# VoiceStudio — card for AI agents

Local desktop app for voice cloning, voice design, video dubbing, dictation and transcription in 646 languages, with a local API and an MCP server for agents.

Category: Voiceover (text-to-speech) · maker Palash Debnath (Yupcha Softwares) · checked 2026-10-17

## Price
- Price: Free
- Free tier: The AGPL-3.0 app is free and runs on your own hardware (CUDA, Apple Silicon or CPU).
- Usage price: Optional Pro build (Commercial License, closed source): $99 per user a year or $299 once.
- Pricing page: https://voicestudio.sh/pro

## License
AGPL-3.0 (app); default OmniVoice weights CC-BY-NC — https://github.com/debpalash/VoiceStudio/blob/main/LICENSE-NOTICE.md

- commercial: **conditions** (Commercial use with conditions)
  > Commercial use under the AGPL-3.0 is free.
  (https://github.com/debpalash/VoiceStudio/blob/main/LICENSE-NOTICE.md)
- attribution: **varies** (Credit depends on the item)
  > Model weights, tokenizers, and other third-party assets retain their own terms.
  (https://github.com/debpalash/VoiceStudio/blob/main/LICENSE-NOTICE.md)
- in app: **conditions** (Ship inside your app with conditions)
  > the complete corresponding source code of your modified version
  (https://github.com/debpalash/VoiceStudio/blob/main/LICENSE-NOTICE.md)
- youtube: **conditions** (Monetized YouTube with conditions)

## Rules to follow
- The app is AGPL-3.0: commercial and internal use are allowed, but a modified version offered over a network must ship its source.
- Model weights keep their own terms: the default k2-fsa/OmniVoice pretrained model is CC-BY-NC (non-commercial).
- Clone voices only with permission.

## How an agent gets files
- Local desktop app: API at http://127.0.0.1:3900 (capabilities at GET /.well-known/voicestudio-speech); MCP over HTTP at POST /mcp/ or stdio via python -m backend.mcp_shim. (sign-in: none)
- Not allowed: App is AGPL-3.0: a modified version offered over a network must give users its source.
- Not allowed: Model weights keep their own terms; the default OmniVoice weights are CC-BY-NC.
- Not allowed: A separate paid commercial license exists for closed-source embedding.

## How to connect
- Official MCP server
- MCP server (official): http://localhost:3900/mcp/
- API docs: https://github.com/debpalash/VoiceStudio/blob/main/docs/speech-platform.md
- API limits: Local: limited only by your hardware.
- Repository: https://github.com/debpalash/VoiceStudio
- Website: https://voicestudio.sh

## Notes
LICENSE-NOTICE: 'Commercial use under the AGPL-3.0 is free'; the paid Commercial License is for closed-source embedding and the Pro build. OmniVoice model card: 'The pre-trained model is licensed under the CC-BY-NC due to constraints from its training data'. Other engines (CosyVoice 3, IndexTTS 2.5, KittenTTS and more) can be chosen in the app, each with its own terms. MCP tools: generate_speech, clone_voice, design_voice, transcribe and more.

## Numbers (AgentGid data)
- Gid Score: 56.0
- GitHub stars: 79,301

## Lines to paste into your prompt or AGENTS.md

```
- VoiceStudio (Voiceover (text-to-speech)). Local desktop app for voice cloning, voice design, video dubbing, dictation and transcription in 646 languages, with a local API and an MCP server for agents. Price: free; commercial use with conditions; credit depends on the item; ship inside your app with conditions.
  Connect: official MCP server http://localhost:3900/mcp/; API docs https://github.com/debpalash/VoiceStudio/blob/main/docs/speech-platform.md. Limits: Local: limited only by your hardware.
  Get files: Local desktop app: API at http://127.0.0.1:3900 (capabilities at GET /.well-known/voicestudio-speech); MCP over HTTP at POST /mcp/ or stdio via python -m backend.mcp_shim.
  Not allowed: App is AGPL-3.0: a modified version offered over a network must give users its source.
  Not allowed: Model weights keep their own terms; the default OmniVoice weights are CC-BY-NC.
  Not allowed: A separate paid commercial license exists for closed-source embedding.
  Rule: The app is AGPL-3.0: commercial and internal use are allowed, but a modified version offered over a network must ship its source.
  Rule: Model weights keep their own terms: the default k2-fsa/OmniVoice pretrained model is CC-BY-NC (non-commercial).
  Rule: Clone voices only with permission.
  License: https://github.com/debpalash/VoiceStudio/blob/main/LICENSE-NOTICE.md (checked 2026-10-17; details: https://agentgid.com/assets/voicestudio/)
```

Not verified: Commercial terms of the other engines' weights; output rights for generated audio with the default model beyond the CC-BY-NC note.

## Links
- AgentGid page: https://agentgid.com/assets/voicestudio/
- Is it free for commercial use? https://agentgid.com/assets/voicestudio/commercial-use/
- Source: https://github.com/debpalash/VoiceStudio
- Source: https://github.com/debpalash/VoiceStudio/blob/main/LICENSE-NOTICE.md
- Source: https://github.com/debpalash/VoiceStudio/blob/main/docs/mcp.md
- Source: https://github.com/debpalash/VoiceStudio/blob/main/docs/speech-platform.md
- Source: https://huggingface.co/k2-fsa/OmniVoice
- Source: https://voicestudio.sh/pro

---
AgentGid's catalog is curated: every entry was checked against the maker's official pages (registry, repository, license and pricing pages) on the date shown; a license value shows the quote it stands on wherever AgentGid has one. Usage numbers come from AgentGid's own scan of public repositories set up for AI coding agents. Claims by makers are reported, not tested by AgentGid.