System Speech Recognition, MS Speech Platform and Kinect

Question

I recently started working on a project that requires speech recognition (SR) in the Portuguese language. Initially, the plan was to use Kinect´ss SR but later we found that that Kinect´s Language Pack (LP) didn´t include Portuguese (PT).

So i turned to MS Speech Platform that supports quite a few more languages, including PT but for reasons that i haven´t quite figure out i can´t get the MS samples working and there´s not much information or tutorials about it on the Web.

I´m quite confused regarding about language packs and their role within SAPI´s: from what i´ve read from another thread, Microsoft.Speech.Recognition uses the Server version of SAPI (which has its own language packs, Microsoft Speech Platform - Server Runtime Languages), while System.Speech.Recognition uses the Desktop version of SAPI, which has what LP´s? I know my Windows has en-US since i´ve tried out a SR app and it worked just fine by using the where to get extra LPs? Is it even possible to use SAPI´s LP with a Kinect? If so, what are its limitations?

Regards, Nuno

Answer 1

You can't use the Desktop SR engine (System.Speech.Recognition) with Kinect. The desktop SR engines come with the Windows language packs (for Windows Vista, 7, 8, and 8.1).

Kinect uses the Server SR engine (Microsoft.Speech.Recognition); you should be able to use the Portuguese engine, but I haven't tried it myself.

Answer 2

SAPI and Microsoft Speech Platform are different things. SAPI is an interface, which can be used by voice engine developers. MS Speech Platform is an independent speech engine, non-compatible with SAPI. LP for SAPI is not compatible with MS Speech Platfrom and LP for MS Speech Platform is not compatible with SAPI. But Platform has it's own SDK. You can use Kinect SDK for non-language purposes and MS Speech SDK for speech recognition in one project, without SAPI. MS Speech Platform also has a good text-to-speech.