YazSes v2.4.0-dev.1 — Wave H ship-now tier (developer preview)¶

Pre-release · not published (publish workflows are pinned to v0.x/v1.x; v2 dev tags do not ship). v1.4.1 remains the stable release. Every feature is OFF by default.

Opens Wave H — a fresh 2026 SoA round (10 candidate features distinct from the prior 62) in design/vision/v2-research/10-wave-h.md + design/adr/adr-v2-055..060. The six ship-now, dependency-free features:

New features (all off by default)¶

• Emoji & Symbol by Voice (ADR-v2-055, [commands] symbols) — "right arrow" → →, "shrug emoji" → 🤷, "degree sign" → °. Fills the Unicode gap ASCII voice-punctuation leaves.
• Voice Unit Conversion (ADR-v2-056, [convert]) — "twenty miles in kilometers" → "32.19 kilometers" (incl. temperature), evaluated inline, fully offline.
• Spoken Temporal Normalizer (ADR-v2-057, [temporal]) — "next Friday" → a concrete date, using only the local clock.
• Mid-Utterance Self-Repair (ADR-v2-058, [commands] self_repair) — "email Sarah no I mean Sara" → "email Sara", applied before typing; resolves chained repairs.
• Spoken Spreadsheet (ADR-v2-059, [spreadsheet]) — cell-addressed dictation + grid navigation: "next row", "cell down", "go to B7".
• Clipboard-History by Voice (ADR-v2-060, [cliphistory]) — "paste the second thing I copied", "the URL I copied".

yazses features now lists 71 capabilities.

Wave H backlog (medium / hardware-gated)¶

Ambient Audio-Event Guard, On-Device Condense, Slot-Filling Dictation, Few-Shot Command Spotter.

Quality¶

1186 tests green across Linux × macOS × Windows on Python 3.11 and 3.12. Every Wave H ship-now core is unit-tested to 100% with no new dependency; the default install and the v1 dictation path are unchanged.