Multi-pass speech recognition

摘要

According to example configurations, a speech recognition system is configured to receive an utterance. Based on analyzing at least a portion of the utterance using a first speech recognition model on a first pass, the speech recognition system detects that the utterance includes a first group of one or more spoken words. The speech recognition system utilizes the first group of one or more spoken words identified in the utterance as detected on the first pass to locate a given segment of interest in the utterance. The given segment can include one or more that are unrecognizable by the first speech recognition model. Based on analyzing the given segment using a second speech recognition model on a second pass, the speech recognition system detects one or more additional words in the utterance. A natural language understanding module utilizes the detected words to generate a command intended by the utterance.