# Changelog OpenVoiceUI uses calendar versioning (`YYYY.M.D`; a second release on the same day gets `-N`). Every release is a git tag `vYYYY.M.D` on `main` with a GitHub Release that carries the full notes. This file is the index: one short summary per release, newest first, linking to the full notes. See [docs/RELEASING.md](docs/RELEASING.md) for how releases are cut. ## [Unreleased] - Repository organization pass: root docs moved into `docs/`, stale duplicates moved into `archive/`, README rewritten, community files added (Code of Conduct, this changelog, `.editorconfig`, issue-template config). See [docs/REPO-ORGANIZATION.md](docs/REPO-ORGANIZATION.md). ## [2026.9.29] - 2026-09-29 - Gateway reconnect window now covers a real gateway restart; boot warm-up retries until the gateway answers and the container entrypoint waits for the gateway port. - One-click vault Connectors (Supabase, GitHub, Netlify) with validate-before-save. - Suno submit refusals (for example, out of credits) are reported to the user instead of failing silently. - TTS fixes: the end of a reply is no longer cut off, no more double audio from interim TTS; push-to-talk works without a running call. - Auth and canvas hardening; least-privilege permissions on CI workflows. - Bundles OpenClaw `2026.5.7`. ## [2026.9.7] - 2026-09-07 - On-box speech-to-text (Parakeet) and Pocket TTS as a free local voice-clone provider. - Opt-in wake-word auto-arm with a tunable wake threshold; faster first audio. - Audio clip creator, camera capture with job albums, image splitter and grid creator rework. - Grok/xAI providers and the booking plugin. - Canvas pages locked by default; canvas data feeds require auth; upload CSP by file type. - Bundles OpenClaw `2026.5.7`. ## [2026.7.15] - 2026-07-15 - Security hardening pass: upload stored-XSS, OAuth callback CSRF and auth fail-open fixes. - Real Hume Octave TTS provider and an opt-in xAI Grok Realtime adapter. - Full Suno suite (stems, personas, voice cloning, uploads) with durable task-id records. - Canvas Styles (per-instance design-system picker) and a desktop/canvas overhaul. - Rebuilt admin panel. ## [2026.6.9] - 2026-06-09 - Resemble TTS hardening (chunking, timeouts, retries, no silent voice substitution). - `/api/auth/me` and `/api/history` endpoints. - Dependency updates. ## [2026.5.30] - 2026-05-30 - Z.AI-direct slow/empty-turn fallback; persona-driven greetings replace canned fallbacks. - Story engine API; mic mute held until the response stream ends. - Dependency updates and community contributions. ## [2026.5.4] - 2026-05-04 - Fixed a cost leak where one song request could fire multiple paid Suno calls. - Fixed bare "YES"/"NO" replies leaking to the user. - Made an optional backend import non-fatal so a fresh image build starts cleanly. ## [2026.4.28-1] - 2026-04-28 - Removed the robotic browser speech filler during retry cascades. ## [2026.4.28] - 2026-04-28 - `bump-openclaw-version.sh` keeps the OpenClaw pin in sync across the Docker, Pinokio and native installers; the Pi coding agent ships as the default coding CLI. - Deepgram marked optional in the installer (browser Web Speech is the default STT). ## [2026.4.19] - 2026-04-19 - Session recovery and interrupt handling: context-preserving recovery, empty-reply retries and steer-during-inference recovery. ## [2026.4.13] - 2026-04-13 - Deepgram and Web Speech push-to-talk reliability, server-side garbage-STT filter, Resemble TTS reliability. ## [2026.4.10] - 2026-04-10 - Plugin config API and settings panel, plugin catalog stubs, custom faces system, page-icon meta tags for canvas pages. ## [2026.4.7] - 2026-04-07 - Remote plugin catalog, conversation interject, external STT provider, text/voice mode toggle, sub-agent visibility, AI config admin panel. ## [2026.4.2] - 2026-04-02 - External (bring-your-own) STT provider; AI models and API keys admin panel; gateway auth and STT accumulation fixes. ## [2026.3.31] - 2026-03-31 - Text/voice mode toggle, sub-agent visibility, canvas screenshots, version display and update detection; Docusaurus documentation site. ## [2026.3.29-1] - 2026-03-29 - Re-publish of 2026.3.29 with an updated npm package. ## [2026.3.29] - 2026-03-29 - GLM-5-turbo default, admin panel overhaul, README overhaul; avatar content moved into the plugin system. ## [2026.3.28] - 2026-03-28 - Faster installs (heavy face-recognition dependencies made optional), clearer Docker errors, Dependabot and security scanning in CI, device-pairing fix for stock installs. ## [2026.3.27] - 2026-03-27 - First official release: one-click self-update, plugin system, canvas pages, voice studio and the initial TTS provider set. [Unreleased]: https://github.com/MCERQUA/OpenVoiceUI/compare/v2026.9.29...HEAD [2026.9.29]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.9.29 [2026.9.7]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.9.7 [2026.7.15]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.7.15 [2026.6.9]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.6.9 [2026.5.30]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.5.30 [2026.5.4]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.5.4 [2026.4.28-1]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.28-1 [2026.4.28]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.28 [2026.4.19]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.19 [2026.4.13]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.13 [2026.4.10]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.10 [2026.4.7]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.7 [2026.4.2]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/v2026.4.2 [2026.3.31]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/2026.3.31 [2026.3.29-1]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/2026.3.29-1 [2026.3.29]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/2026.3.29 [2026.3.28]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/2026.3.28 [2026.3.27]: https://github.com/MCERQUA/OpenVoiceUI/releases/tag/2026.3.27