About Me

I am a fifth year PhD student at the University of Michigan, where I work with Dr. Emily Mower Provost on speech processing and machine learning for healthcare monitoring. My current focus is on disfluency detection, where disfluencies are interruptions in the typical flow of speech. Disfluency detection may be used to help track the progression of stuttering disorders and cognitive impairment.

Before I was a graduate student, I earned my BS in math from University of Michigan. I also spent four years as a research associate at the Institute for Defense Analyses.

Research

Broadly speaking, I am interested in using signal processing and machine learning methods to improve assistive technology. Below are highlights from some projects I am working on.

Off-the-shelf disfluency detection

triplet_example
  • Evaluated the use of within-corpus versus off-the-shelf disfluency detection models for pathological speech.
  • Improved baseline repetition and revision detection by adding a fine-tuning step with triplet loss.

Detecting reading errors and disfluencies

distance_matrix
  • Applied Levenshtein distance to count the number of errors and disfluencies (ex. # of inserted words) from a transcript of participants reading a template.
  • Modeled features with linear regression to predict cognitive impairment severity for people with Parkinson's Disease.
  • Currently exploring methods for improving the ASR transcription of these errors and disfluencies.

Evaluating vowel distortion

vowel_dist_pipeline
  • Introduced method to decompose vowel waveform into trend + fluctuation component.
  • Demonstrated use of Hurst exponent for capturing vowel irregularities for people with manifest Huntington disease.
  • Currently exploring how these vowel irregularities vary across different types of vowels and if we can combine this method with automatically segmented vowels.

Publications

Amrit Romana, Minxue Niu, Matthew Perez, Angela Roberts and Emily Mower Provost. ”Enabling Off-the-Shelf Disfluency Detection and Categorization for Pathological Speech.” Interspeech.Incheon, South Korea. September 2022.

Amrit Romana, John Bandon, Matthew Perez, Stephanie Gutierrez, Richard Richter, Angela Roberts and Emily Mower Provost. ”Automatically Detecting Errors and Disfluencies in Read Speech to Predict Cognitive Impairment in People with Parkinson’s Disease.” Interspeech. Brno, Czech Republic. August 2021.

Matthew Perez, Amrit Romana, Angela Roberts, Noelle Carlozzi, Jennifer Ann Miner, Praveen Dayalu and Emily Mower Provost. ”Articulatory Coordination for Speech Motor Tracking in Huntington Disease.” Interspeech. Brno, Czech Republic. August 2021.

Amrit Romana, John Bandon, Noelle Carlozzi, Angela Roberts, Emily Mower Provost. “Classification of Manifest Huntington Disease using Vowel Distortion Measures.” Interspeech. Shanghai, China. October 2020.