# `design/research/state-of-the-art/` — the SoA studies behind the v2 waves

Five state-of-the-art / market studies that fed directly into ADR-v2-129 and the v2
cognitive-layer features. Each one is a snapshot of the field as of its date, kept
as a historical research record rather than updated in place — a later study
supersedes an earlier one in prose, the way an ADR does.

| File | What it holds |
|---|---|
| [`2026-06-14-yazses-10-features-soa.md`](2026-06-14-yazses-10-features-soa.md) | Feasibility review of ten candidate feature ideas, evidence-graded, with a ready-now/partial/too-early verdict for each. |
| [`2026-08-07-stt-engine-sota.md`](2026-08-07-stt-engine-sota.md) | Local/offline STT engines beyond faster-whisper, streaming approaches, hotword biasing, and on-device personalization. |
| [`2026-08-07-hci-input-sota.md`](2026-08-07-hci-input-sota.md) | Webcam gaze, consumer EMG, consumer BCI/EEG, multimodal fusion, and accessibility input, scoped to a CPU-only offline daemon. |
| [`2026-08-07-competitor-landscape.md`](2026-08-07-competitor-landscape.md) | A neutral market survey of voice-dictation and voice-control products as of mid-2026 — not a strategy document. |
| [`2026-08-07-killer-features-10x-study.md`](2026-08-07-killer-features-10x-study.md) | The synthesis of the three studies above plus a codebase reachability audit, into the program that became ADR-v2-129. |

Companion: [`../2026-08-11-hci-reference-corpus.md`](../2026-08-11-hci-reference-corpus.md)
for the underlying citation corpus, and [`../../adr/`](https://github.com/MSKazemi/yazses/blob/main/design/adr) for the decisions
these studies informed.