AI Foundationspredict · compress · act

concepts → Task

Task

Speech recognition

frontieras of 2026-092 connections

sound in, text out — the task that scale first made reliable

Where it sits

The prism has six jobs across and eight layers down. Its primary cell is no mode × L1. Hatched cells cannot exist — a GPU does not learn, an institution does not infer.

Overlays: Evaluation & methodology.

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 pointSpeech recognitionnothing depends on it yet — a leaf in the reading order

What kind of thing it is — and what it is made of

It is part of: Audio model.

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

paperRobust Speech Recognition via Large-Scale Weak SupervisionAlec Radford, Jong Wook Kim, Tao Xu, Greg Brockman, Christine McLeavey, Ilya Sutskever · 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.

Structure · 1
is part ofAudio modelModel
Flow · 1
usesSpeech tokenQuantity

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.