Project Listing  


1.      Enhancing Speech Quality for IDEN Phones
Funding: Motorola
Students: Harsha Sathyendra

We are developing techniques for enhancing the intelligibility of speech on Motorola cell phones. Various directions include automatic techniques for boosting the perceived loudness and intelligibility of speech (without increasing signal energy), beamforming and bandwidth extension.


2.      Cell phone-based hearing aid
Funding: Motorola
Collaborators: Dr. Alice Holmes
Students: Meena Ramani, Qing Yang

     Being able to communicate via a cell has become an integral part of living in our society regardless of age.  Unfortunately in many instances hearing on the cell can be difficult due to the noise environment the person is in or due to reduced hearing abilities of the listener. Currently, one in every ten Americans (more than 28 million people) is estimated to have hearing loss.  The prevalence of hearing loss increases with age to nearly one in every three adults over 65.  In addition, cell phones are used by people of all ages in noisy environments, i.e., teenagers at the mall, adults in the workplace, etc.   We are studying the effects of hearing loss and noise on communication abilities on a cell phone and the development of a cell phone that can be adapted to an individual's unique hearing abilities and the noise environment in which the cell phone is being used.


3.      Automatic Accent reduction
Funding: KAIM
Students: Kwansun Cho

      We are developing tools to reduce foreign-accents.


4.      Disordered speech evaluation
Funding: UF Seed Grant
Collaborators: Dr. Rahul Shrivastav and Dr. Chris Sapienza
Students: Savyasachi Singh

Speech quality assessment methods are necessary for evaluating and documenting treatment outcomes of patients suffering from degraded speech due to Parkinson's disease, stroke or other disease processes. Subjective methods of speech quality assessment are more accurate and more robust than objective methods but are time-consuming and costly. We propose a novel objective measure of speech quality assessment that builds on traditional speech processing techniques such as dynamic time warping (DTW) and the Itakura-Saito (IS) distortion measure. Initial results show that our objective measure correlates well with the more expensive subjective methods.


5.      Biologically inspired speech recognition
Collaborators: Dr. Mark Skowronski
Students: Ismail Uysal

The performance of today's automatic speech recognition systems pale in comparison to human ability due to the adaptability of the auditory system to the many sources of variability associated with the both the speech signal and noisy recording environments. Though the biological basis for speech recognition is largely unknown, we believe that the role of spike-based coding and processing is crucial for noise robust recognition. This paper describes a biologically plausible algorithm that exploits the use of spikes in both the feature extraction and the recognition stages.