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

Figure 5

Training is done for 200 epochs. The cost or loss of training in different configurations is shown. (a) The trend of loss function for training and validation of the model in configuration A, (b) the trend of loss function for training and validation of model in configuration B, and (c) the minimum loss for training and validation of model in configuration C.

(a)

(b)

(c)