There are several major players that I have watched or worked with:
Voxeo . Earlier versions used Nuance ASR, but I believe that now they use their own. Uses VoiceXML and CCXML to manage calls. They become expensive as they expand, but a development license (2 ports) is free.
Loquendo : I just list it because: a) I did it, and b) they have the best TTS engine I've ever heard. Seriously worth the demo to hear the quality.
Nuance Voice Platform (NVP) . I work with this almost daily. The nuance is by far the most powerful ASR on the market, but it comes at a great price. If you need to ask about the price, then do not worry about falling in love with the quality of the engine. It uses a slightly improved version of VoiceXML.
Microsoft Speech Server You already know that, of course. The blast for the dollar is a great platform. VoiceXML support is borderline (limited control over the engine from VXML properties), but if you want to use Workflows, you can do whatever you want. ASR is good, although it has almost no accuracy with large or complex grammars such as Nuance. This is currently my preferred platform for most applications. And, like Voxeo, the developer license is free.
source share