voice--v0.0.4

This release significantly improves wake-word detection reliability: a pre-roll buffer prevents clipping the start of utterances, a sliding-window fuzzy matcher handles filler words and per-word edit tolerance, Whisper is biased toward your configured trigger vocabulary, and loudness is normalized before transcription to help quiet Bluetooth/HFP captures.

Changed

• dispatcher: trigger matching — sliding window + per-word tolerance — the matcher now tries up to three starting offsets so filler words prepended by Whisper ("um hey jarvis") don't prevent a match. Tolerance is now one allowed edit per trigger word (previously a character-count heuristic), making it more predictable across short and long triggers.
• dispatcher: default Whisper model tiny → base — better transcription accuracy out of the box at a modest CPU cost. Override with whisper.model: tiny in config.yaml if needed on low-power hardware.

Added

• dispatcher: pre-roll audio buffer — 320 ms of audio is buffered before VAD onset so the first syllable of an utterance is never clipped before Whisper sees it.
• dispatcher: Whisper initial-prompt biasing — the configured trigger phrases are fed to Whisper as an initial_prompt, nudging the model toward your vocabulary and reducing mis-transcriptions of trigger words.
• dispatcher: normalize_gain config option — loudness-normalizes captured audio before transcription (default true). Particularly effective for AirPods and other Bluetooth sources that capture at low gain. Disable with audio.normalize_gain: false in config.yaml.
• dispatcher: trigger_tolerance config option — override the per-word edit tolerance globally (audio.trigger_tolerance: N in config.yaml). Leave unset to use the automatic one-edit-per-word default.

Upgrade Instructions

Upgrade dispatcher only — no plugin changes in this release. No re-pairing required. The default Whisper model changes from tiny to base; if transcription feels slower on your hardware, add whisper: {model: tiny} to config.yaml to revert.

voice--v0.0.3

This release brings full macOS compatibility: the entire setup flow — pairing, TLS cert pinning, and project-local state isolation — now works on macOS without manual workarounds. Both platforms also gain a new trigger-beep feature.

Changed

• compatibility: macOS support — voice-dispatcher and /voice:configure are now fully tested and supported on macOS; pairing, cert-pinning, and state-isolation all work out of the box.

Added

• dispatcher: trigger beep — plays the OS notification sound immediately when a trigger phrase is matched, giving the operator an instant "got it" cue before TTS begins. Uses afplay on macOS, pw-play/paplay on Linux, and a synthesized 880 Hz tone via sounddevice as a universal fallback. Enabled by default; disable with notifications.trigger_beep: false in config.yaml.

Fixed

• plugin: configure macOS PEM fix — decode and verify the pairing string in a single bun command so the cert check never hand-parses the PEM, eliminating the spurious "pem has lines of non-standard length" error on macOS.
• plugin: project-local state dir — /voice:configure now pins VOICE_STATE_DIR to <project>/.claude/channels/voice in the project's settings.local.json, automatically isolating each project's token and gitignoring the state dir so the token can't be committed by accident.
• plugin: error hardening — a malformed pairing-string paste now emits a clean ERROR line (no stack trace); the status skill and README explain the "works but shows not configured" case for installs that predate project-local state dirs.

Upgrade Instructions

Upgrade both dispatcher and plugin together. No re-pairing is required.

To gain per-project token isolation, re-run /voice:configure in each project — it will write a fresh settings.local.json entry pointing to a project-local state dir. Existing setups that skip this continue to work from the old global state dir.

voice--v0.0.2

Changed

• security: in-band cert pinning — v2 pairing strings now include the dispatcher public cert PEM, and the plugin validates WSS using Bun's TLS ca path before sending the bearer token. This removes the previous two-connection fingerprint preflight and closes the token-leak TOCTOU window.
• plugin: v2 config — secure wss:// configs now require dispatcher_cert_pem plus matching dispatcher_cert_sha256; legacy fingerprint-only configs fail closed with re-pair guidance.
• dispatcher: auth failure logging — token rejection now logs the claimed agent ID so operators can diagnose misconfigured agents without enabling debug mode.
• plugin: permanent 4001 close — the plugin no longer retries after a 4001 Unauthorized close; it writes a clear error with fix instructions to status.json and exits cleanly, preventing reconnect storms on bad tokens.

Upgrade Instructions

Upgrade both dispatcher and plugin, then re-pair every agent with a fresh voicepair_... string from voice-dispatcher config rotate-token <id> or voice-dispatcher config add-agent <id> ....

voice--v0.0.1

Initial public release.