AT&T has announced it will offer developers access to its Watson speech recognition technology through new Application Programming Interfaces (APIs).

Watson can detect spoken words, regardless of accent or dialect and turn them into text. The API would allow developers to include speech recognition in their apps, without having to build the functionality themselves.

According to a video announcing the new APIs, speech recognition works best when it’s focused on specific categories. As a result, AT&T engineers will create different categories of Watson APIs, which will enable the software to know what types of words to expect.

Those categorized APIs will soon be deployed on AT&T’s U-verse system, allowing users to do voice searches for television shows and movies.

Watson could also play a part in AT&T’s goal of providing real-time speech translation on mobile devices. AT&T Research is currently developing a real-time speech-to-speech translation technology that begins translating as soon as it detects speech, without the latency incurred while waiting for an utterance to complete before translating.

At AT&T’s developer conference in January, which precedes the Consumer Electronics Show (CES), Chief Marketing Officer David Christopher introduced the 2,500 developers in attendance to the company’s new developer platform and API catalog.

Christopher stressed that new APIs released this year would offer developers a set of tools to integrate apps with AT&T’s U-verse set-top boxes.

AT&T expects 10 billion API transactions this year, an exponential increase from the 300 million it saw in 2009.