OpenClaw Voice Control

openclaw-voice-control 0.1.9 - Expanded SKILL.md documentation to add instructions for changing the macOS TTS voice. - Clarified that the user must ensure their preferred voice is downloaded via System Settings before selection. - No changes to code or user-visible features; documentation only.

- Reworded setup instructions: "foreground validation" is now "direct-run validation" - Clarified requirement to stop direct-run tests before proceeding with background deployment or additional validation - Added warnings about running multiple simultaneous voice runtimes and the risks of duplicate replies - No code or logic changes; documentation updated only

- Documentation updated in SKILL.md; text edits and clarifications. - No code or functional changes, only documentation changes in this release.

- Updated environment and binary requirements in the `metadata` section, adding variables such as `OPENCLAW_BASE_URL`, `SENSEVOICE_MODEL_PATH`, and `SENSEVOICE_VAD_MODEL_PATH`, and including `launchctl` as a required binary. - Added new instruction to explicitly ask for user confirmation before performing any system-changing steps (e.g., repository fetch, pip install, model downloads, enabling launchd/auto-start). - Clarified default route variable requirements for `.env` with a new section listing essential environment variables for setup. - Improved instructions on handling optional Porcupine wakeword integration, emphasizing not to prompt for Porcupine-specific variables unless requested. - Enhanced clarity on not assuming system state or reusing old environments or assets without user input.

openclaw-voice-control 0.1.5 - Updated setup instructions: Foreground validation now requires launching both the voice service and overlay process. - Added command examples for running both processes simultaneously during foreground validation. - Clarified that both commands must be run from the same installed skill workspace for proper validation. - Foreground validation steps and execution rules now explicitly mention starting both main and overlay processes before enabling background behavior.

- Changed default wakeword provider to openWakeWord and updated all instructions and prerequisites accordingly. - Added detailed main installation path using openWakeWord, including SenseVoice and VAD model download steps. - Made Picovoice/Porcupine route optional and documented only as a fallback, not the default. - Added a safety warning section about voice access and recommended security practices. - Clarified daily maintenance commands and background deployment flow. - Improved execution rules, setup prerequisites, shutdown intents, and configuration guides for clarity and security.

- Added homepage and metadata fields to SKILL.md, documenting repository, supported OS, required binaries, environment variables, and primary token variable. - No functional or behavioral changes to the skill logic. Documentation and metadata only.

openclaw-voice-control v0.1.2 - Clarified documentation requirements for model and wakeword files in SKILL.md - Updated language to specify that SenseVoice and VAD model directories should be downloaded or populated per repository setup, not guessed manually - No functional or code changes; documentation only

- Significant repository cleanup: removed 37 files, leaving only SKILL.md in the skill workspace. - Installation, configuration, and operation are now strictly gated on having the full repository synced—not just SKILL.md. - Instructions and execution rules in SKILL.md have been updated to emphasize that partial copies are not sufficient and to reinforce starting from a complete repository sync. - Install and deployment steps now require validation of the full repository presence before proceeding.

Initial release of OpenClaw Voice Control for macOS, providing local voice access integration for OpenClaw agents. - Adds wakeword activation, local ASR via FunASR/SenseVoice, and macOS TTS playback. - Supports overlay UI and background resident mode using launchd with auto-start after login. - Requires user-provided wakeword/model files and secrets; does not bundle sensitive assets or tokens. - Documentation details strict installation, deployment, and token handling rules. - Includes scripts for deployment, service management, and troubleshooting. - Foreground/manual operation and background/auto-start are both supported, with choice prompted during setup.