Users interact with virtual assistants by issuing voice queries. Prior to processing, the user’s voice query is converted to text using automatic speech recognition (ASR). For a variety of reasons, ASR can result in errors such that the converted text query does not match the original voice query at least partially. Query misrecognition disrupts the flow of interaction between the user and the virtual assistant, requiring users to reissue the voice query multiple times or enter it using an alternate input mechanism. This disclosure describes techniques to predict misrecognized voice queries and avoid the same misrecognition on the subsequent query attempt.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Weisz, Ágoston, "Automatic Contextual Adjustment to Speech Recognition to Reduce Query Misrecognition", Technical Disclosure Commons, (January 29, 2021)