VoiceStudio: local voice cloning, dubbing and transcription
Added NewCopyable prompt below
Published Last verified Released
Our verdict
A good fit when you want voice work without per-character API bills or uploading recordings: it bundles several open speech engines (OmniVoice by default) behind one Electron app, local API and MCP server. It is mainly a studio around those engines, so quality depends on the engine, language and your GPU, and on CPU it is usable but slow. The app is AGPL-3.0, but model weights and voices carry their own licences, so check them before commercial use, and clone voices only with permission.
- Who it's for
- Builders and creators who want local TTS, dubbing or dictation, or want an agent to speak in a chosen voice
- Leverage
- Local voice generation without per-character API fees
- Effort
- Medium
- Date added
What it does
VoiceStudio (debpalash/VoiceStudio, site voicestudio.sh) is a desktop app for macOS (Apple Silicon), Windows and Linux, plus Docker. It covers voice cloning from a reference clip, voice design from a text description, timed video dubbing, a floating dictation widget, transcription, and long-form audiobook and batch jobs. The running backend exposes a local API and an MCP server at /mcp with tools such as generate_speech, clone_voice, design_voice and transcribe. Models download on first use (the default OmniVoice model is about 2.3 GB). Usage analytics require consent.
How builders use it
- Install the latest release from the official repo or voicestudio.sh (DMG, EXE, AppImage or .deb), or use the one-line installer after reading it.
- Open Voice cloning, add a clean reference recording you have permission to use, and generate a test line.
- Point your agent at the local MCP endpoint (http://127.0.0.1:3900/mcp/ by default; the app exports ready-made client configs) and set OMNIVOICE_MCP_BASE_PATH to limit file access.
- Keep the MCP endpoint on localhost: the docs say its transport is not authenticated.
What to expect
Voiceovers, dubbed clips and transcripts produced on your own hardware, callable from your agent.
Copy this: Install + agent ask
# Official sources only: github.com/debpalash/VoiceStudio, voicestudio.sh
# macOS / Linux (read the script first):
curl -fsSL https://voicestudio.sh/install | sh
# Windows (PowerShell):
irm https://voicestudio.sh/install | iex
# Optional agent skill:
npx skills add debpalash/VoiceStudio
# Local MCP endpoint (app running):
# http://127.0.0.1:3900/mcp/
Ask your agent:
Using the VoiceStudio MCP server, list my voices, then generate a 20-second WAV of <script> in <voice> and save it under my base path.Related discoveries
- ToolCreative & Media
ComfyUI
Open-source node-graph engine for image, video and audio generation that runs locally and exposes an API.
NewEffort: High#Image#Video
- ToolCreative & Media
Chatterbox TTS
Open-source text-to-speech models from Resemble AI with voice cloning, installable with pip.
NewEffort: MediumPrompt#Audio#Local / BYO model
- ToolCreative & Media
Invoke (InvokeAI)
An open-source, locally run image studio with a canvas, inpainting and node workflows for open models.
NewEffort: Medium#Image#Local / BYO model
- ToolCreative & Media
LTX-2: open audio and video generation
Lightricks' open-weights model for video with synchronized audio, runnable locally or inside ComfyUI.
NewEffort: High#Video#Audio
Want more like this?
Practical discoveries, without the noise.