Here’s some stuff you should check out:

• Dictation Resource Kit (DRK):  This is fantastic if you have a speech application that requires a specialized vocabulary.  For example, say you’re building a product for the medical or legal professions, where there’s a lot of specialized terminology.  You can use this tool to build a language model for your domain that can be used with the speech recognition engine in Windows.  Maybe in a few months I’ll put together a sample and blog about it.
• I’ve been playing a lot with the Bing client on my iPhone, which uses our speech engine in the cloud so you can tell it what you want to search for.  Very *very* cool.  Try it out.  It’s also available on Windows Mobile – actually it was there first, and it’s also excellent.
• TellMe.  This is an astoundingly impressive operation, delivering state-of-the-art IVR phone services (i.e. automated phone applications that work extremely well).  For example, there’s a good chance that when you dial 411 (directory assistance) in the US, it’s a TellMe app that’s answering the phone. 
• The speech macro recorder for Windows is a lot of fun.  Basically, you just fill it with your own custom voice commands so that you can customize the way you control programs with your voice.
• If you’re using speech recognition on Windows and you get a new computer, you don’t need to retrain the new computer to learn your voice.  Just use the profile tool to back-up your profile from your old computer and put it on your new one.