Skip to main content

Shonda Bernadin, Ph.D.

Shonda Bernadin
Associate Professor
Education
  • Ph.D., Electrical Engineering, Florida State University, Tallahassee, FL, 2003
  • M.S., Electrical and Computer Engineering, University of Florida, Gainesville, FL, 1999
  • B.S., Electrical Engineering, Florida A&M University, Tallahassee, FL, 1997
Research Interests

Digital Signal Processing, speech recognition, characterization of speech and sound waves using wavelets and other time-frequency analysis methods, language modeling, data analysis techniques.

Learning behaviors of engineering students; assessment and evaluation of student performance; teaching and learning strategies that produce an effective learning environment.

Engineering education research on increasing retention, progression, graduation rates of students, improving quality of teaching and learning engineering, increasing diversity.

Broadening Participation Research on interest, motivation and perception in minority engineering students.

 

Current Research Projects

Speech and Image Processing Research Projects:

Engineering Education Research:

  • Characterizing acoustic correlates for electroglottographic (EGG) signals;
    In this project, we examine the acoustic and electroglottographic features in the characterization of passaggio in female singers. Three groups of female singers were instructed to sing the notes of the scale for one octave using an "ah" vowel. When singing this octave they sang through a register shift, which is called 'passaggio'. Their singing voices were recorded in a two-channel dataset. The first channel captured the acoustic signal using a microphone and the second channel captured the electroglottographic (EGG) signal using an EGG instrument. This study used voice analysis software to analyze the features of the two-channel dataset that contribute to the characterization of the passaggio in female singers. Glottal measurements can give more robust information on precise glottal opening and closing moments using the derivative of the EGG signal (DEGG). The results of this investigation will also provide an analytical framework for calculating the DEGG in female singers.
  • Speech and image interfacing with autonomous quadcopter;
    The primary goal of this project is to design and build an autonomous quadrotor helicopter (i.e. quadcopter) with interface capabilities to incorporate intelligent applications for image and speech processing. This work will help to develop a research platform for exploring perception methods for multimodal intelligent autonomous systems, which are extremely useful in aerial surveillance and search-and-rescue operations. An autonomous system that supports aerial motion using an Arduino microcontroller for enhanced base layer and navigation controls was constructed. Software interfacing and speech recognition are currently being investigated for system implementation.
  • Speech intelligibility calculations for surface electromyographic (EMG) speech data
    In previous work (Wohlert and Smith, 2002) it was determined that variability in children's speech production is reflected in upper lip muscle activity using electromyographic (EMG) data across repetitions of a phrase. Later studies (MacPherson and Smith, 2013) showed that a lip aperature variability index, which represents the difference in upper lip displacement and lower lip displacement, can also be used as a reliable variability measure from kinematic data to determine the effects of multiple repetitions of the same utterance on speech motor production. This project is an extension of previous work and examines the orofacial muscle activity patterns (i.e. EMG data) during speech production in efforts to quantify EMG variability. This information can yield significant insights into how well the speech motor system is functioning in different groups of speakers (e.g., healthy young adults vs. healthy older adults).
  • 3D image reconstruction of cultural artifacts
    In this project we create 3-dimensional (3D) digital images of selected artifacts from the Slavery in the Old South collection located at the prestigious Meek-Eaton Southeastern Regional Black Archives Research Center and Museum (Black Archives) at FAMU These items include a slave breeding bed, pickaninny yoke, topsy-turvy doll, leg shackles, hand shackles, and a leather whip. Digital photogrammetry will be used to recreate the images. Digital photogrammetry is an imaging process that uses cameras to take several 2-dimensional (2D) photographs of an object at many different angles to capture the special intricacies of the object. The photographs are then converted to 3D images using image processing techniques. According to researchers, digital photogrammetry is an ideal method to use for preserving artifacts, including archeological findings and increasing community access and engagement to artifacts.
  • Evidence-Centered Instructional Design in Signal Processing Courses;
    Many engineering courses are designed based on course objectives and outcomes which typically correlate to program goals and outcomes. It is implied that students' have gained the necessary knowledge and skills they need in order to be successful in subsequent courses. Yet, in some of these courses students struggle with basic, fundamental concepts that impact their understanding and success rate in subsequent courses. These "foundational knowledge gaps" (i.e. pre-requisite knowledge gaps) may impact a student's identity development as a professional engineer. In this project we investigate a way to accurately measure foundational knowledge and identify knowledge gaps in the learning process using the Evidence-Centered Design (ECD) approach. ECD is a structured method based on Bayesian inference networks that is used to develop assessment tasks to accurately measure student proficiency or understanding based on "inference-by-observation". ECD is explored in the design of a Signal Processing course to measure student proficiency and understanding of course content.
  • Impact of Achievement Emotions on Student Performance and Persistence in Engineering
    This project seeks to investigate the underlying issues that impact the persistence of engineering students using a mixed-methods design approach that explores the impact of achievement emotions delineated in the control value theory. In Tinto's retention model (Tinto, 1975) he identifies academic and social integration factors that are predictive of student persistence and retention. These factors are studied in this project to identify psychosocial factors related to persistence of engineering students at the FAMU-FSU College of Engineering.
Publications
  1. D. Ludger, D. Carey, N. McNeal-Parham, S. Bernadin, "Intelligent Autonomous Systems for HCI", Florida Conference on Recent Advancements in Robotics, Florida International University, Miami, Florida, May 8-10, 2014
  2. Bernadin, Shonda, Megan MacPherson, Tejal Udhan, "Optimizing Performance of a Speech Analysis System for Orofacial Muscle Activity Data", Abstract Proceedings of the 2014 International Association of Journals and Conferences (IAJC)/ISAM Joint International Conference in Orlando, FL, September 27-29, 2014
  3. Bernadin, Shonda, Stephen Alexander, "Investigating Kalman Filtering for Charge Estimation in Battery Management Systems", Abstract Proceedings of the 2014 IAJC/ISAM Joint International Conference in Orlando, FL, September 27-29, 2014
  4. Bernadin, Shonda, Tejal Udhan, "Development of a Task Model using Evidence-Centered Assessment Design in a Speech signal Processing Course", Proceedings of the American Society for Engineering Education (ASEE) Southeast Regional Conference 2014, Mercer University, March 31-April 1, 2014