Saying that Microsoft is into machine intelligence would be something of an understatement. Skype Translator and Cortana are two prominent examples of Microsoft investment in creating our future AI overlords, and Bing is in on the game with outward facing research and betting tools like Bing Predicts.
Microsoft's Project Oxford is another set of artificial intelligence and machine learning tools that Microsoft has been working on, and the company has taken to its blog to announce the impending release of new development options to putting this intelligence into their own software solutions. Microsoft's Research Cambridge (UK) head Chris Bishop demonstrated the emotion tool during the keynote at Microsoft's Future Decoded conference, running this week.
Project Oxford is aimed at creating software that can identify sounds, words, images, and facial expressions, which can be used for a host of applications, from marketers ascertaining customer reactions to messaging apps that automatically express emotion based on an attached photo. The system learns to recognize these things from intensive training, and then applies what it learns to identifying new traits. And it's not just simple things like recognize a face--the emotion tool can so far tell when that face is expressing anger, contempt, fear, disgust, happiness, sadness, surprise, or is simply neutral.
One currently shipping example of how Project Oxford can identify specific facial attributes is their entertaining MyMoustache site in honor of beard-growing fundraising movement Movember. Other tools that will be available on a limited free trial basis include:
Spell check: intelligent spell check utility that can recognize slang, brand names, common and hard-to-spot errors (e.g., "four" vs. "for") and is continuously updated as new terms are added to the vernacular. In public beta today.
Video: track faces, detect motion, and stabilize video utilizing the same tech as in Microsoft's Hyperlapse app. In public beta by the end of 2015.
Speaker recognition: identifies who's doing the talking by learning a given person's voice patters, useful for security apps. In public beta by the end of 2015.
Custom Recognition Intelligent Services: a.k.a., CRIS, overcomes environmental factors to make speech recognition more efficient and accurate, useful for noisy places and for overcoming speech disabilities and non-native speech patterns. Invite-only beta by the end of 2015.
Updates to face APIs: Project Oxford will also be updated in general, including facial hair and smile prediction, gender identification, and visual age estimation.
Developers can get all the details at Microsoft's Project Oxford site. In the meantime, just rest assured that Microsoft is doing everything it can to usher in the AI apocalypse.