End-To-End Automated Speech Recognition on Numeric Sequences

A system comprising:

• data processing hardware;
• and
• memory hardware in communication with the data processing hardware, the memory hardware storing instructions that when executed on the data processing hardware cause the data processing hardware to perform operations comprising:
• receiving audio data for an utterance containing a numeric sequence;
• decoding, using a sequence-to-sequence speech recognition model, the audio data for the utterance to generate, as output from the sequence-to-sequence speech recognition model, an intermediate transcription of the utterance;
• processing, using a neural corrector/denormer, the intermediate transcription output from the sequence-to-sequence speech recognition model to generate a final transcription that represents the numeric sequence of the utterance in a written domain, the neural corrector/denormer trained on a set of training samples, each training sample comprising a speech recognition hypothesis for a corresponding training utterance and a ground-truth transcription of the corresponding training utterance, wherein the ground-truth transcription of the corresponding training utterance is in the written domain;
• and
• providing the final transcription representing the numeric sequence of the utterance in the written domain for output.