# VoiceStudio > Open-source, fully-local desktop app for voice cloning, voice design, video dubbing, dictation, transcription, and audiobook creation in 646 languages. VoiceStudio is an open-source, fully-local voice AI desktop application built by Palash Dev and released under the AGPL-3.0 license. It runs entirely on your own hardware — no API keys, no cloud account, and no data leaving your machine — and supports 646 languages across its core workflows. The project was formerly known as OmniVoice Studio before a name change documented on the project site. ## What It Is VoiceStudio is a local-first voice workflow engine that bundles seven distinct capabilities into a single Electron desktop application: voice cloning, voice design, video dubbing, multi-voice story narration, audiobook creation, a voice gallery, and audio/video transcription. It positions itself as a self-hosted alternative to cloud-based voice AI services, giving users full control over their voice data and no usage counters. The core platform is written primarily in Python and ships as a downloadable installer for macOS, Linux, and Windows (via WSL), as well as a Docker image. ## Core Workflows Each workflow in VoiceStudio targets a distinct production task: - **Voice clone** — Mirror a voice from a short reference clip; the project states three seconds is typically sufficient. - **Voice design** — Describe a new voice in a sentence, specifying gender, age, accent, pitch, and emotion. - **Video dubbing** — Transcribe, translate, and re-voice video while preserving per-speaker identity and timing alignment. - **Stories** — Cast characters and generate multi-voice audiobooks from a script. - **Audiobook** — Convert long scripts or EPUB files into chaptered audiobooks. - **Transcripts** — Turn audio or video into editable, searchable text. - **Voice gallery** — Browse and preview ready-made designed voices filtered by accent, age, and style. ## Local API and Agent Integration VoiceStudio exposes an OpenAI-compatible local API at `http://localhost:3900`, documented under contract version `0.5.0`. Developers can query `/openapi.json` to discover available routes covering audio synthesis, model management, files, diagnostics, settings, and application administration. The project also ships an MCP (Model Context Protocol) server and supports agent skill installation via `npx skills add debpalash/VoiceStudio`, making it composable with AI coding agents and multi-agent workflows. A hosted Cloud API is listed as a preview contract (`1.0.0`) with no production endpoint currently active. ## Architecture and Deployment The application runs as an Electron desktop app backed by a local Python server. Users can install via a shell script (`curl -fsSL https://voicestudio.sh/install | sh`), platform-specific installers from GitHub Releases, or Docker Hub and GHCR container images. The engine layer is swappable — the default engine is powered by k2-fsa/OmniVoice, and additional engines can be installed from the in-app Model Catalogue. Hardware acceleration via CUDA and MLX is supported where available, with performance varying by engine and device. Remote worker nodes are an optional extension for teams needing distributed compute. ## Update: v0.5.3 — Electron Migration The latest release as of September 2026 is **v0.5.3**, titled "A new look. Now on Electron." Version 0.5.3 introduced the Electron-based desktop application and was simultaneously the final Tauri release. Existing Tauri users must install the Electron app separately and follow the documented migration guide. The local API contract is versioned at `0.5.0`. The GitHub repository shows active development with the last push on September 18, 2026, and the project has accumulated over 32,900 GitHub stars and more than 221,000 release downloads according to figures published on the project homepage. ## Open-Source Deployment Model VoiceStudio is licensed under AGPL-3.0, meaning the source code is publicly available and modifications must be shared under the same terms when deployed as a network service. Personal use is free with no usage limits. Commercial use requires reviewing both the application license and the individual model licenses, which vary by engine. A Pro tier for commercial rights and a hosted Cloud offering are listed on the website as enquiry-only and early access respectively, with no public pricing published at this time. ## Features - Voice cloning from short reference clips - Voice design from text description (gender, age, accent, pitch, emotion) - Video dubbing with per-speaker identity and timing alignment - Multi-voice story and audiobook creation - EPUB to chaptered audiobook conversion - Audio and video transcription - Voice gallery with accent, age, and style filters - OpenAI-compatible local API at localhost:3900 - MCP server for agent integration - Swappable speech engine architecture - In-app Model Catalogue with install actions - CUDA and MLX hardware acceleration support - Docker deployment support - 646 language support - Fully offline — no API keys or cloud account required - Agent skill installation via npx skills add ## Integrations Docker, Hugging Face, CUDA, MLX, OpenAI-compatible API clients, MCP (Model Context Protocol), k2-fsa/OmniVoice engine, Docker Hub, GHCR (GitHub Container Registry) ## Platforms WINDOWS, MACOS, LINUX, WEB, API, CLI ## Pricing Open Source, Free tier available ## Version v0.5.3 ## Links - Website: https://voicestudio.sh - Documentation: https://voicestudio.sh/docs - Repository: https://github.com/debpalash/VoiceStudio - EveryDev.ai: https://www.everydev.ai/tools/voicestudio