Telegram Voice Pipeline Overview This skill handles Telegram voice messages end-to-end: 1. Detect voice message via in channel tag 2. Download OGG/Opus file via Telegram MCP tool 3. Transcribe with Whisper ( ) 4. Process transcribed text as the user query — generate a text response 5. Generate TTS audio via ElevenLabs (fallback: OpenAI TTS) → MP3 6. Reply with the MP3 audio file via Telegram MCP tool When to Invoke Invoke when: - A Telegram channel tag has an attribute (voice or audio message) - The channel agent needs to respond with audio rather than text - Any step in the voice pipeline ne…