IBM’s Watson gets an upgrade to understand speech and recognise objects

6 Feb 2015

IBM's Watson supercomputer as he would have appeared on Jeopardy. Image via Atomic Taco/Flickr

IBM’s Watson supercomputer has five new strings to its powerful digital bow after it received its latest upgrade, which includes text-to-speech and visual recognition.

Watson’s five new abilities are currently in beta testing, but include speech-to-text, text-to-speech, visual recognition, concept insights and trade-off analytics, the latter two of which would give the user a much better picture of different factors that go into making a particular choice.

The Watson supercomputer is currently used by more than 5,000 different clients, ranging from individuals to insurance companies, looking to better advise their customers on their financial futures.

Now, these groups will be able to test Watson’s new capabilities, with much being made of its text-to-speech function which first made a public appearance as far back as 2011 when it appeared on the US game show Jeopardy.

Now however, the user can input any English or Spanish text to generate speech output, a service IBM says has potential applications for the visually impaired, as reading-based education tools and for multiple mobile apps.

Meanwhile, Watson’s visual-recognition capability, according to InfoWorld’s test, saw an image of a cat presented to Watson come back with a probability of it being a cat at 67pc. While this percentage is low, the hope is a growing user base will help fine tune the capability.

That is why last September IBM began offering Watson’s services as part of a ‘freemium’ to allow more companies to try out and potentially invest in future Watson developments.

Colm Gorey was a senior journalist with Silicon Republic