voice-notes

name	voice-notes
description	Send voice notes and transcribe incoming audio. Use when a user sends a voice note, asks for a voice reply, or wants text read aloud. Combines local Whisper transcription with ElevenLabs TTS.
metadata	{"clawdbot":{"emoji":"🎙️","requires":{"bins":["sag","bash"]}}}

Voice Notes

CRITICAL: How to send voice notes

DO NOT use the built-in tts tool. It generates audio but NO_REPLY blocks delivery — the user will never receive the voice note.

ALWAYS use sag via exec, then include MEDIA:/path in your reply text. This is the ONLY method that works for WhatsApp delivery.

Correct pattern (ALWAYS use this):

sag -o /tmp/voice-reply.ogg "Your message here"

Then in your reply text, include:

MEDIA:/tmp/voice-reply.ogg

You MUST include MEDIA:/path in your actual text reply. Do NOT reply with NO_REPLY after generating audio — that kills delivery.

For long messages, split into parts:

If the text is longer than ~500 characters, split into 2-3 shorter voice notes:

sag -o /tmp/voice-part1.ogg "First part of the message..."

Reply with: MEDIA:/tmp/voice-part1.ogg

Then:

sag -o /tmp/voice-part2.ogg