Enhancing Performance of End-to-End Gujarati Language ASR using combination of Integrated Feature Extraction and Improved Spell Corrector Algorithm

A number of intricate deep learning architectures for effective End-to-End (E2E) speech recognition systems have emerged due to recent advancements in algorithms and technical resources. The proposed work develops an ASR system the publicly accessible dataset on Gujarati language. approach provided this research combines features like Mel frequency Cepstral Coefficients (MFCC) Constant Q (CQCC) at front-end feature extraction methodologies. Enhanced spell corrector with BERT-based algorithm Gated Recurrent Units (GRU) based DeepSpeech2 architecture are used implement back end portion system. study shown that combining MFCC CQCC extracted from GRU-based model upgraded or enhanced improves Word Error Rate (WER) by 17.46% when compared without post processing.