Quote the operative idea, then translate. Dolby's grant US10503461B2, "Binaural rendering for headphones using metadata processing" (issued December 10, 2019; assignee Dolby Laboratories Licensing Corporation), is a granted patent classifying across H04S 7/304 and 7/306 (binaural rendering) plus G06F 3/162/3/165 (audio interface). The element doing the work is in the title: metadata processing.
Here is the mechanism. Ordinary stereo bakes the mix into two fixed channels — left and right, decided at production time. Binaural rendering with metadata does something different: the content carries information about where each sound object is positioned in a 3D space, and the renderer, at playback, computes what your two ears should hear to place that sound there — using head-related transfer functions, the filters that encode how your head and ears color sound from different directions. The two channels are generated on the fly from the positional metadata, not pre-mixed.
Why the metadata is the claim's spine: it decouples the spatial intent from the playback device. The same metadata-bearing stream can render to any headphones, and — crucially — can react to head tracking, so a sound stays anchored in space as you turn your head. That is exactly the "spatial audio" experience modern earbuds advertise. Pull the metadata layer out and you are back to fixed stereo, outside the claim.
This is why the family became a licensing backbone rather than a single-product feature. A claim anchored to metadata-driven rendering reads on any implementation that carries positional metadata and renders it binaurally — which is most of the spatial-audio ecosystem. That breadth, anchored to a concrete technical mechanism, is what makes it valuable as a license rather than a moat: the rational move for device makers is to license, not design around.
The precision a claims analyst owes: this 2019 grant is foundational metadata-binaural art, but it is not the whole of spatial audio. Head-tracking-specific refinements, earbud-specific implementations, and personalization of the transfer functions live in later, separate filings from Apple, Meta, and others. Scope ends at metadata-driven binaural rendering. But that is the load-bearing layer, and it is the one a licensing conversation starts from.