AI Foundationspredict · compress · act

concepts → Mechanism

Mechanism

Neural audio codec

frontieras of 2026-092 connections

compresses a waveform into a short stream of discrete codes — the move that makes sound a language problem

Where it sits

The prism has six jobs across and eight layers down. Its primary cell is represent × L2. Hatched cells cannot exist — a GPU does not learn, an institution does not infer.

Overlays: Efficiency, cost & energy.

What must come first — and what it unlocks

Left to right is reading order, derived from the prerequisite_of edges. Nothing here is hand-ordered: the diagram is the graph.

nothing comes first — this is a starting pointNeural audio codecSpeech token

Unlocks: Speech token

Where to read it

The chapter that introduces it, and any chapter that uses it again.

60Audio Modelsact XIII · Beyond Text

Where it comes from

paperHigh Fidelity Neural Audio CompressionAlexandre Défossez, Jade Copet, Gabriel Synnaeve, Yossi Adi · 2022

Every connection

All 2 edges touching this node, grouped by relation family — the sections above are highlights from this list. Colours match the relation families inthe atlas.

Order · 1
must come beforeSpeech tokenQuantity
Flow · 1
usesTokeniserMechanism

This page is a projection of one node in src/data/concepts.ts. It has no prose file of its own — 521 declared edges produce all 340 of these pages. Edit an edge and both endpoints change.