Humans adapt speech to computer: Study (fwd)

[Phil Cash Cash]

Humans adapt speech to computer: Study
By Staff, CNETAsia
27/6/2003
URL: http://asia.cnet.com/newstech/systems/0,39001153,39137979,00.htm

A study has shown that humans tend to make their voices sound like the
computers they are speaking to--and when the computer switches to a new
voice, so will the human user.

Researchers at the Oregon Health and Science University in the U.S. were
studying human-computer audio interfaces, important in applications
such as text-to-speech (TTS) and speech-to-text (STT).

"We discovered that people subconsciously tailor their voice to the
computer's voice. So, for example, if the computer voice is talking
more quietly, the person interacting with the computer will talk more
quietly," said according to Sharon Oviatt, an experimental psychologist
who led the study.

"This is a major new source of variability in users' spoken language to
computers, which is opening up new scientific directions for mobile
audio interface design," she said.

Speech convergence, a phenomenon seen in groups of people when their
speech patterns begin to mimic each other, can be observed to occur in
human-machine interaction as well, she said.

In the study, children conversed with animated characters seen on a
handheld computer. The animations had voices from a TTS software
engine. The scientists found that in factors such as amplitude,
duration and pitch, the children tended to copy the TTS voices. Four
different voices were tested to see how closely the children matched
their voices to the computer's.

The team will carry on with tests so that they can model how much speech
changes when interacting with computers, so that computer audio
interfaces can be made more reliable, especially in noisy natural
surroundings.

Recently, a team of Japanese scientists found that when computers were
given voices that sound like the human user, the user tends to like the
computer more, and find it easier to use.