Skill
Text-to-speech via ElevenLabs — turn text into natural voice audio (Hebrew + English). Use for voice summaries to Telegram, video/Reels voiceovers, audio versions of articles, or any “read this aloud / make a voiceover” request.
Playbook (mirrored from disk)
ElevenLabs TTS
Account tier: Creator (131k chars/month). API key: elevenlabs in ~/.openclaw/secrets.json.
Generate speech
EL_KEY=$(python3 -c "import json; print(json.load(open('/home/yossef7875/.openclaw/secrets.json'))['elevenlabs'])")
curl -s -X POST "https://api.elevenlabs.io/v1/text-to-speech/<VOICE_ID>?output_format=mp3_44100_128" \
-H "xi-api-key: $EL_KEY" -H "Content-Type: application/json" \
-d '{"text":"<the text>","model_id":"eleven_v3"}' \
-o /home/yossef7875/.openclaw/media/outbound/<name>.mp3Voices (multilingual — all speak Hebrew)
| Voice | ID | Best for |
|---|---|---|
| George — warm storyteller | JBFqnCBsd6RMkjVDRZzb | summaries, narration (Jarvis default) |
| Sarah — mature, confident | EXAVITQu4vr4xnSDxMaL | marketing voiceovers, Reels |
| River — relaxed, neutral | SAz9YHcvj6GT2YYXdXww | informative content |
| Laura — quirky, energetic | (query /v1/voices) | social media hooks |
More voices: curl -s https://api.elevenlabs.io/v1/voices -H "xi-api-key: $EL_KEY".
Deliver to Telegram
Media must live under ~/.openclaw/media/:
openclaw message send --channel telegram --target 1121942605 \
--media /home/yossef7875/.openclaw/media/outbound/<name>.mp3 -m "<caption>"(SocialRise: remember the dual-delivery rule — also send to telegram:640770095.)
Rules
- HEBREW REQUIRES
eleven_v3— it is the ONLY model that supports Hebrew (verified via /v1/models).eleven_multilingual_v2produces gibberish for Hebrew (learned 2026-06-12). English can use either. - Mind the quota: 131k chars/month shared across ALL agents. A 1-minute voiceover ≈ 700 chars.
Don’t TTS long articles wholesale — summarize first, or ask Yossef. - Voiceover for Remotion videos: generate mp3 → place in the project’s assets → sync timing.
- Music/SFX (background tracks, sound effects) → separate skill
elevenlabs-audio.