×

Speech and noise models for speech recognition

  • US 8,249,868 B2
  • Filed: 09/30/2011
  • Issued: 08/21/2012
  • Est. Priority Date: 06/14/2010
  • Status: Active Grant
First Claim
Patent Images

1. A system comprising:

  • one or more processing devices; and

    one or more storage devices storing instructions that, when executed by the one or more processing devices, cause the one or more processing devices to;

    receive a first audio signal generated by a device based on audio input from a user, the first audio signal including at least a first user audio portion that corresponds to both first background audio and one or more first user utterances recorded by the device;

    access a user speech model associated with the user;

    determine that the first background audio in the first user audio portion is below a defined threshold;

    in response to determining that the first background audio in the first user audio portion is below the defined threshold, adapt the accessed user speech model based on the first audio signal to generate an adapted user speech model that models speech characteristics of the user;

    receive a second audio signal generated by the device based on second audio input from a user, the second audio signal including at least a second user audio portion that corresponds to both second background audio and one or more second user utterances recorded by the device;

    determine that the second background audio in the second user audio portion is not below the defined threshold;

    in response to determining that the second background audio in the second user audio portion is not below the defined threshold, not adapt the accessed user speech model based on the second audio signal; and

    perform noise compensation on a third audio signal using the adapted user speech model to generate a filtered audio signal with reduced background audio compared to the third audio signal.

View all claims
  • 2 Assignments
Timeline View
Assignment View
    ×
    ×