YazSes v2.8.0-dev.1 — Wave L opens: non-speech & prosodic voice interaction (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 L (design/vision/v2-research/14-wave-l.md) — a fresh state-of-the-art round that turns YazSes toward a genuinely new area: non-speech vocal signals (formant/pitch/loudness/breath) and acoustic prosody as first-class interaction channels — for users who can phonate but not articulate discrete words, and for word-free eyes-free control. Ten evidence-anchored features are designed; this drop ships the first two pure cores.
New in dev.1 (all off by default)¶
- Vocal Joystick (ADR-v2-095,
[vocaljoystick]) — continuous analog cursor/scroll control by sustaining vowels: "ahh" → right, "eee" → up, louder = faster, a sharp pitch-jump = click. No words.vowel_to_direction(formants → 8-way) +vocal_control_vector+ aVocalJoystickintegrator. A fundamentally new modality (velocity, not selection) for severe motor impairment (ALS, high SCI, CP). Anchor: Bilmes et al., The Vocal Joystick (ASSETS 2006). - Earcon Feedback Language (ADR-v2-096,
[earcon]) — tiny structured tones signal daemon state eyes-free: a rising two-note motif for "recording started", a muted buzz for "low confidence", a chime for "command done" — faster and less intrusive than spoken read-back.earcon_formotif grammar +render_earconnumpy synth. Anchor: Brewster earcons (TOCHI 1993).
yazses features now lists 107 capabilities. (Vocal Joystick is experimental — features enable requires --force.)
Wave L remaining¶
Mouth-Sound Switch Access · Beam-Steered Spatial VAD · Breath-Paced Dictation · Whisper-Aware Mode · Hesitation-Hold Endpointing · Involuntary-Vocalization Auto-Excision · Pitch-Contour Vocal Gestures · Prosodic Auto-Punctuation.
Quality¶
1344 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 Vocal Joystick's formant/pitch tracker is an optional lazy extra; the pure mapping + integrator are dependency-free). The default install and the v1 dictation path are unchanged.