Mycroft: A Startup is Focusing on Open Source AI for the Home

by Ostatic Staff - Jun. 21, 2016

Ask many people to name the technology categories that are creating sweeping change right now, and cloud computing and Big Data analytics will probably be top of mind for a lot of them. However, there is an absolute renaissance happening right now in the field of artifical intelligence and the closely related field of machine learning. And, open source tools are making a difference in this space.

Now, a startup company is focusing on voice-controlled artificial intelligence applications for the home. Mycroft is billed as the world’s first open-source, open-hardware home AI platform.

Mycroft Core, the software-driven brains behind the effort has just arrived in a new version. According to its developers, it is focused on "responsive listening":

"Today marks the release of a new version of Mycroft....This is a significant update because it contains a large rethinking of the listener that Mycroft uses. The result is our new “Responsive Listening” mode.

When we began experimenting with the Mycroft Core listener, the part of the code that listens for a command, we embarked on creating a concept of passive listening. Passive listening processed an audio buffer, looked for the wake word, and then processed the query. We found in practice that this allowed us to say things in a very unstructured way and they could still be processed, such as:

“What’s the weather Mycroft?”

“Hey there Mr. Mycroft, I’d like to know what time it is.”

This was an interesting paradigm in interacting with the device. But in practice passive listening resulted in long processing time, false positives, and parts of sentences getting processed that weren’t the intended query. The best intentions never go unpunished. As a result, we took it upon ourselves this past week to provide a more consistent experience, stable experience...."

 The new experience that Mycroft offers is represented in a Kickstarter video. Currently, Mycroft responds to basic voice commands. It will play a song you request from YouTube, or report weather conditions for you.

This is just one of the interesting open source AI technologies arriving. Google has open sourced a program called TensorFlow. It’s based on the same internal toolset that Google has spent years developing to support its AI software and other predictive and analytics programs. You can find out more about TensorFlow at its site, and you might be surprised to learn that it is the engine behind several Google tools you may already use, including Google Photos and the speech recognition found in the Google app.

Meanwhile, Facebook is open sourcing its machine learning system designed for artificial intelligence (AI) computing at a large scale.