Now, I do want to say that systems like TellMe Networks, and VoiceXML itself are actually quite cool, if a slight bit non-obvious how to program. TellMe’s systems actually incorporate non-trivial amounts of speech recognition technology, which from what I can tell works on occasion.

But.

Sometimes I want to talk to a travel agent. Sometimes I want to fill in a few web forms. But let me tell you, I’m not optimistic about the future of talking to a web form*. Some analysis to chew on:

1. Speech Vs. Vision: A quick summary of my information transfer theory.
2. Deaf and Dumb: A Critique. An application of the previous document to the standard Telephone Registration Systems at many college universities. Send me feedback.

Incidentally: Would someone *please* create a trivial HTML->VoiceXML translator? I can’t wrap my brain around using a document format for a procedural language just quite yet.