Setup & Installation
Or with the ClawHub CLI, for registry-managed skill folders outside a full OpenClaw workspace:
Version History
telegram-voice-to-voice-macos v0.1.3 - Updated scripts/tts_telegram_voice.sh (details not shown). - No user-facing changes in SKILL.md.
What This Skill Does
Handles Telegram voice note conversations on macOS Apple Silicon. Transcribes incoming .ogg voice notes using Apple's Speech.framework via the yap CLI, generates a spoken reply with say and ffmpeg, and sends it back as a Telegram voice note. Supports per-user toggle between voice and text reply modes.
Uses Apple's on-device Speech.framework for transcription instead of a cloud API, so no audio leaves the machine.
When to use it
- Replying to Telegram voice messages hands-free on a Mac
- Transcribing received voice notes into readable text
- Running a voice-native Telegram bot on an Apple Silicon machine
- Switching specific contacts between voice and text reply modes
- Automating voice responses in a Telegram workflow on macOS
Example Workflow
Here's how your AI assistant might use this skill in practice.
User sends a Telegram voice note asking a question
- 1Locate the inbound .ogg file from ~/.openclaw/media/inbound/
- 2Run scripts/transcribe_telegram_ogg.sh to transcribe the audio using yap
- 3Check voice_state/telegram.json to determine the reply mode for this sender
- 4Generate a reply text, then run scripts/tts_telegram_voice.sh to produce a .ogg voice note
- 5Send the .ogg back to Telegram as a voice note with asVoice: true
A Telegram voice note reply sent to the original sender
Requirements
Accounts, API keys, or tools you or your AI assistant may need to set up while using this skill.