A team of Electrical Engineering researchers from the Indian Institute of Science is working on developing software tools that can help develop better speech therapies. The team from the Signal Processing, Interpretation and REpresentation Laboratory (SPIRE LAB), led by Dr. Prasanta Kumar Ghosh, uses novel signal processing techniques to better understand human speech production.
Signal processing is an area of research which deals with acquisition, analysis, interpretation and modelling of a variety of signals. For example, speech signal models are used to parameterize the human voice signal, which is critical for transmitting the message and getting back the voice signal at the receiving end. Without signal processing, we wouldn't have been able to speak to our loved ones through telephone. Researchers from SPIRE LAB are developing new signal processing technique to help better understand speech problems, and develop appropriate interventions.
“We use a procedure called ‘electromagnetic articulography’ for recording movements of the articulators in the mouth during speech and are in the process of expanding the tool-repertoire”, says Dr. Ghosh. Dr. Ghosh is passionate about translating the lab’s research to solve real-world problems. Articulation of speech involves moving the muscles of our mouth and tongue in an orderly, learnt fashion to produce meaningful sounds. In many speech pathologies, articulation is adversely affected, which leads to wrongly pronounced speech-sounds. For example, in a disease called amyotrophic lateral sclerosis (ALS), nerve cells controlling muscle movement are progressively destroyed, leading to muscle weakness and degeneration. The muscles that help produce speech are also affected, making it difficult for ALS patients to pronounce words correctly. The SPIRE lab is collaborating with various hospitals to record speech articulatory movements from ALS patients. This will give a better understanding of the specific aspect of articulation that gets affected in patients with ALS and other disorders having speech-pathology symptoms.
“A better understanding of the underlying processes can help in building models which may be used to screen patients for specific diseases and start therapy early on”, explains Dr. Ghosh.
Patients with disorders like ALS, Cerebral Palsy, severe asthma or who have had their larynx (voice box) removed, are often are unable to produce more than whispering or slurred sounds to communicate. Dr. Ghosh and his team have recently developed a method to detect whispered speech even in a noisy recording. This can lead to the development of a device which can reconstruct normal speech from the whispered speech of the patient enabling easy vocal communication.
“We hope that the research done in the lab helps in building technology for social use”, comments Dr. Ghosh. His team members are already developing speech analysis software tools which can be used to track the progress of learning a new language. One can see its benefits for the call centre industry, wherein many novice English speakers are initially trained to increase their fluency and such an application will be very helpful in tracking the improvement and getting feedback. With a brimful of research ideas and unbridled enthusiasm in the domain of speech processing, the SPIRE lab is set for a long and interesting scientific voyage.
Dr. Prasanta Kumar Ghosh is an assistant professor at the Department of Electrical Engineering, IISc.
Lab webpage: http://spire.ee.iisc.ernet.in/spire/