Michael Phillips

Sense is building a home energy monitor that lets homeowners know what's happening at home and save energy by providing real-time information on what devices are on and how much energy they are using via hardware and a smartphone app.

Vlingo created the first voice enabled virtual assistants on mobile phones. We started Vlingo in 2006 just as phones were starting to become connected to the internet (so, back when phones were either simple flip phones or blackberrys). We realized that as phones became connected with data networks that there would be a wide range of new uses, and that typing on small keyboards was going to be the primary limitation on what could be done. And, with the data network, we had the opportunity to do distributed processing - so we used large-scale network side speech recognition and natural language processing with the goal of allowing users to say whatever they want into their phone and handling it in a nice way. At Vlingo, we had our own application available on many devices with over 5 million users and then went on to power the voice interfaces on many devices, including the Samsung Galaxy range of phones.

SpeechWorks was acquired by Scansoft in 2003 and I stayed on as the CTO of the combined company for two years. While I was there, we acquired many more speech and language companies, including Nuance in 2005 and changed the name of the company to Nuance.

in 1994, I started SpeechWorks based on technology we licensed from MIT. Speech recognition technology was just getting to the point of being commercially feasible and we found the first market in the telephone call center market. SpeechWorks grew to be one of the main providers of speech technology for telephone based services worldwide and had a successful IPO in August 2000 (SPWX). SpeechWorks was then acquired by Scansoft in 2003.

I was a member of the Spoken Language Systems group at the Lab for Computer Science at MIT. Our group was one of the first to combine speech recognition and natural language processing to allow users to carry on a dialog to perform some specific task. Our first such system was called "Voyager" and allowed users to find restaurants and directions around Cambridge. In many ways this was like what we now know as Virtual Assistants (so, like Vlingo or Siri). However, instead of running on a mobile phone, we were running these systems on computers the size of refrigerators, and each spoken response took 20 minutes for the computer to recognize! Still, it is clear that the work being done at MIT and other research institutions was the basis for the systems we have today.