dsh-live-voice 0.0.2 → 0.2.0

package/CHANGELOG.md

@@ -0,0 +1,74 @@
+# Changelog
+
+All notable changes to DSH Live Voice are documented in this file.
+
+## [0.2.0] - Unreleased
+
+### Release Changes
+
+This release expands conversation control, audio routing, speech filtering, and settings organization.
+
+### Features
+
+- Add hands-free DSH structured questions in voice conversation mode: narrate each prompt, capture the next spoken response as a custom answer, submit it automatically, and keep the Live Voice status bar visible over the question panel.
+- Add audio input and output device preferences.
+- Add a three-state delivery mode for manual review, queued delivery, and immediate steering.
+- Add configurable speech-input filtering, including minimum-word filtering for final recognition results.
+- Add configurable speech-output filtering to omit code blocks from spoken responses.
+- Add exact voice commands for ending a conversation, muting or resuming listening, stopping speech, clearing input, and sending or queuing recognized text. Commands support multiple comma-separated phrases and normalize punctuation, case, and accents only while matching.
+- Add a remaining-speech segment count to the automatic speech control; it remains expanded while speech is queued.
+- Use the same newline-only segmentation and queue for automatic streaming speech and manual Speak actions.
+- Add tabbed Live Voice settings to organize speech, conversation, commands, and advanced options.
+
+### Changes
+
+- Improve the Live Voice controls and settings interface for delivery, device, filtering, command, microphone-input, and speech-queue states.
+- Expand coordinator, settings, component, and filter test coverage.
+- Update compiled client and server bundles for the new functionality.
+- Include this changelog in the published package.
+
+### Bug Fixes
+
+- Keep voice conversation mode active across chat navigation. When the current composer is replaced, the newly mounted chat automatically resumes voice mode; only an explicit End voice conversation action disables it. Composer drafts remain isolated per conversation, while internal controller disposal and hardware handoffs no longer count as user-requested conversation termination.
+- Add lifecycle regression coverage for switching chats while voice mode is active and for preserving an explicit end across subsequent chats.
+- Require at least one recognized word before headphone-mode microphone activity may pause assistant speech; audio activity alone no longer pauses playback.
+- Debounce headphone-mode interruptions to reduce false pauses from short recognition events.
+- Automatically resume speech when an interruption candidate ends without becoming valid user speech.
+- Preserve manual pause behavior separately from automatic interruption handling.
+
+## [0.1.0] - 2026-09-15
+
+### Bug Fixes
+
+- Reliably reset browser speech recognition after it ends or encounters an error.
+- Allow a custom HTTP or HTTPS base URL for the host-local Qwen3 speech service.
+
+## [0.0.2] - 2026-09-15
+
+### Features
+
+- Add host-local Apple MLX support for Qwen3-ASR and Qwen3-TTS.
+- Add Qwen TTS voice selection in Live Voice settings.
+- Expand Live Voice controls and their integration coverage.
+- Add continuous integration, a pre-push hook, formatting configuration, and distribution-artifact validation.
+
+### Changes
+
+- Update the conversation coordinator, microphone, recognition, synthesis, Whisper, and native macOS `say` integrations for the new engines and controls.
+- Include compiled client and server bundles in the published distribution.
+
+## [0.0.1-alpha.1] - 2026-09-15
+
+### Features
+
+- First functional release of the DSH Live Voice plugin.
+- Coordinate microphone capture, speech recognition, assistant-message delivery, and speech playback.
+- Add voice typing in the DSH composer and continuous voice conversations.
+- Support browser SpeechRecognition and authenticated loopback whisper.cpp HTTP recognition.
+- Support browser speech synthesis and native macOS `say` output.
+- Add Live Voice controls, Whisper settings, build and browser-preview scripts, and an initial test suite.
+
+[0.2.0]: https://github.com/victorwads/dsh-live-voice/compare/v0.1.0...HEAD
+[0.1.0]: https://github.com/victorwads/dsh-live-voice/compare/v0.0.2...v0.1.0
+[0.0.2]: https://github.com/victorwads/dsh-live-voice/compare/v0.0.1-alpha.1...v0.0.2
+[0.0.1-alpha.1]: https://github.com/victorwads/dsh-live-voice/tree/v0.0.1-alpha.1

package/README.md

@@ -100,7 +100,7 @@ Speech processing can run locally, but the DSH language model may still be remot
 
 ## Qwen3 HTTP engine
 
-When a compatible Qwen3 speech API is already running on the DSH host, choose **Qwen3 ASR — local MLX server** under Speech recognition and **Qwen3 TTS — local MLX server** under Speech output. Configure its loopback base URL in **DSH Settings → Live Voice**. The plugin supports the OminiX-API contract and standard OpenAI-style speech endpoints at `GET /health`, `POST /v1/audio/transcriptions`, and `POST /v1/audio/speech`; it does not install, start, stop, or manage that external service or its model weights.
+When a compatible Qwen3 speech API is already running on the DSH host, choose **Qwen3 ASR — local MLX server** under Speech recognition and **Qwen3 TTS — local MLX server** under Speech output. Configure its base URL in **DSH Settings → Live Voice**. The Qwen server may use any HTTP or HTTPS base URL reachable from the DSH host. The plugin supports the OminiX-API contract and standard OpenAI-style speech endpoints at `GET /health`, `POST /v1/audio/transcriptions`, and `POST /v1/audio/speech`; it does not install, start, stop, or manage that external service or its model weights.
 
 ## License