Key points are not available for this paper at this time.
A phoneme represents the smallest distinct sound unit within a language, unique to each linguistic system. Phoneme identification finds applications in various speech-related technologies, including automatic speech recognition, language learning, assistive technology for speech impaired individuals. The realm of Automatic Speech Recognition (ASR) stands as a formidable challenge within the field, with a predominant emphasis on extensively spoken languages. However, the intricacies posed by Hindi speech recognition have received limited attention, owing to the language's complex structure and the scarcity of pertinent data. Existing models designed for languages like English cannot be seamlessly adapted to the nuances of Hindi. This study presents an approach to phoneme-based speech recognition in the Hindi language through the implementation of a deep learning model. The proposed model leverages techniques in ASR to transform raw audio data into phoneme sequences, assisting in recognizing digits from a dataset collected in limited capacity consisting of counting in Hindi language.
Singh et al. (Fri,) studied this question.