Mimikastudio, developed by Boltzmannentropy, is an MCP server application for macOS, built with Flutter (Dart) and a FastAPI Python backend. It operates as a local-first agentic voice cloning server, running entirely on-device without external cloud service reliance. This free program offers advanced text-to-speech and voice management capabilities, targeting content creators and authors. Its design emphasizes on-device processing for privacy and performance, making it a powerful tool for local audio production.

What does Mimikastudio connect to?

Mimikastudio functions as a local-first application and an agentic voice cloning server, categorized as an MCP server. It runs fully on-device, ensuring its core functionality does not depend on external cloud services. The application supports HTTP as a transport protocol, demonstrated by curl examples interacting with http://127.0.0.1:8010. It is compatible with AI clients such as Claude Code CLI, Codex, and any other MCP-compatible client, allowing for diverse integration possibilities within existing workflows.

Core features and tools

The MCP server in Mimikastudio exposes over 50 tools, covering various functionalities. These include TTS generation tools like Kokoro, Qwen3, and Chatterbox. Voice management tools allow users to list, upload, delete, update, and preview voices. It also provides tools for audiobook generation and management, system info and monitoring, and LLM configuration. A specific callable tool example is tts_list_voices. The application further offers more than 60 REST API endpoints, integrated as MCP tools for extended capabilities.

Who is it for?

Mimikastudio primarily serves authors aiming to create audiobooks from manuscripts and users interested in local-first voice cloning, text-to-speech, and content creation. It supports converting full documents into audiobooks with queueable chapter generation. Users can clone any voice from short audio samples and generate high-quality text-to-speech. This is not the right fit if you require a cloud-dependent solution or a fully managed service.

How to get started

No credentials are required for Mimikastudio's core functions, except for an optional LLM for IPA transcription. The application supports HTTP as its transport protocol, allowing interaction via standard HTTP requests.

Use cases

• Clone any voice from as little as 3 seconds of reference audio.
• Generate high-quality text-to-speech and read documents aloud with sentence-level highlighting.
• Convert full documents (PDF, DOCX, EPUB, Markdown, TXT) into audiobooks with queueable chapter generation.
• Turn a YouTube clip into a reusable voice prompt.

Mimikastudio provides a local-first solution for voice cloning and audiobook creation, offering a rich set of tools for content generation. Its on-device operation ensures privacy and independence from external services. The application supports various document formats and integrates with MCP-compatible clients.