Page 69 - Winter2018
P. 69

 Those eligible for the ASA cash prize awards were chosen by the three ASA judges from the Intel ISEF finalists. They first searched through the 1,800 titles and abstracts for acoustics relevance (defined in the broadest sense). The relevant proj- ects were then reviewed and cut to the 35 top projects that had acoustics as a primary focus of the research. Most, if not all, of the 13 technical areas of ASA were represented.
The ASA judges then met and discussed the 35 projects, and each judge chose to be a primary or secondary judge for var- ious projects. Next, the primary and secondary judges pre- judged the projects by reading the student’s poster boards (see Figure 1), documentation, and lab notebooks in the absence of the student. From there, the judges selected 20 projects for student interviews. A total of five semifinalists were selected and reinterviewed once or twice more by the entire panel of ASA judges.
After much deliberation, first, second, and third place final- ists were chosen and received cash prizes for themselves, their schools, and their mentors. The first place winner re- ceived $1,500 plus $200 for the school and $500 for the men- tor. The second place finalist received $1,000 plus $100 for the school and $250 for the mentor. The third place finalist received $600 plus $150 cash prize for the mentor. An hon- orable mention certificate was also awarded. Additionally, the student winners have been invited to attend the upcom- ing ASA annual meeting in Victoria, BC, Canada. For pho- tos of this and past Special Award Ceremonies, see the Intel ISEF ASA Flickr album at
The first place ASA winner was Anwesha Mukherjee (Ro- botics and Intelligent Machines Category). Ms. Mukherjee is a 10th grader at Westview High School, Portland, OR. She was mentored by teacher Debbie Cooper with a project enti- tled “A Novel Approach to Recognize Emotion from Speech Using Machine Learning Algorithms to Aid Social Interac- tion of Kids with Autism.” The goal of this project was to aid autistic children in their development of empathy and social interactions by using cues from speech to help compensate for difficulties in reading facial expressions. Ms. Mukher- jee’s contribution over previous work was developing a heu- ristic weighting of the mel-frequency cepstral coefficients (MFCCs) that improve classifier accuracy by an additional 3-12%. She investigated a number of machine-learning al- gorithms and found that the multinomial logistic regression provided the most accuracy for her metrics. Her system was trained and tested using a database of expressive speech. See her full project abstract at Ms. Mukherjee
Figure 1. Poster session at the Intel International Science and Engineering Fair.
also received the Intel ISEF Best of Robotics and Intelligence Category Second Place Award.
The second place ASA winner was Sharmi Shah, an 11th grader from Colonia High School, Colonia, NJ, mentored by teacher James Danch (Physics and Astronomy Category). Ms. Shah’s project was “Speech Intelligibility Analysis of Sound-Modulated Laser Signal Countermeasures.” It is well established that laser light can be used as a spying device by measuring the modulation of the reflected light from a window. Ms. Shah investigated two coatings for their ability to diffuse the laser light and prevent eavesdropping while maintaining optical clarity. She evaluated the two coatings plus a control by using speech-recognition software to ana- lyze speech recorded by the laser. Silica nanoparticle-epoxy residue reduced intelligibility of the chosen speech by 47% while nanoparticle-dimethylsiloxane fully precluded speech intelligibility by the algorithms. Although the coatings ap- peared transparent, transmission spectra were measured for the control and coated samples and found that the light transmission was very minorly impacted. The project ab- stract is available at Ms. Shah also received the US Air Force First Place Award in Physics and Astron- omy, the National Security Agency (NSA) Research Direc- torate Physical Sciences First Place Award, and the Intel ISEF Best of Physics and Astronomy Category Fourth Place Award.
An 11th grader, Gabrielle Liu (Systems Software Category), from Ravenwood High School, Brentwood, TN, and men- tored by teacher Peter Lowen, took third place. Ms. Liu’s project, “Preventing Domestic Violence Using Emotion Recognition in Speech” also used machine learning to moni- tor emotions but as a screen to pick up domestic violence. She envisions incorporating her algorithms in the burgeon- ing prevalent use of artificial intelligence personal assistants.
 Winter 2018 | Acoustics Today | 67

   67   68   69   70   71