Utterance classifier

摘要

Methods, systems and apparatus including computer programs encoded on a computer storage medium for classification using neural networks. One method is to receive audio data corresponding to an utterance, obtain a transcription of the utterance, generate a representation of the audio data, generate a representation of the transcription of the utterance, (i) audio data And (ii) providing to the classifier a representation of the transcription of the utterance, the classifier being based on the given representation of the audio data and the given representation of the transcription of the utterance. Including being trained to output an indication of whether the utterance associated with the given expression is likely or unlikely to be directed to the automated assistant.