×

Speaker identification and unsupervised speaker adaptation techniques

  • US 10,438,595 B2
  • Filed: 10/09/2018
  • Issued: 10/08/2019
  • Est. Priority Date: 09/30/2014
  • Status: Active Grant
First Claim
Patent Images

1. A non-transitory computer-readable storage medium storing one or more programs, the one or more programs comprising instructions for operating a virtual assistant, which when executed by one or more processors of an electronic device, cause the device to:

  • receive current user speech for activating the virtual assistant, wherein the current user speech is associated with current contextual data;

    select, based on the current contextual data, a first set of stored voiceprints from a plurality of sets of stored voiceprints in a speaker profile of the device, wherein the first set of stored voiceprints is annotated to indicate first contextual data;

    determine whether a current voiceprint derived from the current user speech matches the first set of stored voiceprints within a predetermined threshold; and

    in accordance with a determination that the current voiceprint matches the first set of stored voiceprints within the predetermined threshold;

    add the current voiceprint to the first set of stored voiceprints in the speaker profile;

    annotate the current voiceprint to indicate the first contextual data; and

    activate the virtual assistant to process a spoken command received subsequent to the user speech.

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