immersive room an upmix of a fold-down — and the mix's spatial intent is gone at the first hop. What delivery needs is a stream that is fixed on the wire and still a scene at the far end.
HOW IT WORKS
How it works
Fixed bandwidth by design
Whether the master carries eight objects or a hundred, the stream is six channels. Links, storage, CDN caching and player buffers are provisioned once, for the whole catalogue — the figure never moves.
Every endpoint, one feed
The receiving side renders the scene to what is actually there — a full speaker layout, a soundbar, or binaural headphones with head-tracking — instead of playing someone else's downmix.
Quality that holds
Localization, width and ambience survive the trip: the render is validated against the reference rendering of the original master, layout by layout — not against a compromise downmix.
UNDER THE HOOD
Going deeper
ATX condenses an object-based master into a fixed six-channel scene representation plus a light stream of scene data. The condensation happens at the encoder, where the original mix is still fully known — the decisions that matter are taken against ground truth instead of being guessed at the receiver.
The result is the property platforms care most about: a constant, known bandwidth. Six channels is the figure for the whole catalogue — an object-heavy blockbuster costs the wire exactly what an acoustic session costs — so provisioning, caching and decoder requirements stop depending on how each title was mixed.
At the receiving end, the same feed renders to whatever the endpoint owns. Speaker layouts from stereo to 9.1.6 get a rendering computed for their actual geometry; a soundbar gets its own; headphones get binaural with head-tracking. No endpoint plays another endpoint's downmix.
The same container serves files and live links: encode once and archive, or encode in real time and stream — the player treats both identically.
AT A GLANCE
At a glance
| Input | Object-based immersive masters — ADM / BW64 |
| Transport | Six channels — fixed bandwidth, independent of object count |
| Output | Speaker layouts stereo → 9.1.6 · soundbars · binaural with head-tracking |
| Modes | File-based and real-time |
| Availability | Encoder & player apps — in development |
PARAMETERS
Hands on the algorithm
| Input | ADM / BW64 immersive master |
| Transport | Six channels — fixed |
| Output layouts | Stereo → 9.1.6 · soundbar · binaural with head-tracking |
| Modes | File · real-time link |
| Monitoring | Preview any target layout at encode time |
POSITIONING
Compared to the alternatives
vs shipping the full master
A hundred-channel session moves only where a hundred channels fit; ATX carries the same scene on six, over links and storage that already exist.
vs a fixed downmix
A downmix freezes one compromise for every listener; ATX delivers the scene and lets each endpoint render it for its own system.
vs object codecs — MPEG-H & AC-4
Object codecs were built for the consumer decoder: a licensed bitstream, dedicated decoder support at every endpoint, profiles negotiated device by device. ATX stays in plain audio — six channels that ride anything that already carries audio, from SDI and AoIP to files and a streaming stack — rendered by a lightweight player under one licence, from one vendor.
vs per-layout deliverables
One file per target multiplies versions, storage and QC, and still misses tomorrow's layout; ATX ships one deliverable — one bandwidth — that renders to all of them.
APPLICATIONS
Where it fits
Streaming platforms
One fixed-bandwidth stream for the whole immersive catalogue: provisioning, bitrate ladders and caching sized once, rendering left to the device.
Broadcast
An immersive programme travels production-to-transmitter on six channels — plant links and routers sized for surround carry the full object mix.
Multi-room & venues
One feed serves many rooms: each endpoint renders the same scene to its own system, from a stereo foyer to an immersive theatre.
Consumer endpoints
Set-top boxes, soundbars and mobile apps receive one light stream and render locally — including head-tracked binaural on headphones.
INTEGRATION
Built to live inside your product
| Delivery | C · C++ · MATLAB · .dsp — full source code |
| Platforms | macOS · iOS · Windows · Linux |
| DSP platforms | Flow DSP · Audio Weaver — solutions in preparation |
| Documentation | Whitepaper — every algorithm explained, in the clear |
OEM LICENSING
- ■ One-time payment per brand
- ■ Full source code — C, C++, MATLAB, .dsp
- ■ Whitepaper — all algorithms explained
- ■ Integration support included
- ■ Free updates
- ■ Volume discounts on multiple licenses
SHIPS IN