Voice SIG: Natural Prompt Design
General features of naturally occurring conversations that should be incorporated into the design of prompts.
The Voice SIG is currently being considered for inclusion as a special interest group of the Software Development Forum, join us FREE of admission for this upcoming meeting and show your support for this new group.
Cubberly Community Center
4000 Middlefield Rd. Room H-1
Palo Alto, CA
6:00-6:30pm Registration / Networking / Refreshments (Free Pizza!)
6:30-8:00pm Presentation / Q&A
The telephone is an icon of conversation in our culture. Callers naturally bring conversational expectations to all phone calls, and speech recognition applications are no exception. This talk will focus on general features of naturally occurring conversations that can and should be incorporated into the design of prompts. There are a number of benefits to this approach. When engineered dialogs conform to familiar conventions of spoken discourse, the user's experience is assuredly more comfortable. I also argue that this "naturalness" approach facilitates listening comprehension. Come find out a linguist's low-down, for example, on what's strange about the following conversation:
411 Operator: Hi, this is Claire. What city?
Caller: Mountain View.
411 Operator: All right.
Caller: Cost Plus.
Recording: The number you requested, six five zero, nine six one, six zero six six, can be automatically dialed by pressing one now. An additional charge will apply."
James Giangola works in the Dialog R&D group at Nuance Communications. He comes to Nuance from General Magic, where he was VUI Designer for Portico, a very natural-sounding virtual assistant, and before that, from Sensory, Inc. James has worked on dialog design, prompt writing, concatenation planning, and voice actor direction for a number of projects in languages other than American English, including Brazilian Portuguese (for Banco Itaú and Companhia Telefônica do Brasil Central), Swiss German (UBS), French (TOMY Europe), Japanese (Sega), and British English (British Telecom). Besides his work in speech technology, James has ten years of experience as a classroom language teacher. He holds linguistics degrees from Brown University, the Monterey Institute of International Studies, and UC San Diego. He has recently received an offer for publication of his research on the pronunciation patterns of Brazilian Portuguese.Attendance: 80+Pizza Sponsored by Tellme Networks, Inc.This 2nd meeting of the SIG built upon the great momentum established in October. Over 80 people came to Tellme Network's corporate headquarters to hear Brian Siberman from SpeechWorks speak about their Open Speech Web intiative.This was also a quite commercial talk, focusing tightly on the SpeechWorks product stack. Q&A centered around their long-awaited open source VoiceXML browser project (in association with Carnegie Mellon University), and how application server vendors will figure into the ultimate success of broad enterprise adoption of voice in the context of multi-device deployments.