OVOS and Mycroft: A Fork That Wasn't Meant to Be

OpenVoiceOS added a wake word verifier plugin that identifies enrolled household speakers at the wake gate, rejecting television audio and unrecognized voices, as part of a set of updates to the open-source voice assistant ecosystem. OVOS STT, TTS, and translation servers now describe themselves over UTCP and expose an MCP endpoint so any agent speaking either protocol can discover and call them without custom client or wrapper code, and ovos-installer can now deploy OpenVoiceOS on macOS on both Intel and Apple Silicon via launchd and virtualenv. The project also rewrote several C, C++ and Java programs — espeak-ng's G2P, Cotovia, AhoTTS and HermiT — as pure Python with an AI reading the original source and a human orchestrating, keeping the upstream licences while leaving the ownership and derivative-work questions open.

Discover the latest updates, guides, and insights from the open-source voice assistant ecosystem. Join our journey of innovation and collaboration. A new wake word verifier plugin teaches OpenVoiceOS to recognise who is speaking, not just what was said. Enrolled household voices get through the wake gate; the TV advert and the guest who thinks he's funny don't. A full translation platform for OpenVoiceOS skills: context-aware editor, automated PR pipeline, a validation engine tuned to voice-assistant file types, and open ML datasets, running entirely on GitHub, with zero infrastructure of our own to operate. Just a GitHub account, and forkable by anyone. How a 2018 Mycroft fairytale skill, two false starts, and one offhand comment from Jarbas turned into an 8-provider common-reading pipeline for OVOS. We rewrote several C, C++ and Java programs espeak-ng's G2P, Cotovia, AhoTTS, HermiT as pure Python, with an AI reading the original source and a human orchestrating. No human on our side read the originals. That raises two separate questions: can the output be owned at all, and is it a derivative of the input? We kept the upstream licences because that was cheaper than answering. We still think the question is open. Why I built a TUI client for testing OVOS without a mic or speaker - and a few real bugs it caught along the way that no amount of log-tailing would have surfaced on its own. OVOS STT, TTS, and translation servers now describe themselves over UTCP and mount an MCP endpoint. Any agent that speaks either protocol can discover and call them — no custom client, no wrapper code. ovos-installer can now deploy OpenVoiceOS on macOS Intel and Apple Silicon using launchd + virtualenv. Here’s the boring good path to a working OVOS stack on your Mac. Share your knowledge with the OpenVoiceOS community We welcome contributions on topics related to OpenVoiceOS, voice assistants, open-source technology, and more.