Google researchers boost speech recognition accuracy with more datasets

Google researchers propose a simple solution for improving speech recognition accuracy: training one large model on many datasets.