While looking for a way to use speech recognition with flash I crossed paths with voice xml. I've gone through the wc3 description of it but still I got a few big doubts.
Can I simply create a vxml doc and place it on my webserver? what are the requirements?
Will it work?
Can I input with a microphone over the internet?
I cant seem to find a direct answer to it.
You need the equivalent of a browser that knows how to process VoiceXML.
I work in telecom industry, so that usually means software that connects to the public phone network through either one of the old style telephony connections or via VoIP. There are many commercial and some open source solutions in this area.
There are some other implementations, like Opera and some research initiatives in the accessibility area, but I haven't seen them gather much ground.
I wouldn't look at VoiceXML as the easiest way to approach speech recognition. That said, there aren't easy ways nor many free/open source solutions. The easiest path on a Microsoft platform would be to look at Microsoft's SAPI layer and the free, minimal ASR they provide. On the Linux side, check out CMU Sphinx.
VoiceXML is a way of marking up content that is intended to be delivered over voice, usually the telephone. Just like HTML is how you format content you want to display over a web browser, VoiceXML is how you deliver it to the phone.
VoiceXML by itself won't do speech recognition. But many of the VoiceXML providers out there also bundle a speech recognition (also known as ASR) engine with their VoiceXML platform.
If you are looking for a way to recognize voice input coming from a microphone on your computer into some sort of desktop application, then VoiceXML probably isn't what you're looking for.
Wikipedia has a list of some speech recognition software at http://en.wikipedia.org/wiki/List_of_speech_recognition_software
If you do want to learn more about VoiceXML, there's some good documentation at http://vxml.org/