Speech Recognition and Natural Language Processing

Summary

Traditional machine interfaces such as keyboards, tablets and joysticks have long been the weak point in fully integrating computers into our everyday lives. Speech recognition and speech response systems were too limited, had high error rates and were offensive to listen to because they sounded like mechanical speech.

Now that is all changing. Major technology breakthroughs on both the input (speech recognition) and output (speech synthesis) sides have made interactive speech suitable for everyday use. Interestingly enough, these breakthroughs involve neural networks as part of their input and output functions. In addition, neural networks can also be the source of the content that is part of a response.