← ClaudeAtlas

voice-hud-wayland-skilllisted

Hands-free Voice-to-Text HUD architecture for workbench technicians under Linux Wayland: local faster-whisper streaming, KWin overlay glassmorphism, and post-transcript LLM polishing.
4ntiDandruff/zero-bloat-skills · ★ 1 · AI & Automation · score 74
Install: claude install-skill 4ntiDandruff/zero-bloat-skills
# Voice HUD Wayland Skill Design pattern for a hands-free speech-to-text heads-up display (HUD) allowing technicians to dictate repair notes while hands are occupied holding soldering irons, tweezers, or multimeter probes. --- ## 1. Low-Latency Pipeline: Local Whisper + LLM Polishing ``` [ Bench USB Microphone / Headset ] ↓ [ PyAudio / SoundDevice (Silero VAD: Voice Activity Detection) ] ↓ [ faster-whisper (CTranslate2 INT8 Model on Local CPU - Latency <200ms) ] ↓ [ Raw Text Stream ] ↓ [ LLM Post-Processor (Repair Electronics Jargon: MOSFET, VCC, ALW) ] ↓ [ Wayland KWin HUD Overlay & Direct Keystroke Injection ] ``` --- ## 2. Floating Transparent Overlay (PyQt6 / Wayland) Under Wayland sessions (KDE Plasma 6 / GNOME), windows cannot be arbitrarily positioned without protocol flags: ```python from PyQt6.QtWidgets import QWidget from PyQt6.QtCore import Qt class WaylandHud(QWidget): def __init__(self): super().__init__() self.setWindowFlags( Qt.WindowType.FramelessWindowHint | Qt.WindowType.WindowStaysOnTopHint | Qt.WindowType.Tool | Qt.WindowType.WindowDoesNotAcceptFocus ) self.setAttribute(Qt.WidgetAttribute.WA_TranslucentBackground) self.setStyleSheet(""" background-color: rgba(15, 23, 42, 0.85); border: 1px solid rgba(255, 255, 255, 0.1); border-radius