Mycroft AI Launches Mimic 3 Neural Text-to-Speech Platform
Open-source voice assistant developer Mycroft AI has released Mimic 3, its newest neural text-to-speech (TTS) software. Mimic 3 offers more than 100 different voices speaking more than two dozen languages and can operate on a device without needing to interact with the cloud, reinforcing Mycroft’s pitch as a privacy-focused AI developer.
Mimic 3 is essentially a customizable TTS feature that developers can employ within their own hardware or platforms. Users can adjust the language, voice, and details of the speech audio to best suit their needs thanks to Mycroft’s phoneme-based approach to speech synthesis, which mimics human speech patterns better than a standard robotic voice. The low-energy and offline demands make it relatively low-cost to integrate and The software operates on Mycroft’s own platform, as a docker container, or on Linux using an open-source license. Mimic 3 can be integrated into compatible smart devices for a licensing fee and is built into its own smart speakers, including the Mark II, which is set to ship in September.
“Mycroft AI has pre-trained voices for Mimic 3 in 25 different languages, with over 100 individual English speakers available. ,” Mycroft senior software developer Mike Hansen explained in a blog post announcing Mimic 3. “In addition to basic voice controls like speaking rate and variability, Mimic 3 supports a subset of Speech Synthesis Markup Language (SSML), allowing you to script who’s speaking and how. With SSML, you can create a single document that switches voices (and even languages), includes timed pauses between sentences, and manually adjusts volume, speed, etc.”
Mycroft has been very busy in the last few months. Ahead of the Mimic 3 release, founders Joshua Montgomery and Kris Adai stepped down from day-to-day involvement at Mycroft and moved to Hawaii, though Montgomery remains on the board. More crucially for the company’s future, the USPTO Patent Trial and Appeal Board ruled in favor of Mycroft regarding ownership of both patents at the center of a lawsuit battle that began in 2020 with a company Mycroft has characterized as a patent troll. Though the legal fight is not entirely resolved, the win with the USPTO grants Mycroft at least some assurance that its plans for voice tech and smart speakers will not end prematurely.