Ubuntu and Mycroft better together: How you can help


Mycroft is the Open-Source Intelligent Virtual Assistant. We aim to share tools and data while at the same time becoming an agent for user privacy. The core open source Mycroft technologies are developed jointly by Mycroft AI and the community members. Skills and Enclosures can be created by individuals, community groups or companies under whatever license they deem fit. One distinct advantage we have over other solutions is our community of developers. Through their work, there are many interesting and unique skills that showcase the benefits of open source software.


Mycroft is made of several key components. The Mycroft team is primarily responsible for improving and maintaining Mycroft-Core which provides the backbone of interaction via voice.

The 3 main components of Mycroft-Core consist of a Speech to Text Engine, an Intent Parser, and a Text to Speech Engine. While these elements also are improved by the community, the main area of community development is in adding Skills. Skills simply represent all the capabilities Mycroft and offer the greatest opportunity to add new use-cases to Mycroft. A few examples of these include interacting with Gmail, searching Youtube, or commanding Mycroft to create a new poem, all via voice interaction.


All of this work would be much more difficult without a platform like Ubuntu. Ubuntu is a key component in nearly every aspect of the company. Ubuntu is more than just the operating system in which we run Mycroft; we develop using Ubuntu, we run Mycroft wherever Ubuntu can run, we use Ubuntu to run our servers responsible for the backend, and most importantly we use Ubuntu’s pre-existing community members with a passion for Natural Language to develop skills.


Our current plans to improve on Mycroft involve a few key areas.


The first way we would like to improve Mycroft is by working on an Open Source Speech to Text Engine. Some work has been done in this area, but most STT systems are controlled by large companies like Amazon, Google, Microsoft, and Apple. By encouraging an open source alternative, we can allow people and companies to not be forced to give their data to another company.


Another area we are working on is getting Mycroft in more systems. There is a lot of room in the voice control space for a system that works well. One example of this is in-vehicle systems. Auto manufacturers are far behind tech companies when it comes to interfaces. One key improvement would be to add intelligent voice controls to vehicles.


We also plan to open the door so as many devices as possible can use Mycroft enabled voice services. With a little processing power and an internet connection, any device can be improved to add voice control. All it takes is a Raspberry Pi or equivalent to process the speech services. With that type of added capability, we want to enable any device that can benefit from voice to be able to have the capability.


The final area in which we want to improve Mycroft is naturalizing the interaction. Current voice interfaces still require trigger words and are limited by their pre-programmed skill-set. We want to develop a conversational interface that can respond in intelligent ways and adapt to you. By integrating machine learning techniques and better semantic processing, Mycroft could become the digital assistant we have always dreamed of having.

Together with the open source community, Mycroft wants to develop an AI for everyone. By basing our scope on Ubuntu, Mycroft can easily be adapted to work on many different devices.

Ballroom A
Thursday, March 2, 2017 - 13:30 to 14:30