In the IP IVR 3.X and later release the VXML browser supported VXML with speech only. However with IP IVR 4.0 we support speech as well as DTMF. There is a separate DTMF only voice browser also which can allow you to process VXML without requiring an ASR engine. Also with IP IVR we now support VXML 2.0 With IP IVR you can use the voice browser step for VXML based applications.
Normally, a VoiceXML Server application can only have one locale associated with it, in the application's property page. But I believe you can change the locale on the fly by setting the Cisco-proprietary VoiceXML property com.cisco.tts-server
