WASHINGTON, D.C., NOVEMBER 5, 2018 – With thousands of fans clapping, chanting, shouting and jeering, college basketball balls can be…
WASHINGTON, D.C., NOVEMBER 5, 2018 – With thousands of fans clapping, chanting, shouting and jeering, college basketball balls can be almost deafly loud. Some arenas have decibel meters, which, exactly or not, give a certain indication of the volume of volume provided by spectators and audio systems. But the audience’s sound is rarely the focus of scientific investigation.
“When it comes to literature, it’s mostly been a researcher trying to get around,” noted Brooks Butler, a physics student at Brigham Young University and a member of the BYU team who will present research at the US 176th meeting in Acoustical Society, held in conjunction with the Canadian Acoustic Association 2018 Acoustic Week in Canada, November 5-9, at the Victoria Conference Center in Victoria, Canada.
“Crowds noise is typically treated as background interference ̵
1; something to sharpen out.” But BYU researchers thought the audience’s sound was worthy of their own investigation. In particular, they wanted to see if machine learning algorithms were able to select patterns within the raw acoustic data indicating what the audience did at a given time, thereby giving clues about what happened in the game itself. A possible application of this may be an early discovery of unpopular or violent behavior – even though this idea has not been tested.
The BYU team made high-quality acoustic measurements during men’s and women’s basketball games at the university, which later did the same for football and volleyball games. They broke the games in half-second intervals, measure the frequency content (as shown on spectrogram), sound levels, the ratio of maximum and minimum noise levels within a certain time interval and other variables. Then, they applied signal processing tools that identified 512 different acoustic functions consisting of different frequency bands, amplitudes, etc.
The group used these variables to construct a 512 dimensional space using machine learning techniques to perform a computerized cluster analysis of this complicated multidimensional empire .
BYU physics professor Kent Gee was a main researcher on the project together with professors Mark Transtrum and Sean Warnick. Together they led a team of several students focusing on various aspects of the problem, including data collection, analysis and machine learning.
Explain the process with a simple analogy. “Suppose you have a plot of points on a two-dimensional, x-y chart and measure the distance between those points,” he said. “You may see that the points are joined into three lumps or clusters. We did something similar to our 512-dimensional space, but of course you need a computer to keep track of all this.”
The so-called “K-means clustering” analysis as they ran revealed six separate clusters corresponding to what happened in the arena, depending on whether people were cheering, singing, boiling, silent or letting the speakers dominate the sound image.
Thus, Gee and his colleagues could measure the audience’s emotional state, simply from a machine-based analysis of audio data. “An important possible application of our research,” he said, “may be early detection of worrying or violent patterns of behavior.”
Presentation # 1pSP11, “Clustering Analysis of Public Audio from Collegial Basketball Games,” by Brooks A. Butler, Mylan R. Cook, Kent L. Gee, Mark K. Transtrum, Sean Warnick, Eric Todd and Harald Larsen will be Monday 5 November, 16:25 in the Shaughnessy (Fe) room at the Victoria Conference Center in Victoria, British Columbia, Canada.
———————– MORE MEETING INFORMATION ——————— –
Main meeting website: https: /
Meeting Technical Program: https: /
Hotel Information: https: /
WORLD WIDE PRESS ROOM
In the coming weeks, ASA’s World Wide Press Room will be updated with further tips on dozens of news stories and spell-paper, which are 300-800 words summaries of presentations written by researchers for a public audience and accompanied by photos, audio and video. You can visit the site, beginning in late October, at http: // acoustics.
We will grant free registration to journalists and professional freelance journalists. If you are scratch and want to participate, contact Rhys Leahy or AIP Media Line ([email protected], 301-209-3090). We can also help you set up interviews and get pictures, audio clips or background information.
LIVE MEDIA WEBCAST
A press release with a selection of newsworthy research will be live live from the conference Tuesday 6 November 2018 Times and topics to be announced. Media members should sign up in advance at http: // aipwebcasting.
The Acoustical Society of America (ASA) is the premier international scientific community in acoustics devoted to science and technology of sound. Its 7,000 members worldwide represent a wide range of acoustics. ASA publications include Journal of the Acoustical Society of America (The World’s Leading Acoustics Magazine), Acoustics Today The Newspaper, Books and Standards for Acoustics. The society also holds two important scientific meetings each year. For more information about ASA, visit https: /
The Canadian Acoustic Association (CAA) is a professional interdisciplinary organization that promotes communication among people working in all areas of acoustics in Canada; promotes growth and practical application of acoustics knowledge; Encourages education, research, environmental protection and employment in acoustics; and is an umbrella organization through which general issues in education, employment and research can be managed at national and interdisciplinary level. For more information about CAA, visit http: // caa-aca.