Improvement in Automatic Speech Recognition of South Asian Accent Using Transfer Learning of DeepSpeech2

Table 1

Literature survey of previous methodologies.

Year	Paper	Methodology	Language	Accent	WER
2021	[45]	Unsupervised CNN	English	—	12.83
2021	[46]	RNN encoder-decoder + CTC	English	—	8.3
2021	[47]	Quantifying bias	Dutch	—	16.85
2022	[48]	Transformers	English	—	39.9
2022	[49]	Semisupervised learning	English	—	9.4
2022	Proposed	Multi pipeline feature extractor learning	English	Native + South Asian	7