CN-122024739-A - Main recognition method and related device of electronic equipment

Abstract

The invention provides a main recognition method and a related device of electronic equipment, wherein the method comprises the following steps of acquiring a voice sampling data set acquired by the electronic equipment in a feeding mode in a preset statistical period; the method comprises the steps of screening each voice fragment contained in a voice sampling data set to obtain a plurality of target voice fragments related to feeding actions, extracting voiceprint characteristics of each target voice fragment to obtain a plurality of first voiceprint characteristics, determining a plurality of first target voiceprint characteristics which have long-term feeding relation with electronic equipment based on the occurrence times of each first voiceprint characteristic, and storing each first target voiceprint characteristic into a master voiceprint library. The invention deeply simulates the main recognition process of the real pet, improves the simulation degree of main recognition of the electronic equipment, and has higher intelligent degree.

Inventors

• CHENG BING
• ZOU BO

Assignees

• 深圳市噜咔博士科技有限公司

Dates

Publication Date

20260512

Application Date

20251223

Claims (10)

1. 1. A method of authenticating an electronic device, the method comprising the steps of: acquiring a voice sampling data set acquired by electronic equipment in a feeding mode in a preset statistical period; Screening each voice segment contained in the voice sampling data set to obtain a plurality of target voice segments related to feeding actions; extracting voiceprint features of each target voice segment to obtain a plurality of first voiceprint features; Determining a plurality of first target voiceprint features with long-term feeding relationship with the electronic equipment in the preset statistical period based on the occurrence times of the first voiceprint features; And storing each first target voiceprint feature into a master voiceprint library so as to set the working mode of the electronic equipment as a master mode after the first target voiceprint feature is subsequently detected.
2. 2. The method of claim 1, wherein after said storing each of said first target voiceprint features in a master voiceprint library, the method further comprises: acquiring a plurality of second voice characteristics acquired by the electronic equipment in a feeding mode in a current statistical period and the occurrence times of the second voice characteristics, wherein the current statistical period is later than the preset statistical period; determining at least one second target voiceprint feature having a long-term feeding relationship with the electronic device within the current statistical period based on the number of occurrences of each second voiceprint feature; and for each first target voiceprint feature in the master voiceprint library, if a second target voiceprint feature matched with the first target voiceprint feature exists, updating the first target voiceprint feature by using the second target voiceprint feature.
3. 3. The method of claim 2, wherein the method further comprises: For each of the first target voiceprint features in the master voiceprint library, if there is no second target voiceprint feature matching the first target voiceprint feature, removing the first target voiceprint feature from the master voiceprint library. And for each second target voiceprint feature, if the matched first target voiceprint feature does not exist in the master voiceprint library, storing the second target voiceprint feature into the master voiceprint library.
4. 4. A method according to any one of claims 1 to 3, wherein determining a number of first target voiceprint features having a long-term feeding relationship with the electronic device within the predetermined statistical period based on the number of occurrences of each of the first voiceprint features comprises: clustering the first voiceprint features to obtain an initial clustering result; Classifying each first voiceprint feature in the initial clustering result to obtain a category of each first voiceprint feature, wherein the category of each first voiceprint feature is a core point, a boundary point or a noise point; Removing first voiceprint features belonging to noise points from the initial clustering result, and determining a plurality of clustering clusters according to each core point and boundary point in the initial clustering result; Selecting cluster clusters with a large number of first voiceprint features as target cluster clusters, and respectively taking the voiceprint features obtained by fusing the first voiceprint features contained in each target cluster as the first target voiceprint features.
5. 5. A method according to any one of claims 1 to3, wherein said performing voiceprint feature extraction on each of said target speech segments to obtain a plurality of first voiceprint features comprises: Preprocessing each target voice segment to obtain a plurality of voice sub-segments corresponding to each target voice segment; extracting primary voiceprint features of each voice sub-segment to obtain initial voiceprint features which correspond to each voice sub-segment and contain energy features; Inputting each initial voiceprint feature corresponding to each target voice segment into a preset time sequence depth neural model for each target voice segment to obtain advanced voiceprint features of the target voice segment; and carrying out preset optimization treatment on each advanced voiceprint feature to obtain each first voiceprint feature.
6. 6. A method according to any one of claims 1 to 3, wherein said screening each speech segment contained in said speech sample dataset to obtain a number of target speech segments associated with a feeding action comprises: Feeding keyword extraction is carried out on each voice fragment contained in the voice sampling data set; and taking each voice segment containing the feeding keywords as each target voice segment.
7. 7. A method according to any one of claims 1 to 3, wherein the method further comprises: Receiving current dialogue data input by a user; Extracting voiceprint features from the current dialogue data to obtain current voiceprint features; Matching the current voiceprint features with a plurality of first target voiceprint features contained in the master voiceprint library; Detecting that a first target voiceprint feature matched with the current voiceprint feature exists in the master voiceprint library, and setting the working mode of the electronic equipment as a master mode.
8. 8. A recognition apparatus of an electronic device, characterized in that the recognition apparatus of an electronic device comprises: The feeding data acquisition module is used for acquiring a voice sampling data set acquired by the electronic equipment in a feeding mode in a preset statistical period; The voice segment screening module is used for screening each voice segment contained in the voice sampling data set to obtain a plurality of target voice segments related to the feeding action; The voiceprint feature extraction module is used for extracting voiceprint features of each target voice segment to obtain a plurality of first voiceprint features; the voiceprint feature screening module is used for determining a plurality of first target voiceprint features with long-term feeding relation with the electronic equipment in the preset statistical period based on the occurrence times of the first voiceprint features; and the data management module is used for storing each first target voiceprint feature into a master voiceprint library so as to set the working mode of the electronic equipment as a master mode after the first target voiceprint feature is subsequently detected.
9. 9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps in the method of recognizing an electronic device according to any one of claims 1 to 7 when the computer program is executed by the processor.
10. 10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon a computer program which, when executed by a processor, implements the steps in the authentication method of an electronic device according to any of claims 1 to 7.

Description

Main recognition method and related device of electronic equipment Technical Field The application relates to the technical field of artificial intelligence, in particular to a main recognition method and a related device of electronic equipment. Background At present, the setting of the host mode of the intelligent pet toy is mostly finished based on physical keys or one-key setting functions in mobile phone application programs. The real pet generally carries out the recognition based on the caretaking behaviors of the daily user for feeding the pet and the like, and the non-explicit recognition function setting of the host mode of the intelligent pet toy is not in line with the actual situation of the linear cultivation recognition in the real pet health scene, and the host set based on a physical key or a one-key in a mobile phone application program is possibly inconsistent with the actual host of the intelligent pet toy, so that the intelligent pet toy and the actual host cannot communicate according to the host mode, the simulation degree of the intelligent pet toy is insufficient, the intelligent degree is low, and the use requirement of the user is difficult to meet. Disclosure of Invention In view of this, the application provides a method for recognizing main of electronic equipment, which screens a voice sampling data set collected by the electronic equipment in a feeding mode to obtain a plurality of target voice fragments related to feeding action, so as to reduce interference of background boring voice collected in the feeding mode to recognizing main of the electronic equipment, and add voiceprint features extracted from each target voice fragment into a main voiceprint library, so that users with long-term feeding relationship with the electronic equipment can interact with the electronic equipment in the main mode, the process deeply simulates a main recognizing process of a real pet, improves the reality of recognizing main of the electronic equipment, can strengthen interaction between the electronic equipment and actual feeding users, and ensures that the intelligent degree of the electronic equipment is higher. In a first aspect, the present application provides a method for recognizing an electronic device, the method comprising the steps of: acquiring a voice sampling data set acquired by electronic equipment in a feeding mode in a preset statistical period; Screening each voice segment contained in the voice sampling data set to obtain a plurality of target voice segments related to feeding actions; extracting voiceprint features of each target voice segment to obtain a plurality of first voiceprint features; Determining a plurality of first target voiceprint features with long-term feeding relationship with the electronic equipment in the preset statistical period based on the occurrence times of the first voiceprint features; And storing each first target voiceprint feature into a master voiceprint library so as to set the working mode of the electronic equipment as a master mode after the first target voiceprint feature is subsequently detected. Optionally, after the storing each of the first target voiceprint features in the master voiceprint library, the method further includes: acquiring a plurality of second voice characteristics acquired by the electronic equipment in a feeding mode in a current statistical period and the occurrence times of the second voice characteristics, wherein the current statistical period is later than the preset statistical period; determining at least one second target voiceprint feature having a long-term feeding relationship with the electronic device within the current statistical period based on the number of occurrences of each second voiceprint feature; and for each first target voiceprint feature in the master voiceprint library, if a second target voiceprint feature matched with the first target voiceprint feature exists, updating the first target voiceprint feature by using the second target voiceprint feature. Optionally, the method further comprises: For each of the first target voiceprint features in the master voiceprint library, if there is no second target voiceprint feature matching the first target voiceprint feature, removing the first target voiceprint feature from the master voiceprint library. And for each second target voiceprint feature, if the matched first target voiceprint feature does not exist in the master voiceprint library, storing the second target voiceprint feature into the master voiceprint library. Optionally, the determining, based on the number of occurrences of each first voiceprint feature, a number of first target voiceprint features having a long-term feeding relationship with the electronic device in the preset statistical period includes: clustering the first voiceprint features to obtain an initial clustering result; Classifying each first voiceprint feature in the initial clustering result to obtain a c