Speech Tactons Improve Speech Warnings for Drivers
Josh Mills
2016-02-25
Focus of this study
This study evaluated a set of speech and tactile warnings, looking primarily at multi-modal warnings that combined both Speech warnings and Tactile warnings (Speech Tactons)
What are speech tactons?
They retain the rhythm of speech and used different levels of roughness and intensity to convey urgency
This study evaluated a set of speech and tactile warnings, looking primarily at multi-modal warnings that combined both Speech warnings and Tactile warnings (Speech Tactons)
How is speech perceived in warnings in terms of its urgency, annoyance, and alerting effectiveness?
Speaking signal words urgently
They found that the signal word “Danger” was perceived as more urgent compared to the words “Warning” and “Caution”,
which in turn were perceived as more urgent compared to “Notice”. Also, Signal words spoken urgently created higher
ratings compared to non-urgently, which in turn were higher compared to words spoken in a monotone manner
How to increase richness in tactile cues/warnings?
Vary parameters such as rhythm, roughness and intensity
Varying parameters such as rhythm and roughness (amplitude modulation on the original waveform that provides the vibration)
of such messages enabled the design of richer cues without cost in their recognition accuracy, as long as a reduced number
of different levels for roughness was used.
In this study..
In this study, tactile cues are used along with audio, varying the rhythm of the cues, by imitating the rhythm of speech
Three urgency levels:
The words “Danger!”, “Warning!”, “Notice!” were added before each Lh, Lm and Ll. the ac- tor was instructed
to speak messages of LH in an urgent manner, as if a loved one was in imminent danger. Accord- ingly, LM messages
were spoken non-urgently, as if in a friendly conversation with nothing interesting about the situation and LL messages
were spoken in a monotone, deadpan manner
Experiment 1
The first experiment investigated the subjective responses provided by participants when exposed to the warnings.
A 6×3×4 within subjects design was used with Message, Modality and Design as the independent variables and Perceived Urgency (PU), Perceived Annoy- ance (PA) and Perceived Alerting Effectiveness (PAE) as the dependent ones
A 6×3×4 within subjects design was used with Message, Modality and Design as the independent variables and Perceived Urgency (PU), Perceived Annoy- ance (PA) and Perceived Alerting Effectiveness (PAE) as the dependent ones
Set-up
participants were exposed to the 54 cues (6 A, 24 T, 24 AT) in a random order, to familiarize them with the signals. Afterwards, they were again presented with the cues and asked to rate them all in terms of PA, PU and PAE, by
completing a 5-point Likert scale for each rating
Results
Contrasts revealed that the significant differences in ratings of PU described above were not present in modality T
From Experiment 1 it was clear that AT was rated higher in all measures compared to A and T, showing clear evidence of the usefulness of the designed cues when modalities were combined
Further, it was evident that the urgency designed in the warnings was reflected in their PU. PAE escalated according to PU, indicating that messages signi- fying situations of higher importance were regarded as more useful
PA was higher for messages of both LH and LL and lower for LM
a further indi- cation of the higher utility of the cues when presented mul- timodally
Experiment 2
Experiment 2 investigated the recognition accuracy of participants when exposed to the T warnings
Results
The results of PU are a clear indication that the messages de- signed conveyed the desired urgency. AT messages had higher
ratings of PU, which is an improvement compared to A or T messages, when a situa- tion of high criticality needs to be
conveyed
This study shows that tactile cues can improve responses to speech warnings. This adds to the existing body of work,
suggesting en- hanced responses to multimodal signals versus unimodal ones
T messages did not present highly differ- ent ratings of PU, adding to the argument that such cues work better when used multimodally
In terms of Tactile designs..
In terms of tactile designs, it was clear that intensity was the main factor that led to higher PU
In terms of Tactile designs..
This strengthens the evidence that intensity of the tactile part is useful to create more urgent messages and can be
compared with [2,3,9], where high intensity of audio af- fected PU ratings
Results
Results showed that the addition of these new cues improved subjective responses of drivers to speech warnings.
The warnings were clearly distinguished in terms of urgency, their annoyance was low and their alerting effectiveness
changed similarly to urgency, increasing for more urgent messages and for multimodal cuesWe suggest the use of Speech
Tactons to accompany speech warnings, so as to make use of the observed ad- vantages of multimodal cues, but not for
low urgency situations, to avoid annoyance
Common Themes/What I took away from this..
Multi-modality is king, and perceived annoyance is actually a positive, but only when considering Lh scenarios
