Telegram Voice → Text (STT)

When a user sends a voice note on Telegram, OpenClaw transcribes it before the model runs.

How it works

1. Telegram delivers .ogg voice file to .openclaw/media/inbound/
2. Gateway runs transcribe-audio-cli.js (Google Cloud Speech-to-Text)
3. Transcript replaces <media:audio> in the user message
4. Optional echo: user sees 📝 Heard: "..." in Telegram
5. Assistant answers the transcribed text via Bifrost → genapp-proxy → Dialogflow

Rules for the assistant

• Never say "I cannot process audio" if a transcript is present
• Reply to the transcript content as the user's actual question
• For Hindi/English mix, answer in the language the user spoke or ask which they prefer
• If transcription is empty, ask the user to type their message or re-record

Stack

• OpenClaw: tools.media.audio CLI model → ~/.openclaw/gcp/transcribe-audio-cli.js
• Fallback: genapp-proxy also transcribes voice before Dialogflow (8192 token limit)

Manual test (gateway machine)

node "$env:USERPROFILE\.openclaw\gcp\transcribe-audio-cli.js" "C:\path\to\voice.ogg"