Detecting and suppressing voice queries
A computing system receives requests from client devices to process voice queries that have been detected in local environments of the client devices. The system identifies that a value that is based on a number of requests to process voice queries received by the system during a specified time interval satisfies one or more criteria. In response, the system triggers analysis of at least some of the requests received during the specified time interval to trigger analysis of at least some received requests to determine a set of requests that each identify a common voice query. The system can generate an electronic fingerprint that indicates a distinctive model of the common voice query. The fingerprint can then be used to detect an illegitimate voice query identified in a request from a client device at a later time.
1. A client device-implemented method, comprising:
detecting, by a client device, an utterance of a hotword in a local environment of the client device;
receiving, by the client device and in response to the detecting, a voice query including one or more words following the hotword;
generating, by the client device, an electronic fingerprint for the received voice query;
determining, by the client device, whether the received voice query is blacklisted based on a comparison of the electronic fingerprint of the received voice query with electronic fingerprints of blacklisted voice queries stored in a local database of the client device, wherein each of the electronic fingerprints of the blacklisted voice queries corresponds to a respective common voice query occurring across multiple voice query requests, and each of the electronic fingerprints of the blacklisted voice queries is determined based on a volume of voice query processing requests received over a time interval; and
suppressing, by the client device, performance of an operation indicated by the received voice query when the received voice query is determined to be blacklisted.
2. The method of claim 1 , further comprising:
performing, by the client device, the operation indicated by the received voice query when the received voice query is determined not to be blacklisted.
3. The method of claim 1 , wherein the suppressing includes, when the operation indicated by the received voice query was not performed, blocking, by the client device, the operation from being performed.
4. The method of claim 1 , wherein the suppressing includes, when the operation indicated by the received voice query was performed, reversing, by the client device, the operation.
5. The method of claim 1 , further comprising:
storing, by the client device, first information data associated with the received voice query and second information data associated with the suppressed operation associated with the received voice query into a suppression log; and
transmitting, by the client device, the first information and the second information in the suppression log to a server system including one or more computers.
6. The method of claim 1 , wherein the received voice query is one of a plurality of voice queries received by the client device, and the method further comprises:
determining, by the client device and for each of the plurality of voice queries, whether the respective voice query corresponds to one of the blacklisted voice queries stored in the local database.
7. The method of claim 1 , wherein the received voice query is one of a plurality of voice queries received by the client device, and the method further comprises:
selecting, by the client device, a subset of the plurality of voice queries; and
determining, by the client device and for each of the subset of the plurality of voice queries, whether the respective voice query corresponds to one of the blacklisted voice queries stored in the local database.
8. The method of claim 1 , further comprising:
receiving, by the client device, the electronic fingerprints of the blacklisted voice queries from a server system including one or more computers; and
storing, by the client device, the electronic fingerprints of the blacklisted voice queries into the local database.
9. The method of claim 1 , further comprising:
receiving, by the client device, additional electronic fingerprints to add the local database from a server system including one or more computers.
10. The method of claim 1 , further comprising:
transmitting, by the client device, audio data of the received voice query and a voice query processing request corresponding to the received voice query to a server system including one or more computers.
11. A non-transitory computer-readable storage medium including computer executable instructions, wherein the instructions, when executed by a computer, cause the computer to perform a method, the method comprising:
detecting an utterance of a hotword in a local environment of a client device;
receiving, in response to the detecting, a voice query including one or more words following the hotword;
generating an electronic fingerprint for the received voice query;
determining whether the received voice query is blacklisted based on a comparison of the electronic fingerprint of the received voice query with electronic fingerprints of blacklisted voice queries stored in a local database of the client device, wherein each of the electronic fingerprints of the blacklisted voice queries corresponds to a respective common voice query occurring across multiple voice query requests, and each of the electronic fingerprints of the blacklisted voice queries is determined based on a volume of voice query processing requests received over a time interval; and
suppressing performance of an operation indicated by the received voice query when the received voice query is determined to be blacklisted.
12. The non-transitory computer-readable storage medium of claim 11 , further comprising performing the operation indicated by the received voice query when the received voice query is determined not to be blacklisted.
13. The non-transitory computer-readable storage medium of claim 11 , wherein the suppressing includes blocking the operation from being performed when the operation indicated by the received voice query was not performed.
14. The non-transitory computer-readable storage medium of claim 11 , wherein the suppressing includes reversing the operation when the operation indicated by the received voice query was performed.
15. The non-transitory computer-readable storage medium of claim 11 , further comprising:
storing first information data associated with the received voice query and second information data associated with the suppressed operation associated with the received voice query into a suppression log; and
transmitting the first information and the second information in the suppression log to a server system including one or more computers.
16. The non-transitory computer-readable storage medium of claim 11 , wherein the received voice query is one of a plurality of voice queries received by the client device, and further comprising:
determining, for each of the plurality of voice queries, whether the respective voice query corresponds to one of the blacklisted voice queries stored in the local database.
17. The non-transitory computer-readable storage medium of claim 11 , wherein the received voice query is one of a plurality of voice queries received by the client device, and further comprising:
selecting a subset of the plurality of voice queries; and
determining, for each of the subset of the plurality of voice queries, whether the respective voice query corresponds to one of the blacklisted voice queries stored in the local database.
18. The non-transitory computer-readable storage medium of claim 11 , further comprising:
receiving the electronic fingerprints of the blacklisted voice queries from a server system including one or more computers; and
storing the electronic fingerprints of the blacklisted voice queries into the local database.
19. The non-transitory computer-readable storage medium of claim 11 , further comprising:
receiving additional electronic fingerprints to add the local database from a server system including one or more computers.
20. An apparatus, comprising:
processing circuitry configured to
detect an utterance of a hotword in a local environment of the apparatus,
receive, in response to detecting, a voice query including one or more words following the hotword,
generate an electronic fingerprint for the received voice query,
determine whether the received voice query is blacklisted based on a comparison of the electronic fingerprint of the received voice query with electronic fingerprints of blacklisted voice queries stored in a local database of the apparatus, wherein each of the electronic fingerprints of the blacklisted voice queries corresponds to a respective common voice query occurring across multiple voice query requests, and each of the electronic fingerprints of the blacklisted voice queries is determined based on a volume of voice query processing requests received over a time interval, and
suppress performance of an operation indicated by the received voice query when the received voice query is determined to be blacklisted.