Patience with a slow main, delivery diagnostics, steadier connection sensor
Recordings live in the recognition log; the Unknown tab is gone
Speech-to-text services: main, fallback chain, shadows (0.11)
Document Home Assistant STT entities as a backend Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Document the false-wake settings and what they cannot fix Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Document the media-ducking blueprint and why Murdock does not do it Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Document local Parakeet, the language hint, and the whisper noise floor Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Document the whisper state on MQTT, the integration, and assignment Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Correct the stale note about whisper profiles
Document the whisper score and whisper voice profiles
Document the experimental features and the profile-health coach
Document transcript correction, HACS install and the chip editor
Point the docs at the new settings groups
Document where to inspect the mirrored vocabulary
Document both delivery paths, the MQTT requirement and the timing fix
Fix mangled command in the API-check snippet
Document the 0.1.1 alias fix and the API verification script
Fix internal link to the transcript quality tiers
Document sidecar mode for ambiguity markers
Document the HA custom integration, margin gate and 0.8.0 event payload
Document transcript quality tiers
STT backends: add OpenRouter preset
Document STT backends: OpenAI-compatible, local fallback, A/B shadow
Clarify MQTT stays on in both speaker-context modes
Add Speaker Context page (delivery modes) + link from Home
Embed how-it-works diagram on Home + Architecture
Full wiki: installation, MQTT, training, tuning, backup, architecture
Initial Home page