On the Effect of Purely Synthetic Training Data for Different Automatic Speech Recognition Architectures

Open in new window