YazSes v2.9.0-dev.1 — Wave M opens: minimal-bandwidth AAC & text-intelligence (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 M (design/vision/v2-research/15-wave-m.md) — a fresh state-of-the-art round spanning the lowest-bandwidth AAC input and new text-intelligence layers. Ten evidence-anchored features are designed; this drop ships the first two pure cores.
New in dev.1 (all off by default)¶
- Vocal Morse (ADR-v2-105,
[morsevox]) — type by Morse using two distinguishable vocal sounds (short/long): full text from a single reliable vocalization, with adaptive timing that re-fits to your rhythm.classify_pulse/classify_gap+MorseDecoder(international Morse) +adapt_dot_threshold. The lowest-bandwidth input in the whole set — for locked-in, late-stage ALS, and severe-CP users who have essentially no other channel. Anchor: Google Gboard Morse (2018). - Checksum-Validated Data Entry (ADR-v2-106,
[checkdigit]) — dictated account/ID numbers run through the field's check-digit algorithm (Luhn, ISBN-10/13, Verhoeff), so a mis-heard digit is caught before it lands — with single-edit fix suggestions.validate+suggest_fix. Catches STT's single most damaging silent error. Anchor: Luhn/ISO-7812, IBAN MOD-97, ISBN, Verhoeff.
yazses features now lists 117 capabilities. (Vocal Morse is experimental — features enable requires --force.)
Wave M remaining¶
Diagrams-as-Code · Interruptible Read-Back Proofreading · Style-Consistency Enforcer · Screenplay Auto-Format · Semantic Line Breaks · Spoken Spaced-Repetition Capture · Suggestion-Mode Dictation · Acronym/Glossary Manager.
Quality¶
1378 tests green across Linux × macOS × Windows on Python 3.11 and 3.12. Both new cores are unit-tested to 100% with no new dependency (pure timing/arithmetic; the Morse pulse stream comes from the existing hold detector). The default install and the v1 dictation path are unchanged.