Training of Automatic Speech Recognition System on Noised Speech

In this paper, two techniques of automatic speech recognition system training on noised speech are compared with technique of training on clean speech. The comparing has been made by means of speech recognition accuracy measure, with usage of fourteen kinds of noise. These were noises of household appliances and computers, street and transport, teaching rooms and lobbies. The superiority degree of noised speech training techniques over the competitive technique has been assessed. It is shown that training on noised speech allows reaching the 95% recognition accuracy for minimal signal-to-noise ratio 10 dB, whereas training on clean speech allows reaching the same recognition accuracy for minimal signal-to-noise ratio 20 dB.