Skip to main content
PromptQuorumBuilt for humans. Structured for AI.
Home/Power Local LLM
Voice, Speech & Multimodal

VoiceStudio Review: Free Local Voice Cloning and Dubbing Studio

·8 min readBy · Founder of PromptQuorum · Discovery engine for open-weight & open-source AI

VoiceStudio (stable version 0.5.8) is an AGPL-3.0 desktop studio that clones and designs voices, dubs video, transcribes and dictates, and its maker says the free app runs on your own machine without an account. It also provides a local API, a CLI and MCP support, and every statement here comes from the maker's website, privacy and terms pages and GitHub repository.

Key Takeaways

  • Maintenance: the main branch last changed on 18 November 2026 and the newest tag is a pre-release from the same day, so no stale-project warning applies.
  • Version: the latest stable release is 0.5.8 from 12 November 2026; faster-moving 0.5.9 beta builds are published as pre-releases.
  • Cost: the app is free under AGPL-3.0; a separate closed-source Pro build with a commercial license is paid, and the pages we read show no price.
  • Platforms: stable on macOS (Apple silicon), Windows and Linux plus Docker images; Android and iOS are labelled preview.
  • Maker: the repository maintainers, with an individual developer named as creator in the site data; the legal terms fall under the law of India and no company imprint was found.

📍 In One Sentence

VoiceStudio is an open-source desktop app that clones and designs voices, dubs video, transcribes, dictates and narrates audiobooks on your own computer, with a local API, a CLI and MCP support.

💬 In Plain Terms

Think of it as a local alternative to a cloud voice service: you install an app, download models and do the voice work on your own machine instead of a subscription site.

📌Note: This review is based on voicestudio.sh, its privacy and terms pages, the GitHub repository and its release page, checked on 19 November 2026. PromptQuorum has not installed or run the app, so voice quality, dubbing quality and speed are not assessed.

VoiceStudio (voicestudio.sh, also called "OmniVoice Studio" in the website's structured data) is an open-source desktop app that clones and designs voices, dubs video, transcribes speech, dictates text and narrates books, with models running on your own computer. The core app is free under the AGPL-3.0 license, a separate paid Pro build adds a commercial license, and it also offers a local API, a command line and MCP support. This review is based on the website, its privacy and terms pages, the GitHub repository and its releases, checked on 19 November 2026; PromptQuorum has not tested the app hands-on.

VoiceStudio Review: Free Local Voice Cloning and Dubbing Studio

What Is VoiceStudio?

VoiceStudio is a desktop studio for voice work that runs on your own computer: cloning a voice from a short sample, designing one from a description, dubbing video, transcribing, dictating and turning scripts or EPUB files into audiobooks. The website calls it a free, open-source alternative to ElevenLabs with no account required (the free core app; a paid Pro build also exists), and its structured data also lists the name "OmniVoice Studio".

The README names k2-fsa/OmniVoice as the default engine and says the Model Manager offers dozens of other engines, with whisper.cpp for speech recognition. The site and README claim 646 languages but give no per-engine breakdown, so treat that figure as a maker claim.

Get It

VoiceStudio installs like a normal app on macOS, Windows and Linux, so no terminal is needed, and the README also lists Homebrew, one-line installers, Snap, AUR and Docker.

Option
How to get it
macOS.dmg for Apple silicon (an Intel build exists with limits), or Homebrew cask: GitHub Releases
WindowsSetup .exe for x64 (and ARM, experimental), or a PowerShell one-liner: GitHub Releases
Linux.deb or AppImage for x64, also Snap and AUR: GitHub Releases
DockerImage ghcr.io/debpalash/voicestudio:stable on port 3900: README
Android and iOSAPK and unsigned IPA, labelled preview: GitHub Releases
Websitevoicestudio.sh
Privacy policyvoicestudio.sh/privacy
Source codegithub.com/debpalash/VoiceStudio, AGPL-3.0 license

Version reviewed: 0.5.8, the latest stable release published on 12 November 2026 per the GitHub release page. The newest tag is the pre-release 0.5.9-beta.6 from 18 November 2026, and the newest commit also dates from 18 November 2026. GitHub is only the delivery channel for these installers, and models are downloaded separately.

Requirements

The website says about 8 GB of RAM is enough to start and a GPU is optional but faster; the README states no minimum VRAM.

Operating system

What the sources say:
Stable on macOS 13.3+ with Apple silicon (Intel Macs only with a remote backend), Windows 10/11 x64 and Linux x64; Windows on ARM is experimental

Acceleration

What the sources say:
Metal on Mac, NVIDIA CUDA or opt-in AMD ROCm on Windows, CUDA, CPU or Vulkan on Linux; runs on the CPU, just slower

Memory and VRAM

What the sources say:
About 8 GB RAM to start per the website; the README gives no VRAM figure and points to a separate performance guide

Models

What the sources say:
Downloaded inside the app through the Model Manager; each model shows its own license

Mobile

What the sources say:
Android 10+ and iOS 17+ are labelled preview; iOS needs sideloading

Internet

What the sources say:
Needed for model downloads and update checks; the website says the free app connects online only when asked

PromptQuorum did not check which models run on which backend or how much memory a given model needs.

Features Confirmed by the Sources

Every item below comes from the website or the README; none has been independently tested.

  • Clone and design. Create a voice from about three seconds of clean audio, or design one from gender, age, accent, pitch and emotion without a recording; saved voices can be reused.
  • Dub. Transcribe, translate and re-voice video while keeping each speaker separate and aligned with the original timing, with multi-speaker dubbing and dub export guides.
  • Narrate. Turn a long script or an EPUB into a chaptered audiobook read by a cloned, designed or preset voice, and assign voices to roles in stories.
  • Dictate and transcribe. A floating dictation widget with a local recognition engine and searchable history, plus editable transcripts from audio or video.
  • Automate. A documented local API (audio, voices, dubbing, jobs, MCP and more), a CLI, reusable recipes, watch folders and queued batch jobs.
  • Languages. The website itself is offered in twelve languages and claims speech in 646 languages.

Cost and License

The desktop app is free and open source under AGPL-3.0, and a separate closed-source VoiceStudio Pro build adds a paid commercial license; no price appears on the pages we read.

Free app

What the sources say:
AGPL-3.0, no account needed; model weights, voices and outputs carry their own licenses shown per model

VoiceStudio Pro

What the sources say:
Paid commercial license and extra features for teams, 7-day free trial, 7-day refund, no automatic renewal; price not shown

Cloud

What the sources say:
Listed as early access with no pricing

Commercial use

What the sources say:
The website offers Pro for teams that need commercial terms; its exact scope is not itemized on the pages we read

AGPL-3.0 obligations depend on how you use or distribute the software and are not summarized here; ask a lawyer before commercial use.

Data Flow and Privacy

The website says the free app needs no account or cloud and connects online only when asked, such as to check for updates or download a model, and the README describes a fully local tool. The privacy page says the free desktop app processes data locally and contacts the network only for update or model requests or when you use a network-backed adapter.

The same page says the Pro desktop app needs an account, sends licence checks that record your IP address and country, and sends usage and error telemetry to a third-party analytics service that it says cannot be switched off. It also states that the public website uses cookieless analytics and Cloudflare hosting.

📌Note: These are the maker's declarations, not audit results. PromptQuorum has not read the code or inspected the app's network traffic.

Voice Cloning, Consent and Model Licenses

Cloning needs only a short sample, and the maker's own rules say it requires the voice owner's permission and prohibit impersonation and fraud, so consent is your responsibility.

  • Consent and law. Whether cloning a voice is lawful depends on the person, the purpose and the country; this review is not legal advice, and PromptQuorum has not assessed any jurisdiction.
  • Model licenses differ. The README says model weights, voices and outputs carry their own licenses, shown per model in the Model Manager; the AGPL-3.0 license covers the app, not those weights.
  • Governing law. The terms state that they are governed by the law of India, with the courts at Agartala, Tripura; no company name or address is given.

⚠️Warning: Do not clone a voice without the owner's consent. Cloning a person's voice without permission can infringe personality and data-protection rights and, depending on the use, can trigger labelling duties for synthetic audio. Obtain consent first; this is not legal advice.

What We Could Not Verify

  • Output quality and speed. PromptQuorum did not listen to cloned voices, check dubbing accuracy or time generation on any model.
  • Language coverage. The 646-language figure is a maker claim without a per-engine list, and we tested no language.
  • Free versus Pro. The pages say Pro adds extra features and a commercial license but we found no feature-by-feature list or price.
  • Release stability. Pre-releases are published several times a week, so the stable 0.5.8 build may differ from the newest beta.
  • Installer signing. Whether the macOS and Windows installers are notarized or code-signed was not checked.

Competitors and Alternatives

Voicebox

Platforms:
macOS, Windows (Linux from source)
Price / license:
Free / MIT
Key difference:
Voice cloning, text-to-speech and Whisper dictation with an MCP server; its review lists no video dubbing
Articles about Voicebox (1)

Also mentioned in:

YoVoice

Platforms:
macOS, Windows, Linux
Price / license:
Free / Apache-2.0
Key difference:
Speech, voice cloning and AI music with a CLI and MCP server; no paid Pro build found
Articles about YoVoice (1)

Also mentioned in:

XTTS v2

Platforms:
macOS, Windows, Linux
Price / license:
Free / CPML (non-commercial)
Key difference:
A single cloning model used as a library, with no studio, dubbing or dictation features
Articles about XTTS v2 (10)

+6 more not shown

These are same-category apps from the PromptQuorum directory; confirm each app's current price, license and platforms on its own page.

Compare in directory

Frequently Asked Questions

Can I run it from a script or an AI agent?

Yes: the README lists a local API, a CLI and MCP support, and the documentation has an MCP section and an OpenAI-compatible audio endpoint. We did not test any of them.

Which engine does it use by default?

The README names k2-fsa/OmniVoice as the default engine and says the Model Manager offers dozens of others, with whisper.cpp for speech recognition.

Does it support my language?

The maker claims 646 languages without a per-engine breakdown, and the website is offered in twelve languages including German, French, Spanish, Chinese and Arabic. We tested none of them.

Does it run on a computer without a GPU?

The README says it runs on the CPU, just slower, and the website says about 8 GB of RAM is enough to start; we measured no speed.

Is the Android or iPhone app ready to use?

No: the README labels the Android (10+) and iOS (17+) builds preview, and the iOS file is an unsigned IPA that needs sideloading.

Verdict

VoiceStudio is a fast-moving, AGPL-3.0 way to clone voices, dub video, transcribe and dictate on your own computer, with a local API and MCP support, and a separate paid Pro build for commercial terms. Its local-first statements come from the website and README rather than an independent audit, stable macOS support covers Apple silicon only, mobile apps are previews, model weights have their own licenses and the 646-language, quality and speed claims are untested here. It suits creators and developers who want one free local voice tool; anyone who needs proven output quality, clear commercial terms or a vetted data policy should test it and read the Pro terms first.

Sources

← Back to Power Local LLM

MCP