Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
What This Skill Does
Local speech-to-text and text-to-speech on macOS using Apple's native Speech.framework and say command. All processing runs on-device with no network calls or API keys. Supports multiple languages, voice quality tiers, and audio output in ogg/opus or aiff formats.
Unlike cloud STT/TTS APIs, this runs fully offline with no usage costs, no data leaving the device, and no account or key setup required.
When to use it
- Transcribing recorded voice memos without uploading to a cloud service
- Having long documents read aloud while multitasking
- Sending voice notes in chat apps from generated audio files
- Dictating in Chinese, Japanese, or other languages with locale-specific recognition
- Building offline voice interfaces that work without internet
Example Workflow
Here's how your AI assistant might use this skill in practice.
User asks: Read this paragraph aloud in French
- 1Run voices.mjs best fr_FR to find the highest quality available French voice
- 2Run voices.mjs check to confirm the voice is downloaded and ready
- 3Run tts.mjs with the text and selected voice to generate an audio file
- 4Send the generated file path as a voice note using the message tool
Audio file generated at ~/.openclaw/media/outbound/tts_<timestamp>.ogg and sent as a voice note
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.