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

Open Voice OS Review: Open-Source Voice Assistant for Linux and Raspberry Pi

·6 min readBy Hans Kuepper · Founder of PromptQuorum · Discovery engine for open-weight & open-source AI

Open Voice OS (installer v11.0.0, released on 29 August 2026) is an open-source voice assistant platform for Linux, Raspberry Pi and Docker, and it is not fully offline by default. The speech recogniser it starts with needs internet, local speech is optional, and an LLM fallback is an add-on. Every statement here comes from the project documentation, not from testing.

Key Takeaways

  • Maintenance: the installer repository was last pushed to on 1 November 2026, so no stale-project warning applies.
  • License: Apache-2.0 open source and free; donations are optional.
  • Origin: it started as MycroftOS and keeps an enhanced fork of Mycroft-core.
  • Setup: one terminal command, a browser-guided installer or a Raspberry Pi image, not a double-click app.
  • Speech recognition: the default engine needs internet; local engines are optional.

📍 In One Sentence

Open Voice OS is an open-source voice assistant platform that runs on your own Linux machine, Raspberry Pi or container and can send unmatched requests to an LLM.

💬 In Plain Terms

A wake word starts it, a speech engine turns your words into text, a skill answers or an optional LLM steps in, and a voice reads the reply back.

📌Note: This review is based on the Open Voice OS website and the GitHub documentation of ovos-core, ovos-installer and ovos-docker, checked on 1 November 2026. PromptQuorum has not installed or timed it, so recognition quality, latency and LLM reliability are not assessed.

Read in:
🇺🇸en🇩🇪de🇫🇷fr🇯🇵ja🇨🇳zh🇪🇸es🇧🇷pt🇸🇦ar🇰🇷ko

Open Voice OS (OVOS) is a community-built, open-source voice assistant platform from OpenVoiceOS that began as a fork of Mycroft. It runs on Linux machines, Raspberry Pi boards and Docker, takes skills as plugins, and can hand requests it cannot match to an LLM. This review is based on the project website and its GitHub documentation, not on hands-on testing.

Open Voice OS Review: Open-Source Voice Assistant for Linux and Raspberry Pi

What Is Open Voice OS?

Open Voice OS is a community-run voice assistant platform for smart speakers and other voice-first devices. The website says it began as MycroftOS, was renamed to avoid trademark issues and now maintains an enhanced fork of Mycroft-core. Its core component, ovos-core, is released under Apache-2.0.

Beyond the core there are an installer, a Raspberry Pi image and a Docker setup. The core repository lists the latest GitHub release as 2.1.1 (5 November 2025), while the installer is at v11.0.0 and the default branch of the core keeps receiving commits.

Get It

Open Voice OS has no single download: you choose a browser-guided installer, a terminal script, a Raspberry Pi image or Docker.

Option
How to get it
Browser installerstart.openvoiceos.org: guided setup, called the easiest route
Installer scriptovos-installer: one command on Linux, WSL2, macOS 15+
Raspberry Pi imageRaspOVOS; development images are marked "expect bugs"
Dockerovos-docker, Apache-2.0
Websiteopenvoiceos.org
Data sharingTelemetry notes; no privacy policy page was found
Source codeovos-core, Apache-2.0

Versions reviewed: ovos-installer v11.0.0 (29 August 2026) and ovos-core 2.1.1 (5 November 2025), per GitHub. The installer needs a terminal: there is no double-click app. The project is free; the website only offers optional donations.

Requirements

The sources give a system list and one hardware figure, so sizing beyond that is guesswork.

Operating system

What the sources say:
Debian/Ubuntu and derivatives, Fedora, Arch, openSUSE, WSL2, macOS 15+ on Apple Silicon (extra setup)

Tools

What the sources say:
curl, git, sudo and Bash 4 or later for the installer script

Local speech

What the sources say:
A Raspberry Pi 5 with 8 GB RAM or something at least as capable

Boards and devices

What the sources say:
Raspberry Pi 3, 4 and 5 on the downloads page; Mark I and Mark II devices

Speech engines

What the sources say:
Default recognition needs internet; Vosk and DeepSpeech (offline) and Mimic3 (local voice) are named

Configuration

What the sources say:
One plain file: mycroft.conf

No supported-language list appeared on the pages checked, so confirm your language before you invest time.

How to Get Started

These steps follow the documentation; PromptQuorum has not run them.

  1. 1
    Pick an install path
    Why it matters: Use the browser installer, the one-line installer script, the Raspberry Pi image or the Docker setup.
  2. 2
    Answer the installer screens
    Why it matters: Choose virtualenv or container, a release channel, a profile and features such as Home Assistant or an LLM.
  3. 3
    Decide on the extras
    Why it matters: Optionally connect Home Assistant, an OpenAI-compatible LLM endpoint or HiveMind satellites for more devices.
  4. 4
    Review data sharing
    Why it matters: Accept or decline installer telemetry and usage metrics, then edit mycroft.conf for anything else.

Features Confirmed by the Sources

Everything below comes from the Open Voice OS documentation; none of it has been independently tested.

  • Skills as plugins. Some skills ship with ovos-core and most are installed separately from PyPI or the OpenVoiceOS GitHub organization; most classic Mycroft skills also work.
  • LLM fallback. Through ovos-persona and solver plugins, requests no skill handles can go to any OpenAI-compatible endpoint, including a self-hosted one.
  • Home Assistant. The installer can set up an optional Home Assistant integration.
  • HiveMind satellites. Several devices can share one assistant.
  • Terminal client. ovos-tui-client lets you type to the assistant without a microphone.

Data Flow and Privacy

The installer states that Open Voice OS is not a fully offline mode: if recognition fails or returns nothing, the recording goes to public servers, and a part that cannot run locally falls back to public speech. Local recognition and voice are described for the alpha channel. The website also says only that the platform has a "focus on privacy and security".

Two kinds of data sharing are asked about separately and are off unless accepted. Installer telemetry is sent once during installation and, if accepted, uses a third-party IP lookup over plain HTTP to fill in the country. Usage metrics report the intents the assistant matches to a community server and are published on an open-data portal; removing the open_data block from mycroft.conf stops them.

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

Who Should Use It

  • Raspberry Pi and Linux tinkerers who want an open smart-speaker stack and accept a terminal install.
  • Mycroft users looking for an actively maintained descendant that still runs many classic skills.
  • Home Assistant owners who want a separate, extensible voice assistant with an optional LLM behind it.

What We Could Not Verify

  • Recognition quality and speed. No timings are given in the sources and PromptQuorum measured nothing.
  • Language coverage. No supported-language list appeared on the pages checked.
  • Release cadence. The core's newest GitHub release is from November 2025 while commits continue; the documentation points to prerelease quirks.
  • Community server handling. Retention and use of data sent to the telemetry and metrics servers is not documented, and no privacy policy page was found.

Competitors and Alternatives

Home Assistant Assist

Platforms:
Linux, web, iOS, Android
Price / license:
Free / Apache-2.0
Key difference:
Voice pipeline inside the Home Assistant smart-home server
Articles about Home Assistant Assist (1)

Voxa

Platforms:
Windows, macOS, Linux
Price / license:
Free / MIT
Key difference:
Local-first desktop voice assistant with optional cloud voice
Articles about Voxa (4)

Also mentioned in:

These are the nearest voice assistants in the directory rather than direct substitutes; confirm each app's current price, license and platforms on its own page.

Compare in directory

Frequently Asked Questions

Is Open Voice OS the same as Mycroft?

No, but it is descended from it: the project says it began as MycroftOS and maintains an enhanced fork of Mycroft-core, and the core README says most classic Mycroft skills also work.

Can it use a local LLM?

Yes, as an optional fallback: the installer accepts any OpenAI-compatible endpoint, including a self-hosted one, for requests that skills do not handle.

Do I need a Raspberry Pi?

No. Linux distributions, WSL2, macOS 15+ on Apple Silicon and Docker are also listed; only local speech is tied to a Raspberry Pi 5 with 8 GB RAM or better.

Can I use it without a microphone?

The installer sets up ovos-tui-client, a terminal client for typing to the assistant.

How do I stop usage metrics?

Remove the open_data block from mycroft.conf and restart the services; the documentation does not explain how to withdraw installer telemetry once accepted.

Verdict

Open Voice OS is an actively developed, Apache-2.0 route to an open voice assistant on hardware you control, with optional local speech, Home Assistant and LLM add-ons. Against that, the default recogniser needs internet, installation happens in a terminal, no language list appeared on the pages checked, and speech quality and latency are untested here. It suits tinkerers and Mycroft veterans; anyone who wants a double-click app should look at the alternatives above.

Sources

← Back to Power Local LLM

MCP