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

Table 7

Results of all three-layer configurations of the proposed model in terms of WER, MER, and WIL.

Config	South Asian accent			Western accent
	WER (%)	MER (%)	WIL (%)	WER (%)	MER (%)	WIL (%)
A	35.35	29.68	30.98	10.36	12.65	11.22
B	20.419	18.54	19.35	9.68	8.69	9.85
C	18.0859	15.36	15.25	7.0	6.8	7.2