Computers applied for the recognition of Hindi syllables

The adaptive pattern recognition theory has been programmed on the IBM 1130 computer and applied for the recognition of speech patterns initially in the form of sonograms of CVC Hindi syllables spoken in isolation. A set of thirty six Hindi syllables spoken by three male speakers was used in the learning and recognition phases of the system. Two experiments were performed. In the first, the computer was trained by using the syllables spoken by a single speaker only; in the second, training was based on syllables spoken by two speakers. In the recognition phases, all the syllables spoken by the speakers of the learning phases were correctly recognized. However, for those speakers whose syllables were not used in the learning phase, the percentage of correct recognition was 19.4% and 61.1% for the first and second experiment, respectively.