
This Kickstarter project allows people to program their own digital assistants using audio recognition and reaction software.
MOVI, which ran on Kickstarter in July of 2015, is now available for purchase. It’s built on Arduino IDE, and parent company Audeme claims that it’s the first cloudless speech recognizer and synthesizer for Arduino.
“We believe there’s a need for ‘Siri-like’ speech recognition with everyday objects, but it shouldn’t be held back with an Internet connection requirement,” said Bertrand Irissou, CEO at Audeme.
In order to allow the MOVI digital assistant to understand commands, up to 150 full sentences or simple commands can be programmed from the Arduino IDE. Since it resides entirely on the Arduino platform, it doesn’t require an internet or PC connection. Audeme’s speech recognizer training methods are still patent-pending, while the software interface is open-source. It takes as few as four lines of code to get the system to recognize and respond to a sentence.
MOVI has also been ported to the Raspberry Pi and the Adafuit Huzzah esp8266, thanks to private members of the open source community. Irissou said that he imagines the speech recognition software being integrated into such diverse functions as home alarm systems, games, home automation, and systems to support the disabled.
The Kickstarter supported the creation of the first engineering prototypes and the initial production, which started in December 2015. With the bugs worked out and a Maker Faire Editor’s Choice Award on the shelf after the Bay Area Maker Faire 2015, they are now in mass production.
MOVI is being sold for $89.90 on Audeme’s website. This could be an interesting project for people who want custom voice assistants or simply to see how what they build can stack up to Siri.