Acoustic Features for Profiling Mobile Users of Conversational Interfaces
David Toney, David Feinberg, Korin Richmond
conversational interfaces for tourist information
personalising interaction based on what we can tell from their voice – gender
age
accent
emotion
(location)
measuments – pitch
jitter
shimmer
harmonics-to-noise ratio
mobiles are used in noisy environments
gender 94.4% accuracy
age – no useful learning
variation found, but too much background noise
(macrophone speech sample database)
looking at finding height from voice(!)
*****
Butler: A Universal Speech Interface for Mobile Environments
Botond Pakucs
user’s needs may rapidly change in dynamic and mobile environments
need to switch transparently and seemlessly between services
multitask
lack of speech interface consistency
many different mental models and therefore increased cognitive load
usability and HCI needs to be considered for whole environemtn, not individual services
‘light on’ – which lights?
service may take proactive initiative
coordination among services is necessary, and support user control
one speech interface for each user, not each product/service
how?
integrated into personal device (phone, pda)
standardised domain data and dialogue management
test system
lunch info, public transport info
based on already existing web services
http://www.speech.kth.se/~botte
contact
email:
chris is at anti-mega.com
Twitter:
@antimega
iChat/AIM:
antimega77