抄録
Voice stress analysis (VSA) is a technique which tries to detect speakers’lies from their voice.
One of the VSA-based computer programs called“Layered Voice Analysis(LVA)”is the core engine
of some lie detector software and has been introduced to practical scenes such as airports, police,
and private companies in western countries; though its scientific validity and reliability are
controversial.
On the other hand, some studies made the suggestions that LVA could work as the monitor of
negative mental states(stress, strain, or anxiety). Present study tested the possibility that LVA is
sensitive to speakers’negative mood. Seven male participants seated in front of the personal
computer and read aloud the questions of Profile of Mood States(POMS) presented on the screen,
then replied to them with their voice and mouse-click. We applied LVA to participants’voice of
each questions and replies. Then we calculate Spearman's rank correlation between LVA parameters
and POMS subscales.
Our data showed that two subscales of POMS, T-A and A-H, significantly correlated with the
LVA parameter, Lie Stress, representing levels of stress associates telling lies. There was also
significant correlation between F subscale of POMS and JQ parameter, indicating anxiety and stress
in LVA. However, later result is contrary to the previous study reported that F subscale was
negatively correlated with JQ parameter. This suggests that LVA could have some problems in
measuring stability and therefore could not detect negative mental mood effectively from healthy
adults’voice.
