Pediatric Voice Program

The Center for Pediatric Auditory and Speech Sciences (CPASS) is dedicated to translational research in the audiological and speech sciences with the goal of advancing the practice of pediatric health care in related areas, while simultaneously advancing the underlying science.

The Craniofacial Outcomes Research Laboratory (CORL), within CPASS, is one of a few labs of its kind in the United States using state-of-the-art equipment to study speech and resonance disorders related to cleft palate and craniofacial anomalies.

Work with measures of vocal track pressure and airflow using PERCI-SARS (a software-based instrumentation system), InvTool (a program used to record speech for synthesizing) and Nasometry (a non-invasive technique used to measure the size of the velopharyngeal opening combined with the measurement of nasal emission) continues to prove invaluable in advancing diagnostic abilities to assess resonance and speech disorders and to objectively evaluate the outcomes of the interventions undertaken to improve them.

These high-tech instruments are also used by Nemours clinicians to determine treatment planning and monitor progress for patient care.

Work within the Speech Research Laboratory under CPASS is focused on the development of computer-based solutions for individuals, especially children, with speech and language disorders and the clinicians who care for them. Many studies are funded by the National Institutes of Health (NIH), among others.

One line of investigation seeks to refine and commercialize a voice banking system called ModelTalker that allows individuals to easily capture their own voice for later speech synthesis. Children and adults who may be losing their voice due to neuromuscular disorders can use ModelTalker to “bank” their own personalized voice for use when they are no longer able to speak. The software is in the final stages of refinement, and plans are underway for formal system evaluation with children, adults, and elderly participants.

The laboratory is also working with colleagues at Northeastern University to allow ModelTalker voices to be made to sound more like the natural voice of disabled children who are able to make a few vowel sounds but are too dysarthric to bank their own voice. Technology developed for utterance verification within the ModelTalker system is also being used to design automatic evaluation tools for functional hearing tests.

In conjunction with a small business partner, Compreval, and the House Ear Institute (HEI), lab researchers have been developing an automatic speech recognition (ASR) system to grade the responses of individuals taking the Hearing-In-Noise Test (HINT). Currently, subject responses in this test must be manually graded by a human listener. The ASR system automates the grading, removing the need for a trained grader, and reduces the chance of grader bias during the test. The automatic grading process is currently being tested at HEI.

Another line of investigation involves research on computer speech processing and speech recognition to support clinical speech assessment and speech training. Investigators are extending previous work on the Speech Training Assessment and Remediation system (STAR) to develop a computer-based speech therapy assistance program for children with cochlear implants. This system uses speech recognition technology to guide children into correct speech. As part of that project, the lab is currently conducting a longitudinal study to collect and analyze the developing speech of children with cochlear implants. Because many projects involve the use of synthesized speech for aural rehabilitation, the lab is planning a study to systematically investigate how hearing-impaired listeners differ from normal-hearing listeners in their perception of synthesized speech of various qualities. The lab is also continuing to analyze a large database of children’s speech collected in the laboratory. And, the lab is currently investigating the use of hidden Markov models (HMMs) to automatically segment the speech for further analysis.