IBM Puts More Machine Learning Services in Watson Cloud

IBM has added five new services in beta to Watson Developer Cloud, which offers developers a way to build applications that incorporate Watson's cognitive computing capabilities. IBM made some of those capabilities available to the masses in 2013 and now claims that over 6,000 applications tapping them have been made to date.

The new services are speech to text, text to speech, visual recognition, concept insights and trade-off analytics. Watson Developer Cloud now has 13 services available in beta. These services aim to enable a new class of applications that employ machine learning, understand natural language, and identify hidden patterns.

About the new services, available today:

Speech to Text: IBM claims this is one of the first real-time services with accurate low-latency speech recognition capabilities. Watson is adept at understanding natural language. The service applies machine intelligence related to grammar and language structures within a specific context for more accurate transcription. In other words, it’s good with homonyms. If you’re converting Jimmi Hendrix lyrics from speech to text, the likelihood of the transcription reading “'scuse me while I kiss the sky” instead of “excuse me while I kiss this guy” is greatly increased.

Text to Speech: This service supports translation in both English and Spanish with three optional voices across the two languages. One of those options is the voice Watson used in its famous 2011 Jeopardy match. You can get Watson to say “I'm sorry, Dave. I'm afraid I can't do that,” in your modern telling of '2001: A Space Odyssey.'

Visual Recognition: Visual recognition analyzes various forms of media and helps to collect and organize large sets of visual data to build semantic associations. Again using machine learning technology, it recognizes an array of visual entities like settings, objects, and events to understand content of images and videos. It can even tell the difference between Bill Paxton and Bill Pullman.

Concept Insights: The service looks at text in a conceptual manner. It has a search capability that discovers new insights on text compared to traditional keyword searches. Concept Insights links user provided documents with a pre-existing graph of concepts based on Wikipedia. These links are explicit (directly mentions a concept) and implicit, which links the user’s documents to relevant concepts not directly mentioned. Kids, when I was young, I had to do this kind of thinking manually.

Trade-off Analytics: This service helps you make better choices by dynamically weighing multiple and often conflicting goals. It uses Pareto filtering techniques to identify optimal alternatives across multiple criteria. It uses various analytical and visual approaches. Here, IBM might have finally built something that can help married couples figure out what to eat for dinner when one person says “they’re up for anything” but rejects everything the other suggests.

Cognitive computing has implications in many fields, a point highlighted by some of the promotional work IBM did prior to making Watson widely available to developers. Some notable visible mainstream uses include the appearance on Jeopardy, staring in late-night monologues, and coming up with unique recipes at SXSW in Austin.

Watson is versatile when it comes to wider industry applications. In healthcare, for example, Watson can make some recommendations a doctor makes, loaded to a smartphone.

IBM already has numerous cloud services that leverage Watson and provides APIs developers can use to tap into the capabilities in their own applications through its Bluemix Platform-as-a-Service. Watson is one of the most active communities within the IBM Bluemix PaaS ecosystem.

In early 2014, the company committed $1 billion to investments in the various businesses around Watson and opened a headquarters building in New York City dedicated exclusively to Watson.