Granted Patent B2

US 12,512,098 · App. 17/820,660 · Granted Dec 30, 2025

Inventor: Chong Wang (Shenzhen, CN)

Assignee: SoundHound AI IP, LLC.

G10L15/22G06F16/9535G10L15/16G10L2015/225

Abstract

Claims (66)

1 . A data processing apparatus, comprising:

an interface configured to receive user-related information of a user from a plurality of channels; and

a processor coupled to the interface and configured to:

extract, via a speech recognition system, intention data from the user-related information from the plurality of channels;

generate a plurality of intention probes based on the intention data related to the user, each intention probe comprising a user task to be performed within a predetermined period of time and associated data items related to the user task;

receive, within the predetermined period of time, changes of the associated data items related to the user task via the user-related information from the plurality of channels;

continuously infer a plurality of importance scores of each intention probe of the plurality of intention probes by calculating a sum of the plurality of importance scores of associated data items for each intention probe related to the user task over the predetermined period of time;

determine an intention probe with a highest score based on the calculated sum for each intention probe;

provide, via an audio output interface, audio output associated with the determined intention probe to remind the user about the user task; and

terminate the intention probe at the expiration of the predetermined period of time.

2 . The apparatus of claim 1 , wherein the user-related information includes speech-based information, and the processor is configured to extract an abstract from a transcript of the speech-based information, and obtain the intention data based on the abstract.

3 . The apparatus of claim 1 , wherein the processor is configured to:

upon detection of a user task in the data related to aspects of the user, generate a corresponding intention probe corresponding to the user task; and

keep the corresponding intention probe active until the user task happens or the corresponding intention probe expires.

4 . The apparatus of claim 1 , wherein the processor is configured to:

adjust the plurality of importance scores dynamically; and

infer the importance of the intention probe based on the sum of scores of the associated data items of the intention probe.

5 . The apparatus of claim 1 , wherein the user task comprises one or more activities, the associated data items comprise timing and location information as the user input data to implement the one or more activities, and the processor is configured to:

change the importance of the intention probe by calculating a score of the timing and location information to implement the one or more activities in the intention probe.

6 . The apparatus of claim 1 , wherein the processor is configured to provide the audio output associated with the intention probe in response to a speech-based trigger of the user.

7 . The apparatus of claim 1 , wherein the processor is further configured to:

update the intention data over the predetermined period of time;

generate a plurality of updated intention probes based on the updated intention; and

provide information associated with an updated intention probe with a highest importance.

8 . A data processing method, comprising:

receiving user-related information of a user from a plurality of channels;

extracting, via a speech recognition system, intention data from the user-related information from the plurality of channels;

generating a plurality of intention probes based on the intention data related to the user, each intention probe comprising a user task to be performed within a predetermined period of time and associated data items related to the user task;

receiving, within the predetermined period of time, changes of the associated data items related to the user task via the user-related information from the plurality of channels;

continuously inferring a plurality of importance scores of each intention probe of the plurality of intention probes by calculating a sum of the plurality of importance scores of associated data items for each intention probe related to the user task over the predetermined period of time;

determining an intention probe with a highest importance score based on the calculated sum for each intention probe;

providing, via an audio output interface, audio output associated with the determined intention probe to remind the user about the user task; and

terminating the intention probe at the expiration of the predetermined period of time.

9 . The method of claim 8 , wherein the user-related information-includes speech-based information, and the method comprises extracting an abstract from a transcript of the speech-based information, and obtaining the intention data based on the abstract.

10 . The method of claim 8 , further comprising:

upon detection of a user task in the data related to aspects of the user, generating an intention probe corresponding to the user task; and

keeping the corresponding intention probe active until the-user task happens or the intention probe expires.

11 . The method of claim 8 , further comprising: adjusting the plurality of importance scores dynamically; and inferring the importance of the intention probe based on the sum of scores of the associated data items of the intention probe.

12 . The method of claim 8 , wherein the user task comprises one or more activities, the associated data items comprise timing and location information as the user input data to implement the one or more activities, and the method further comprises:

changing the importance of the intention probe by calculating a score of the timing and location information to implement the one or more activities in the intention probe.

13 . The method of claim 8 , further comprising providing the audio output associated with the intention probe in response to a speech-based trigger of the user.

14 . The method of claim 8 , further comprising:

updating the intention data over the predetermined period of time;

generating a plurality of updated intention probes based on the updated intention; and

providing information associated with an updated intention probe with a highest importance.

15 . A non-transitory computer readable medium storing instructions that, if executed by one or more processor, would cause an apparatus to perform a method comprising:

receiving user-related information of a user from a plurality of channels;

extracting, via a speech recognition system, intention data from the user-related information from the plurality of channels;

generating a plurality of intention probes based on the intention data related to the user, each intention probe comprising a user task to be performed within a predetermined period of time and associated data items related to the user task;

receiving, within the predetermined period of time, changes of the associated data items relate to the user task via the user-related information from the plurality of channels;

continuously inferring a plurality of importance scores of each intention probe of the plurality of intention probes by calculating a sum of the plurality of importance scores of associated data items for each intention probe relate to the user task over the predetermined period of time;

determining an intention probe with a highest importance score based on the calculated sum for each intention probe;

providing, via an audio output interface, audio output associated with the determined intention probe to remind the user about the user task; and

terminating the intention probe at the expiration of the predetermined period of time.

16 . The non-transitory computer readable medium of claim 15 , wherein the user-related information includes speech-based information, and the method comprises extracting an abstract from a transcript of the speech-based information, and obtaining the intention data based on the abstract.

17 . The non-transitory computer readable medium of claim 15 , comprising:

upon detection of an user task in the data related to aspects of the user, generating a corresponding intention probe corresponding to the user task; and

keeping the corresponding intention probe active until the user task happens or the corresponding intention probe expires.

18 . The non-transitory computer readable medium of claim 15 , further comprising: adjusting the plurality of importance scores dynamically; and inferring the importance of the intention probe based on the sum of scores of the associated data items of the intention probe.

19 . The non-transitory computer readable medium of claim 15 , wherein, the user task comprises one or more activities, the associated data items comprise timing and location information as the user input data to implement the one or more activities, and the method further comprises:

changing the importance of the intention probe by calculating a score of the timing and location information to implement the one or more activities in the intention probe.

20 . The non-transitory computer readable medium of claim 15 , further comprising providing the audio output associated with the intention probe in response to a speech-based trigger of the user.

21 . The non-transitory computer readable medium of claim 15 , further comprising:

updating the intention data over the predetermined period of time;

generating a plurality of updated intention probes based on the updated intention; and

providing information associated with an updated intention probe with a highest importance.

Assignments (5)

Priority Claims (1)

Continuity (1)

References Cited (11)