YazSes v2.8.0-dev.5 — Wave L complete (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.

Completes Wave L (design/vision/v2-research/14-wave-l.md) — all ten features shipped — with the final two pure cores, both accessibility-first:

New since dev.4 (all off by default)¶

• Mouth-Sound Switch Access (ADR-v2-097, [mouthswitch]) — non-verbal mouth sounds (pop = advance the highlight, cluck = select) drive a timed scan-and-select selector — the switch-scanning AAC access method for people who cannot produce reliable speech at all. classify_mouth_sound + ScanSelector (a pure timed state machine, testable with a fake clock). Anchor: Apple Sound Actions for Switch Control (iOS 15).
• Involuntary-Vocalization Auto-Excision (ADR-v2-102, [involuntary]) — detects and deletes involuntary sounds (cough, throat-clear, sneeze) from the stream so they never become garbage tokens or spurious commands. is_involuntary_vocalization + excise_nonspeech_spans. A win for chronic-cough and medical dictation. Anchor: Hyfe Cough Tracker (F1000Research; 91%/98%).

yazses features now lists 115 capabilities. (Mouth-Sound Switch Access is experimental — features enable requires --force.)

Wave L — complete¶

Vocal Joystick · Earcon Feedback · Beam-Steered Spatial VAD · Prosodic Auto-Punctuation · Hesitation-Hold Endpointing · Pitch-Contour Gestures · Breath-Paced Dictation · Whisper-Aware Mode · Mouth-Sound Switch Access · Involuntary-Vocalization Excision. All ten shipped, pure, 100%-covered, off by default.

Quality¶

1369 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 (the scan cursor + span-removal are pure; the acoustic classifiers are optional lazy extras). The default install and the v1 dictation path are unchanged.