elevenlabs-tts

What This Skill Does

The elevenlabs-tts skill brings state-of-the-art text-to-speech capabilities to the OpenClaw agent ecosystem. Leveraging ElevenLabs v3, this skill transforms plain text into hyper-realistic, emotionally rich audio. It is designed to handle multilingual support, nuanced delivery, and specific audio tags to simulate human-like pauses, breaths, and emotional states. Whether for automated customer interactions, content creation, or accessible messaging, this skill ensures your AI agent communicates with natural inflection.

Installation

To install this skill, use the ClawHub CLI command: clawhub install openclaw/skills/skills/shaharsha/elevenlabs-tts. After installation, ensure you have ffmpeg installed on your system as it is a critical dependency for converting ElevenLabs output into Opus format for WhatsApp and other communication platforms. Finally, update your openclaw.json configuration file with your unique ELEVENLABS_API_KEY, preferred voiceId, and modelId settings.

Use Cases

This skill is perfect for dynamic communication tasks. It is widely used for creating personalized audio messages in WhatsApp bots, generating expressive voice-overs for storytelling applications, and crafting professional automated responses. It excels in scenarios where monotone, robotic voices fall short, such as emotional storytelling, multilingual announcements, and building immersive user experiences where the tone of the voice needs to shift dynamically based on the content of the message.

Example Prompts

1. "Speak the following script with an excited tone: [excited] We are thrilled to announce that our new update is live! [laughs] It features everything you asked for."
2. "Generate a suspenseful audio clip for a story: [whispers] The basement was freezing... [pause] [scared] I think something is standing right behind me."
3. "Convert this message to Spanish audio: [soft] Muchas gracias por tu paciencia. [happy] Estamos muy contentos de ayudarte hoy."

Tips & Limitations

For optimal results, experiment with the stability and similarityBoost parameters in your configuration. High stability settings make the voice more consistent but potentially flatter, while lower settings allow for more emotional variance. Ensure you have sufficient credits in your ElevenLabs account. Note that complex emotional tags are most effective when using the v3 model, and overly dense tag usage may occasionally disrupt natural cadence if not spaced correctly with standard punctuation.