MimikaStudio

MimikaStudio is a local-first, on-device voice cloning and TTS platform that runs with an exposed MCP API and a companion UI. It supports cloning voices from short reference samples, generating expressive text-to-speech output, and handling document reading and audiobook workflows entirely on your machine. The MCP interface lets you automate tasks like initiating a voice clone, starting a TTS render, or exporting an audiobook package through API calls, enabling you to chain these operations in scripts or external automation tools. The included dashboard provides model management, run history, and configuration controls for model selection, voice presets, and output settings, making it suitable for batch processing and integration with other local services or pipelines. Use the MCP endpoints to trigger cloning, TTS generation, and document-to-audiobook conversions, all while staying fully local to preserve privacy and minimize latency on Apple Silicon devices.

Prerequisites:

• macOS with Apple Silicon (minimum recommended)
• Docker installed and running (for the recommended docker-based setup) or use a compatible local runtime if provided
• Optional: a local UI or API client to interact with the MCP endpoints

Installation steps:

1. Install Docker from https://www.docker.com/products/docker-desktop and start the application.
2. Pull or build the MimikaStudio MCP image (the repository provides a binary/Tar or a docker image). If using Docker, run: docker pull mimikastudio/boltzmannentropy-mimikastudio:latest
3. Run the MCP server container: docker run -it --rm -p 8080:8080 mimikastudio/boltzmannentropy-mimikastudio:latest This will start the MCP API and UI endpoints on localhost:8080 by default.
4. If you prefer a non-Docker setup (where available), follow the platform-specific binary distribution instructions provided in the repository (download the macOS binary, give it execute permissions, and run the server locally).
5. Verify the server is up by visiting http://localhost:8080 or by querying the MCP endpoints using your preferred HTTP client.
6. Configure authentication/ENV as needed (see environment notes) and begin using the MCP API for cloning, TTS, and document workflows.

Notes and tips:

• This project is optimized for macOS Apple Silicon with native acceleration; performance is best on newer Macs with MLX capabilities.
• If you are using the Docker image, ensure your host has enough GPU/CPU resources allocated for the model workloads.
• Environment variables can be used to control API keys, ports, and model presets; see the mcp_config example for placeholders.
• Windows support is planned; current binaries are macOS-only. If you need to run on Windows, monitor the project for official guidance and alternate build options.
• When running locally, you can leverage the MCP dashboard to manage model downloads, presets, and automation workflows. Ensure you download and store models in accessible locations on your machine.
• If you encounter issues with model downloads, check network access and verify that the on-device storage paths used by MimikaStudio have the correct permissions.