Sound FX (ElevenLabs)

Overview

Generate a sound effect from a text prompt using the ElevenLabs SFX API. Output is MP3 by default; convert to .ogg/opus for WhatsApp mobile playback.

Quick start

1) Set API key:

• ELEVENLABS_API_KEY (preferred) or XI_API_KEY

• Or set skills."sound-fx".env.ELEVENLABS_API_KEY in ~/.clawdbot/clawdbot.json

2) Generate SFX (MP3):

scripts/generate_sfx.sh --text "short audience applause" --out "/tmp/applause.mp3" --duration 1.2

3) Convert to WhatsApp-friendly .ogg/opus (if needed):

ffmpeg -y -i /tmp/applause.mp3 -c:a libopus -b:a 48k /tmp/applause.ogg

Script: scripts/generate_sfx.sh

Usage

scripts/generate_sfx.sh --text "canned laughter" --out "/tmp/laugh.mp3" --duration 1.5

Notes

• Uses POST https://api.elevenlabs.io/v1/sound-generation

• Supports optional --duration (0.5–30s). When omitted, duration is auto.

• Prints MEDIA: <path> on success for auto-attach.

Examples

• Applause: "short audience applause"

• Laughter: "canned audience laughter"

• Whoosh: "fast whoosh"

• Ambience: "soft rain ambience"