elevenlabs-voices_skill

Overview

This skill provides high-quality voice synthesis using the ElevenLabs API, packaged as CLI scripts and utilities. It offers 18 curated voice personas, support for 32 languages, streaming output, batch processing, SFX generation, and a voice design tool. A guided setup wizard stores your API key locally and helps configure default voice, language, quality, and cost tracking. The tooling is designed to integrate with OpenClaw or run standalone from the command line.

How this skill works

The skill exposes small Python scripts that call ElevenLabs endpoints to synthesize speech, generate sound effects, and create custom voices. It inspects local configuration (environment variables, OpenClaw config, or a .env file) to find your API key and applies voice presets, pronunciation rules, and per-voice tuning parameters. Streaming mode emits audio in real time for long text, while batch mode reads newline or JSON lists to produce many files in one run. Usage statistics and character-based cost estimates are collected locally for budgeting.

When to use it

• Create narrated content quickly from scripts or articles
• Produce multilingual TTS for tutorials, announcements, or accessibility
• Generate short AI sound effects for games, videos, or prototypes
• Design and preview custom voice personas before saving them to ElevenLabs
• Batch-convert many lines of text into audio files for podcasts or e-learning

Best practices

• Run the interactive setup on first use to securely store your API key and set defaults
• Use presets for consistent voice selection across projects and to simplify CI automation
• Enable streaming for very long texts to reduce latency and memory footprint
• Keep pronunciation rules in pronunciations.json for brand names and acronyms
• Monitor the --stats output regularly and set a monthly budget cap to avoid unexpected costs

Example use cases

• Generate a narrated chapter: scripts/tts.py --text-file chapter.txt --voice storyteller --output chapter.mp3
• Produce product announcements in multiple languages by looping lang codes and voices
• Create background SFX for a scene: scripts/sfx.py --prompt "Thunder rumbling" --duration 4
• Design a new host voice then save it to your ElevenLabs library and preview: scripts/voice-design.py --description "Warm podcast host" --save MyHost
• Integrate with OpenClaw by enabling the elevenlabs provider in ~/.openclaw/openclaw.json