Retraining voice model for trigger phrase using training data collected during usage

• US 10,885,899 B2
• Filed: 10/09/2018
• Issued: 01/05/2021
• Est. Priority Date: 10/09/2018
• Status: Active Grant

• Current Assignee: Motorola Mobility LLC (Lenovo Group Ltd.)
• Original Assignee: Motorola Mobility LLC (Lenovo Group Ltd.)
• Inventors: Iyer, Boby, Agrawal, Amit Kumar
• Primary Examiner(s): Han, Qi
• Application Number:
  US16/155,275
• Publication Number:
  US 20200111478A1
• Time in Patent Office: 819 Days
• Field of Search: 704243, 704244, 704231, 704250, 704251, 704255, 7042701, 704270
• CPC Class Codes: G10L 15/063 Training
• CPC Class Codes: G10L 15/22 Procedures used during a speech recognition process, e.g. man-machine dialogue
• CPC Class Codes: G10L 2015/0635 updating or merging of old and new templates; Mean values; Weighting
• CPC Class Codes: G10L 2015/223 Execution procedure of a spoken command
• CPC Class Codes: G10L 2015/225 Feedback of the input speech

Abstract

A method includes receiving initial training data associated with a trigger phrase in a device and training a voice model in the device using the initial training data. The voice model is used to identify a plurality of voice commands in the device initiated using the trigger phrase. Collection of additional training data from the plurality of voice commands and retraining of the voice model in the device are iteratively performed using the additional training data. A device includes a microphone and a processor to receive initial training data associated with a trigger phrase using the microphone, train a voice model device using the initial training data, use the voice model to identify a plurality of voice commands initiated using the trigger phrase, and iteratively collect additional training data from the plurality of voice commands and retrain the voice model in the device using the additional training data.

Claims

1. A method, comprising:

receiving initial training data associated with a trigger phrase in a device;

training a voice model in the device using the initial training data;

using the voice model to identify a plurality of voice commands in the device initiated using the trigger phrase;

iteratively collecting additional training data from the plurality of voice commands and retraining the voice model in the device using the additional training data; and

generating a user notification message on the device responsive to retraining the voice model.

2. The method of claim 1, further comprising generating the user notification on the device responsive to the retrained model having a quality metric greater than a first threshold.

3. The method of claim 1, further comprising generating the user notification on the device responsive to a quality metric associated with the retrained model increasing by a predetermined value.

4. The method of claim 3, wherein the quality metric is generated using one of a noise metric, an echo metric, an automated speech recognition metric, or a speaker verification metric.

5. The method of claim 1, wherein collecting the additional training data comprises collecting the additional training data for selected voice commands for which user input was received in the device subsequent to executing the selected voice command in the device.

6. The method of claim 1, wherein collecting the additional training data comprises collecting the additional training data for selected voice commands successfully executed in the device.

7. The method of claim 1, further comprising terminating the iterative collecting of additional training data and the retraining of the voice model responsive to a quality threshold associated with the additional training data exceeding a predetermined threshold.

8. A method, comprising:

receiving initial training data associated with a trigger phrase in a device;

training a voice model in the device using the initial training data;

using the voice model to identify a plurality of voice commands in the device initiated using the trigger phrase;

collecting additional training data from the plurality of voice commands in the device;

retraining the voice model in the device using the additional training data;

generating a user notification message on the device responsive to retraining the voice model; and

identifying subsequent voice commands in the device using the retrained voice model.

9. The method of claim 8, further comprising generating the user notification on the device responsive to the retrained model having a quality metric greater than a first threshold.

10. The method of claim 8, further comprising iterating the collecting and retraining in the device using the subsequent voice commands to further retrain the model.

11. The method of claim 10, further comprising generating the user notification on the device responsive to a quality metric associated with the further retrained model increasing by a predetermined value.

12. The method of claim 8, wherein collecting the additional training data comprises collecting the additional training data for selected voice commands for which user input was received in the device subsequent to executing the selected voice command in the device.

13. The method of claim 8, wherein collecting the additional training data comprises collecting the additional training data for selected voice commands successfully executed in the device.

14. A device, comprising:

a microphone; and

a processor to receive initial training data associated with a trigger phrase using the microphone, train a voice model device using the initial training data, use the voice model to identify a plurality of voice commands received using the microphone and initiated using the trigger phrase, iteratively collect additional training data from the plurality of voice commands, retrain the voice model in the device using the additional training data, and generate a user notification message on the device responsive to retraining the voice model.

15. The device of claim 14, wherein the processor is to generate the user notification on the device responsive to the retrained model having a quality metric greater than a first threshold.

16. The device of claim 14, wherein the processor is to generate the user notification on the device responsive to a quality metric associated with the retrained model increasing by a predetermined value.

17. The device of claim 16, wherein the quality metric is generated using one of a noise metric, an echo metric, an automated speech recognition metric, or a speaker verification metric.

18. The device of claim 14, wherein the processor is to collect the additional training data by collecting the additional training data for selected voice commands for which user input was received in the device subsequent to executing the selected voice command in the device.

19. The device of claim 14, wherein the processor is to collect the additional training data by collecting the additional training data for selected voice commands successfully executed in the device.

20. The device of claim 14, wherein the processor is to terminate the iterative collecting of additional training data and the retraining of the voice model responsive to a quality threshold associated with the additional training data exceeding a predetermined threshold.