Among many people, public speaking is the biggest fear. Now, researchers at the University of Rochester's Human-Computer Interaction Group have developed an intelligent user interface for "smart glasses" that feeds speakers with volume modulation and speech speed in real time, while minimizing distractions.
The Rochester team described the system in a paper presented at the Computer-Machine Intelligent User Interface Association (IUI) conference in Atlanta on Tuesday, March 31, where they used the word "rhema" in Greek.
Smart glasses with Rhema installed can record speakers, transmit audio to a server to automatically analyze volume and speech rate, and then present data to the speakers in real time. This feedback allows the speaker to adjust the volume and speed of speech or continue operation as before.
Ehsan Hoque, a senior author of the paper and an assistant professor of computer science, personally used the system during a lecture last semester. He said: "My wife always tells me that I speak too softly." "Rhema reminds me to keep the volume. It was a great experience." He felt that this practice helped him understand his body better, even if he The same is true without smart glasses.
In this article, Hoque and his students M. Iftekhar Tanveer and Emy Lin explain that providing feedback in real-time during a lecture poses some challenges. "One challenge is to get the speakers to understand their speech performance without distracting them," they wrote. "A large enough distraction can lead to unnatural behaviors such as stuttering or awkward pauses. Second, head-mounted displays are located close to the eyes, which can lead to inadvertent distractions. "
Tanveer, the paper's lead author, explained that overcoming these challenges was their focus. To that end, they tested it with 30 English-speaking natives using Google Glasses. They evaluated different options for providing feedback. They tried different colors (such as a traffic light system), text and graphics, and there was no feedback at all (control). They also tried a continuously changing display and a sparse feedback system by which the speaker could not see the glasses most of the time, and then saw feedback for only a few seconds. After the user test, every 20 seconds in the form of words ("loud", "slower", if the speaker is doing well, nothing) and other forms of feedback are considered by most test users to be the most successful .
The usage scenario of real-time feedback system based on Google Glass. Image source: M. Iftekhar Tanveer, etc.
Researchers also emphasize that, overall, users believe this helps them improve delivery compared to users who receive continuous feedback without feedback at all. They also introduced the system from the audience's perspective and recruited 10 Mechanical Turk workers.
"We want to check whether the feedback that the speaker sees on the glasses is distracting the listener," Hawk said. "We also want the listener to evaluate three conditions: a person appears spontaneous, has too many pauses, and uses too many filler words. And maintain good eye contact, these three conditions are: word feedback, continuous feedback and no feedback.
However, judging by the mechanical Turkish workers, there was no statistically significant difference between the three groups in eye contact, the use of filler words, distraction and stiffness. As part of future work, researchers hope to test their system with a richer audience with Toastmasters International members.
Researchers also believe that real-time feedback displayed in a private, non-invasive way is useful for people with social difficulties (for example, Asperger's syndrome), even for those working in customer service.
