How do I install the necessary dependencies?

The skill utilizes 'uv' with the '--with' flag to manage mlx-audio and pydub dependencies in a cached environment, ensuring a clean installation without global conflicts.

Does this skill require an internet connection?

Only for the initial run to download the ~200MB model from Hugging Face; all subsequent audio generation is performed entirely locally and privately on your machine.

What hardware is required for optimal performance?

This skill is specifically optimized for Apple Silicon (M1, M2, M3 series) using the MLX framework for hardware-accelerated inference.

Are there different voice options available?

Yes, the skill includes a wide range of American and British English presets (both male and female), such as 'af_heart' for warm tones and 'am_adam' for professional narration.

Can I convert large text files into audio?

Yes, the skill includes automatic text chunking and seamless concatenation via pydub to handle long documents and scripts effectively.

Local Text-to-Speech

Name: Local Text-to-Speech
Author: krishagel

bykrishagel

•

数据科学与机器学习

Generates high-quality speech audio locally on Apple Silicon using MLX acceleration and the Kokoro model.

This skill enables Claude to perform high-fidelity text-to-speech synthesis directly on your local machine without relying on external APIs or incurring usage costs. By leveraging Apple Silicon’s MLX framework and the efficient Kokoro-82M model, it provides fast, private audio generation for everything from morning briefings to automated content creation. The skill supports a variety of natural-sounding voice presets across different accents and genders, making it an ideal solution for developers building voice-enabled applications or private automated audio workflows within their local environment.

主要功能

01Hardware-accelerated audio synthesis via Apple Silicon MLX

02High-quality Kokoro-82M model with natural-sounding voice presets

03Privacy-focused local execution with zero recurring API costs

04Automated text chunking for seamless long-form audio generation

05Support for multiple output formats including MP3, WAV, and M4A

063 GitHub stars

使用场景

01Generating personalized audio podcasts from text-based morning briefings

02Creating high-quality voiceovers for documentation and tutorials

03Adding offline speech capabilities to local AI agents and automation scripts

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add krishagel/geoffrey local-tts

For use in Claude.ai and ChatGPT

Download Skill