debpalash/VoiceStudio: VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages.

VoiceStudio is the open-source, fully-local ElevenLabs alternative — voice cloning, voice design, video dubbing, dictation, transcription & audiobook creation in 646 languages. - debpalash/Voic...

Summary

VoiceStudio is an open-source, local-first speech studio (Python backend with an Electron desktop shell) that clones voices, designs synthetic voices, dubs video, transcribes audio, dictates text, and produces audiobooks across 646 languages. The repository describes itself as a fully-local alternative to hosted services like ElevenLabs and is published under AGPL-3.0. The default engine is k2-fsa/OmniVoice, and additional engines can be selected from a catalog documented in docs/feature-catalog.md.

What it is useful for

The bundled workspaces cover the jobs engineers typically hand off to a hosted voice API: voice cloning from a reference recording, voice design from a text description, video dubbing with timed output, transcription, translation, dictation via a floating widget, and long-form audiobook or batch generation. It is useful when you need to keep audio processing on-prem, when you want to swap between TTS/STT engines without leaving the same workflow, or when you want to drive generation from scripts and coding agents. The README notes that local workflows run on your hardware and that remote services are optional, with usage analytics requiring explicit consent.

How engineers can use it

The README documents three install paths. The hosted installer is the shortest:

# macOS / Linux
curl -fsSL https://voicestudio.sh/install | sh
# Windows PowerShell
irm https://voicestudio.sh/install | iex

Prebuilt binaries (.dmg, .exe, .AppImage, .deb) are linked in the README under the platform guides for macOS, Windows, Linux, and Docker. To run the Electron preview from source:

git clone https://github.com/debpalash/VoiceStudio.git
cd VoiceStudio
bun install
bun run setup:api
bun run dev

For agent integrations, the README points to docs/install/agent.md and an npx skills add debpalash/VoiceStudio skill, with a Local API and MCP documented under docs/speech-platform.md and docs/mcp.md.

Hardware sets the experience: NVIDIA GPUs use CUDA, Apple Silicon uses Metal (MPS), and machines without a discrete GPU fall back to a CPU build that downloads about 5 GB of PyTorch and runs slower. The README flags Windows on ARM as experimental (CPU only under emulation) and Intel Mac as supporting the app shell only, which must connect to a remote backend. Electron is the only supported desktop shell — Tauri was retired at v0.5.3, and the legacy Tauri entry points have been removed, so existing installs need a separate Electron install per docs/electron-migration.md. Because the project is AGPL-3.0, any modifications you distribute must stay under compatible terms, and bundled models carry their own licenses that the README says to review before commercial use.