For manufacturers and developers who until now have stayed away from voice assistants due to privacy concerns, Mimic 3 delivers on Mycroft’s privacy-first principle that user data isn’t stored on a cloud and will never be sold. The company’s stated mission is to provide full privacy and security for those who want to be the sole owners of their personal and/or business data.
The new Mimic 3 TTS engine was built with a global mix of users, creators, and translators in mind, and includes more than two dozen languages and 100+ Voice Sets. New ones are being developed by the community all the time, and anyone who wants to add a language can join the vibrant group of developers contributing to the open source, privacy-focused project. As the company highlights, "with Mimic 3, it only takes one user to implement an entirely new language!
The new text-to-speech engine runs completely offline and is able to deliver natural voice quality, pacing and fidelity without an Internet connection. It runs on Mycroft.AI, Linux or as a Docker container and includes its own web server. Users can switch voices, control rate and variability of speech. Mimic 3 converts all input text to phonemes (distinct units of sound in a language that distinguish one word from another) and recognizes and matches the distinct nuances of how an individual speaks.
Beyond Mimic 3, Mycroft’s team is hard at work preparing to deliver the Mark II, its state-of-the-art, privacy-respecting smart speaker that was firs unveiled at CES 2019, and is now expected to ship in September 2022. The company has strong relationships with the open source community and has received financial investments from Jaguar Land Rover, 500 Startups, Techstars and Kickstarter backers - the Mark II was successfully funded in 2018. Mycroft’s corporate initiatives have broad applicability and are being used by both hobbyists and commercial entities to drive voice data innovation.
www.mycroft.ai
