Kokoro Local Text-to-Speech

About

This skill provides a powerful interface for local text-to-speech synthesis, eliminating the need for cloud APIs or internet connectivity. Leveraging the Kokoro-82M model and ONNX runtime, it offers high-performance audio generation that is specifically optimized for Apple Silicon, achieving speeds up to 50x real-time. Whether you are converting short snippets of text or rendering entire books into seamless audio files, this tool handles over 60 voices across eight languages with intelligent chunking to prevent artifacts, making it an essential utility for private and efficient voice synthesis.

Key Features

• Access to a library of over 60 professional-grade voices across eight languages.
• 1 GitHub stars
• High-speed local synthesis reaching 20-50x real-time rendering on Apple Silicon.
• Privacy-focused offline execution using ONNX runtime without API requirements.
• Intelligent long-form rendering for creating seamless audiobooks from markdown files.
• Customizable speech parameters including speed control, voice selection, and silence intervals.

Use Cases

• Integrating private, local voice synthesis into automated developer workflows.
• Generating offline audio versions of long technical documentation or articles.
• Creating high-quality audiobooks from text-based manuscripts without cloud costs.