Recognizing the value of close captioning, researchers from the Electrical and Electronics Engineering Institute of the University of the Philippines Diliman (EEEI-UPD) are developing real-time closed captioning system for the Filipino language. “Research studies have shown that closed-captioned television has educational and sociological implications, including vocabulary development, the opportunity to observe good sentence structure, and motivation for reading,” the project team says.
The closed captioning system consists of an Automatic Speech Recognition (ASR) at the front end, which recognizes spoken sentences, transcribes them to text and subsequently shown in real-time on the television screen. The ASR system uses statistical methods and open source tools while the characterization of the bases of the system and the acoustic and the language models were based on a number of methods. Optimization is based on a lattice scoring technique, where an error region is traversed to find the parameters for the system that would yield the smallest word error rate. The optimization allows for the ASR to be integrated to an actual closed captioning software.
The project is a component of the Interdisciplinary Signal Processing for Pinoys (ISIP) Programfunded by the Philippine Council for Industry, Energy and Emerging Technology Research and Development of the Department of Science and Technology (DOST-PCIEERD) and the University of the Philippines Diliman. For more information, please contact the Program Leader, Dr. Rhandley Cajoteof the Digital Signal Processing Laboratory of the EEEI-UPD.