Voice Portals
There are some 250 million computers connected to the Internet
but there are some 1.3 billion phones which are now capable of
accessing the Internet through any voice portal. The improvements
in speech recognition and speech synthesis and the
adoption of VoiceXML and ECMAScript as standards have made
it practical for the first time to develop telephone speech
applications which provide gateways (voice portals) to the Internet.
In theory public access to the Internet could increase many fold with
it being as easy for those who are computer literate as for those who are not.
A typical architecture for an application in this field might be as shown below.
Anyone with the right phone number (and possibly an access code) can
dial in and start talking to the voice portal. The voice browser
(which sits on the portal) handles the dialogue in accordance with the VoiceXML coding.
If the dialogue is self contained no access to the Internet occurs but,
if some information or dynamic response is required, the voice browser will
use the submit action (based on the http protocol) in much the same way as
a 'visual' browser would. Inside the cgi some code will either produce a
dynamic response (send back a new VoiceXML document for the browser to
interpret) or access a database (on the server or more remotely over the
Internet) and then produce a dynamic response.
Voice portals handle task-oriented dialogues effectively.
These are designed to achieve some quite specific outcome.
It might be a transfer of knowledge, a sale or a payment.
Dialogues of a more conversational nature are not supported but the following
example dialogue (a time-table query) is typical of what lies fairly and
squarely within the domain.
- M: Good Morning, I can help you plan a flight. Where do you want to go?
- P: Sydney.
- M: Where are you flying from?
- P: Darwin
- M: What day do you want to depart on?
- P: Thursday.
- M: Please hold while I check the database.
- M: OK, there is a flight that departs Sydney at 9:50 am and arrives Darwin 6:50 PM. Do you want to hear a later flight?
- P: Yes.
- M: OK, there is a flight that departs Sydney at 12:50 am and arrives Darwin 9:50 PM. Do you want to hear a later flight?
- P: No.
- M: Can I help you with any other enquiry?
- P: No.
- M: I hope we can help you in the future. Goodbye
Voice portals can initiate outward calls just as easily.