Partnerships are “recombinance personified.” So it is with a new service from Microsoft and Yap that enables Blackberry users on the Sprint network to dictate both SMS text messages and email. There is a lot to parse and ponder in this announcement because it signals that there are destined to be multiple players among mobile speech technology providers, and that some of the largest firms in the business (like Microsoft) are not about to settle on a single solution, or set of solutions.

Yap was launched in 2006 by long-time IBM software maven Igor Jablokov and his brother Vic. Igor had been a long time proponent of multimodal and mobile applications coming out of IBM’s labs and its software group. At the start, the Yap platform was pretty much Big Blue through and through, although Igor always maintained that delivering a high quality speech recognition product was based on constant use, testing and refinement, as opposed to the use of any particular recognition engine. Well the Sprint deal shows the sort of opportunities that such agnosticism can promote.

The downloadable app is available on the “business on main” Web site embedded in MSN. Microsoft’s selection of Yap is a bit curious, given that the company obtained a very robust set of mobile software for Blackberries (along with the hosted voice portal software) when it acquired Tellme. What’s more, under “speech Czar” Zig Sarafin, the company has been highlighting its own initiatives around the Tellme, Bing Mobile and Microsoft Speech brands, as signaled in this landing page.

As we noted in this brief posting about Safafin’s video debut, Microsoft has a broad set of initiatives around speech. Yet it becomes increasingly evident that each family of speech-based problems (such as rendering spoken words as text messages or emails) requires a specific set of solutions. In this case, in spite of all its R&D investment and technology acquisitions the company went to a cloud-based service to back end its home-grown app.

Categories: Articles

Tags: Microsoft, mobile speech, Recombinant Communications, speech-to-text