Earphone self-recognizing noise method and system

Abstract

The application relates to a headset self-recognizing noise method and system, wherein the recognizing method comprises creating a filtering whitelist, the data of the filtering whitelist comprising at least one standard sound information; acquiring an environment sound and extracting information of at least one group of sub-sounds in the environment sound, the information of different sub-sounds being emitted by different sound sources; extracting feature data of the sub-sounds, matching the feature data with the data of the filtering whitelist, determining a recognizing type of the sub-sounds according to a matching result, and sending effective sound to a user according to the recognizing type. The method compares the voiceprint data and the speech data of the target sound with the data of the filtering whitelist under preset recognizing rules, improves the accuracy of judging whether the sound is noise, makes the recognizing result more meet the needs of the user, ensures the song listening experience of the user, and helps the user better acquire effective information from the outside world.

Description

Technical Field

[0001] This invention relates to the field of data recognition technology, and in particular to a method for headphones to self-identify noise. Background Technology

[0002] The popularity of digital music products has given rise to a large group of "headphone addicts" who are inseparable from their music. They often listen to music while cycling or taking public transportation, isolating themselves from the external environment. These "headphone addicts" often cannot hear external cues, such as car sounds or alarms, thus facing significant safety hazards and inconvenience.

[0003] To address this issue, existing headphones can activate a "transparency mode" to capture ambient sound. In "transparency mode," ambient sound is picked up by the microphone on the headphones, filtered, and then played back by the headphone speakers, allowing users to hear external sounds while wearing headphones.

[0004] However, the existing "transparency mode" cannot distinguish which external sounds are valid and which are noise. Therefore, it amplifies the overall ambient sound and the noise in the ambient sound as well, which affects the user's listening experience and the acquisition of valid external sounds. Summary of the Invention

[0005] The purpose of this invention is to provide a method for headphones to automatically identify noise, which can solve the problem in the prior art that headphones cannot accurately identify which are valid sounds and noise in transparency mode, thus affecting the user's listening experience and the acquisition of valid sounds from the outside world.

[0006] The above-mentioned objective of this invention is achieved through the following technical solution: A method for headphones to automatically identify noise includes the following steps: Create a filtering whitelist, the data of which includes at least one standard voice information, the standard voice information including voiceprint data information and voice data information; Acquire ambient sound and extract information of at least one set of sub-sounds from the ambient sound, wherein the information of the different sub-sounds is emitted by different sound sources; Extract feature data from the sub-voice, the feature data including target voiceprint data and target speech data; The feature data is matched with the data in the filtering whitelist, and the recognition type of the sub-sound is determined based on the matching result. The recognition type includes valid sound, noise, and undetermined sound. The valid audio is sent to the user based on the identified type.

[0007] By adopting the above technical solution, a filtering whitelist is first set up, and sounds that meet the filtering whitelist criteria are considered valid sounds. After acquiring ambient sound, sounds from different sound sources are extracted and identified separately. Identification is performed from two dimensions: voiceprint data and speech data. Voiceprint data corresponds to the sound source itself, and speech data corresponds to the information content included in the sound. Under preset identification rules, by comparing the voiceprint data and speech data of the sound to be identified with the data in the filtering whitelist, the system can more human-centeredly identify whether a sound is valid, making the identification results more in line with user needs. Finally, based on the identification type, the valid sound is delivered to the user, allowing the user to obtain the information they want. This solution improves upon the problem in existing technologies where headphones in transparency mode cannot accurately identify which sounds are valid and which are noise, thus affecting the user's listening experience and the acquisition of valid external sounds.

[0008] In a preferred embodiment, the present invention can be further configured such that creating the filtering whitelist includes: Acquire the target sound and use the target sound as the standard sound; Extract standard speech data and standard voiceprint data from the target sound; Update the standard voice data and standard voiceprint data to the filter whitelist.

[0009] By adopting the above technical solution, users can obtain effective sounds according to their own needs, such as alarm sounds or the sound of others calling their names. The more target sounds are obtained, the richer the content of the whitelist becomes, thereby covering more sound judgment scenarios and making the headphones more capable of self-identifying noise.

[0010] In a preferred embodiment, the present invention can be further configured to: match the feature data with the data of the filtering whitelist, and determine the recognition type of the sub-voice based on the matching result, including: Determine whether the target speech data matches the standard speech data; If so, then the sub-sound is identified as a valid sound; If not, determine whether the target voiceprint information matches the standard voiceprint data. If they match, identify the sub-voice as a pending sound; otherwise, identify the sub-voice as noise.

[0011] By employing the above technical solution, prioritizing the matching of target speech data with standard speech data helps users quickly obtain the information they want to hear. If the target speech data does not match the standard speech data, it indicates that the target speech data is not the content the user wants to hear. Furthermore, by determining whether the voiceprint information matches the standard voiceprint information, if they match, it means that the sound source is the object of the user's interest and is therefore classified as a pending sound, allowing the user to decide whether to obtain the sound. When it is identified as noise, it means that the sound does not contain any information the user wants to know.

[0012] In a preferred embodiment, the present invention can be further configured such that determining whether the target speech data matches the standard speech data includes: The target speech data is converted into first text data, and the standard speech data is converted into second text data; Semantic analysis is performed on the first text data and the second text data, and the target speech data is determined to match the standard speech data by judging whether the semantic analysis results are consistent.

[0013] By adopting the above technical solutions, users can obtain information that is semantically similar to standard speech data, making the recognition effect more human-like.

[0014] In a preferred embodiment, the present invention can be further configured such that sending the valid tone to the user according to the recognition type includes: Amplify the volume of the sub-sound and send it to the user.

[0015] By adopting the above technical solutions, users can hear the information they want.

[0016] In a preferred embodiment, the present invention may be further configured such that sending the valid tone to the user according to the recognition type also includes: When the sub-sound is identified as noise, the sub-sound is filtered out.

[0017] By adopting the above technical solution, fewer distracting sounds are presented to the user.

[0018] In a preferred embodiment, the present invention may be further configured such that sending the valid tone to the user according to the recognition type also includes: When the sub-sound is identified as an undetermined tone, a prompt is made indicating the existence of the sub-sound, and a determination instruction is received. The determination instruction is used to change the sub-sound to a valid tone or noise.

[0019] By adopting the above technical solution, users can decide how to process the pitch, which improves interactivity and user experience.

[0020] In a preferred embodiment, the present invention may be further configured such that the method further includes: Receive a priority setting instruction, and assign a priority to different standard sound tags according to the priority setting instruction.

[0021] By adopting the above technical solution, when multiple valid tones are identified, the playback priority of the multiple valid tones is determined according to the priority of the standard sound matched by the valid tones, thereby helping users to obtain the most desired information more quickly among multiple valid tones.

[0022] In a preferred embodiment, the present invention may be further configured such that sending the valid tone to the user according to the recognition type also includes: When at least two valid tones are identified, the two valid tones are played sequentially according to the priority.

[0023] By adopting the above technical solution and adjusting the playback order of effective sounds, users can more quickly obtain the information they want from multiple effective sounds.

[0024] The second objective of this invention is to provide a system for headphones to automatically identify noise, which can solve the problem in the prior art that headphones cannot accurately identify which sounds are valid and which are noise in transparency mode.

[0025] The second objective of this invention is achieved through the following technical solution: A system for self-identifying noise in headphones, comprising: The creation module is used to create a filter whitelist. The data in the filter whitelist includes information on at least one standard voice, which includes voiceprint data and voice data. The acquisition module is used to acquire ambient sound and extract at least one sub-sound from the ambient sound, with different sub-sounds emitted by different sound sources; The extraction module is used to extract feature data of the sub-voice, which includes target voiceprint data and target speech data; The recognition module is used to match feature data with data in the filter whitelist and determine the recognition type of the sub-sound based on the matching results. The recognition types include valid sound, noise, and undetermined sound. The sound processing module is used to send valid sounds to the user based on the recognition type.

[0026] By adopting the above technical solution, the creation module creates a filtering whitelist and sends the data of the filtering whitelist to the recognition module; the acquisition module acquires ambient sound and extracts at least one sub-sound from the ambient sound, and sends the sub-sound to the extraction module; the extraction module extracts the feature data of the sub-sound and sends the feature data to the recognition module; the recognition module matches the feature data with the data of the filtering whitelist and determines the recognition type of the sub-sound based on the matching result.

[0027] In a preferred embodiment, the present invention can be further configured such that the creation module includes a data acquisition unit, a processing unit, and an update unit, wherein: The acquisition unit acquires the target sound and uses it as the standard sound. The processing unit is used to extract standard speech data and standard voiceprint data from the target sound; The update unit is used to update the standard voice data and standard voiceprint data to the filter whitelist.

[0028] In a preferred embodiment, the present invention can be further configured such that the identification module includes a first judgment unit, a first identification unit, a second judgment unit, a second identification unit, and a third identification unit, wherein: The first judgment unit is used to determine whether the target speech data matches the standard speech data; The first recognition unit is used to recognize the sub-voice as a valid voice when the first judgment unit determines that the target speech data matches the standard speech data. The second judgment unit is used to determine whether the target voiceprint information matches the standard voiceprint information when the first judgment unit determines that the target voice data does not match the standard voice data. The second identification unit is used to identify the sub-sound as noise when the second judgment unit determines that the target voiceprint information does not match the standard voiceprint information. The third recognition unit is used to recognize the sub-voice as an undetermined sound when the second judgment unit determines that the target voiceprint information matches the standard voiceprint information.

[0029] In a preferred embodiment, the present invention can be further configured such that the first determining unit includes: The conversion subunit is used to convert the target speech data into first text data and the standard speech data into second text data; The analysis subunit is used to perform semantic analysis on the first text data and the second text data, and to determine whether the target speech data matches the standard speech data by judging whether the semantic analysis results are consistent.

[0030] In a preferred embodiment, the present invention can be further configured such that the sound processing module includes an amplification / transmission unit. The amplification / transmission unit is used to amplify the volume of the sub-sound and send it to the user when the sub-sound is identified as a valid tone.

[0031] In a preferred embodiment, the present invention may be further configured such that the sound processing module also includes a filtering unit. The filtering unit is used to filter out sub-sounds when they are identified as noise.

[0032] In a preferred embodiment, the present invention may be further configured such that the sound processing module further includes a determination unit. The determination unit is used to indicate the existence of a sub-sound when it is identified as a pending tone, and to receive a determination instruction, which is used to change the sub-sound to a valid tone or noise.

[0033] In a preferred embodiment, the present invention can be further configured such that the system also includes a setting module: The setting module is used to receive priority setting instructions and assign priorities to different standard sounds according to the priority setting instructions.

[0034] In a preferred embodiment, the present invention may be further configured such that the sound processing module also includes a playback unit. The playback unit is used to play the two valid tones sequentially according to their priority when at least two valid tones are identified.

[0035] In summary, the present invention has at least one of the following beneficial technical effects: 1. By first setting a whitelist of filters, sounds that meet the whitelist criteria are considered valid sounds. After acquiring ambient sound, sounds from different sources are extracted and identified separately. Recognition is performed from two dimensions: voiceprint data and speech data. Under preset recognition rules, by comparing the voiceprint and speech data of the target sound with the data in the whitelist, the sound judgment can be more accurate, thus making the recognition results more in line with user needs and solving the problem in existing headphones that cannot accurately identify valid sounds and noise in transparency mode.

[0036] 2. Users can acquire what they consider to be valid sounds through the microphone, such as alarm sounds or the sound of someone calling their name. The more target sounds acquired, the richer the content of the whitelist becomes, thus covering more sound judgment scenarios and making the headphones more capable of self-identifying noise. Attached Figure Description

[0037] Figure 1 This is a flowchart illustrating a method for headphone self-identification of noise according to one embodiment of the present invention.

[0038] Figure 2This is another schematic flowchart of a method for headphone self-identification of noise according to one embodiment of the present invention.

[0039] Figure 3 yes Figure 2 The flowchart of step S207, determining whether the target speech data matches the standard speech data, is shown in the figure. Detailed Implementation

[0040] The following is in conjunction with the appendix Figure 1-3 The present invention will be described in further detail below.

[0041] Example 1 Embodiment 1 of the present invention provides a method for headphones to automatically identify noise, and the specific implementation solution is as follows: A method for headphone self-identification of noise, see attached document. Figure 1 This includes steps S101-S105: S101. Create a filter whitelist.

[0042] The whitelist is used to determine whether the acquired sound is noise. In this embodiment, the data in the whitelist includes voiceprint data and speech data. By analyzing the ambient sound from both voiceprint data and speech data simultaneously, noise identification can be performed more accurately.

[0043] Voiceprints, or "fingerprints" of sound, can be used for individual identification and are characterized by stability and measurability. By storing different voiceprint data in a whitelist, it can be shown that the sounds emitted by different voiceprints are those that users do not want to be directly filtered by headphones. This is similar to focusing on the voices of different people or specific voices.

[0044] Voice data is the semantic information carried in sound. By storing this voice data in a filtering whitelist, an interest content library can be created, and voices belonging to this interest content library will not be filtered.

[0045] S102. Acquire ambient sound and extract information of at least one set of sub-sounds from the ambient sound.

[0046] Ambient sound refers to all sounds present in a specific environment. For example, in nature, ambient sound can include animal sounds, wind sounds, and water sounds. In urban environments, ambient sound can include street noise, conversations, and traffic noise. In addition, noise from computer towers, air conditioners, and cars are also common ambient sounds.

[0047] Therefore, in order to achieve better sound recognition results, it is necessary to extract the sounds emitted by different sound sources in the ambient sound and analyze and recognize them separately.

[0048] The information of a sub-sound includes the information of the first sub-sound and the information of the second sub-sound. The information of different sub-sounds is emitted by different sound sources.

[0049] For example, when users are in nature, they can use DSP digital signal processing technology to preprocess the ambient sounds they collect. The preprocessing process includes echo cancellation, noise suppression, and filtering to extract sub-sounds of different frequencies. The information of the first sub-sound can be bird call information, and the information of the second sub-sound can be wind sound information.

[0050] For example, if a user is riding the subway, after extraction, the first sub-sound information can be the voice information, the second sub-sound information can be the arrival announcement sound information, and the third sub-sound information can be the subway's running sound information.

[0051] S103. Extract the feature data of the sub-sound.

[0052] For example, if a user turns on the transparency mode of their headphones in the subway, the system extracts feature data from the information of the first sub-voice and the information of the second sub-voice, respectively. The feature data includes target voiceprint data and target speech data.

[0053] Taking the information of the first sub-voice as an example, the information of the first sub-voice corresponds to the speech of the first sound source. The feature data can be extracted by the preset voiceprint recognition algorithm and the preset speech recognition algorithm to obtain the target voiceprint data and target speech data of the first sound source.

[0054] Next, the information of the second sub-voice is extracted, which corresponds to the arrival prompt sound information. After extraction, the target voiceprint data and target speech data of the second sound source are obtained.

[0055] When a user is in nature, the extracted sub-sound feature data can be only the target voiceprint data.

[0056] S104. Match the feature data with the data in the whitelist and determine the recognition type of the sub-sound based on the matching results.

[0057] In this embodiment, the identification types include valid sounds, noise, and undetermined sounds. When the feature data matches the data in the filtering whitelist, it is identified as a valid sound. When the feature data does not match the data in the filtering whitelist, it is identified as noise. When the feature data partially matches the data in the filtering whitelist, it is identified as an undetermined sound.

[0058] After identification, the first sub-sound can be classified as a valid tone, the second as noise, and the third as a tone to be determined. The more data in the whitelist, the wider the application scenarios, and the more accurate the headphones' sound type identification becomes.

[0059] S105. Send the valid tone to the user according to the recognition type.

[0060] For example, if the first sub-voice is the effective sound, the corresponding processing strategy could be to amplify the volume of the first sub-voice or to loop the first sub-voice, thereby helping the effective sound to be better received by the user.

[0061] Example 2 See attached document Figure 2 Embodiment 2 of the present invention provides another method for headphone self-identification of noise, and the specific implementation technical solution includes steps S201-S215.

[0062] S201. Acquire the target sound.

[0063] When creating a filter whitelist, you can first press the start button to activate the microphone and then use the microphone to capture the target sound. The microphone can be installed on the headphones, or the phone can connect to the headphones via a compatible app and use the phone's microphone to capture the target sound; there are no specific limitations here.

[0064] The target sound is the sound that the user wants to retain within the ambient sound environment, such as not wanting to miss the arrival announcement sound when riding the subway. Therefore, the arrival announcement sound needs to be recorded first. For example, if the user wants to retain the arrival announcement sound for station A, then the arrival announcement sound for station A can be recorded.

[0065] If you want to preserve the natural sounds in the environment, you can record natural sounds such as water sounds and wind sounds through a microphone.

[0066] S202. Extract standard speech data and standard voiceprint data from the target sound to serve as standard sound information.

[0067] In this embodiment, the target sound can be analyzed and data extracted using DSP digital signal processing technology, which includes a preset voiceprint recognition algorithm and a preset speech recognition algorithm.

[0068] The process of extracting standard voiceprint data includes: preprocessing the target sound, which includes pre-emphasis, framing, and windowing to enhance the clarity and recognizability of the target sound; extracting the sound that is directly related to the speaker and has high separability and stability from the preprocessed speech signal, which is the standard speech data; and then substituting the extracted sound into a preset speaker model (GMM) to obtain the standard voiceprint data corresponding to the target speech.

[0069] S203. Update the standard sound information to the filter whitelist.

[0070] After updating the standard sound information to the filter whitelist, the ambient sound can be identified according to the latest filter whitelist standard the next time it is acquired, thus making the identification accuracy higher and higher.

[0071] The whitelist can be stored on the DSP's execution unit, or it can be stored in the phone's memory through an app compatible with the headphones, or it can be stored in the cloud. One or more storage methods can be selected according to actual needs, and there are no specific restrictions here.

[0072] S204. Receive priority setting instructions and assign priority to different standard sounds according to the priority setting instructions.

[0073] In this embodiment, the priority setting command can be issued by the user through an app compatible with the headphones. For example, the user captures a subway arrival announcement sound through the microphone and sets the priority of this subway arrival announcement sound to the highest level in the app. Then, the user captures a sound of a friend A calling their name and sets the priority to the second highest level.

[0074] By setting different priorities for standard sounds, when multiple valid sounds are identified, the playback priority of the multiple valid sounds can be determined according to the priority of the standard sounds matched by the valid sounds. This helps users to obtain the most desired information more quickly among multiple valid sounds.

[0075] S205. Acquire ambient sound and extract information of at least one set of sub-sounds from the ambient sound.

[0076] S206. Extract the feature data of the sub-sound.

[0077] In this embodiment, steps S205-S206 are similar to steps S102-S103 above. To avoid repetition, the details will not be repeated here.

[0078] S207. Determine whether the target speech data matches the standard speech data. If yes, proceed to step S207. Identify the sub-voice as a valid sound. If no, proceed to step S208.

[0079] In this embodiment, the system first determines whether the target speech data matches the standard speech data, thereby helping users quickly obtain the information they want to hear. When the speech data is not what the user wants to hear, the system then determines whether the voiceprint information matches the standard voiceprint information to distinguish the importance of the sound.

[0080] When the target speech data matches the standard speech data, it indicates that the target speech data is useful, and step S207 is executed accordingly. When the target speech data does not match the standard speech data, the system further determines whether the target voiceprint information matches the standard voiceprint information, i.e., step S208 is executed, in order to determine whether the source of the sub-sound is the object of the user's interest.

[0081] See attached document Figure 3 Specifically, the methods for determining whether the target speech data matches the standard speech data include S2061-S2062.

[0082] S2071. Convert the target speech data into first text data, and convert the standard speech data into second text data; In this embodiment, DSP technology is used to perform real-time rapid conversion and recognition of target speech data and standard speech data.

[0083] DSP technology enables real-time processing of acquired data to meet various application requirements. In practical applications, DSPs typically need to store the acquired data in SDRAM (Synchronous Dynamic Random Access Memory) for processing when needed. To achieve efficient data storage and transmission, DSPs can utilize EDMA (Extended Direct Memory Access) technology to implement a Ping-Pong buffer.

[0084] Ping-Pong buffering is a common data stream processing method that uses two buffers (one for input data and the other for output data) alternately to achieve fast data processing. The EDMA controller can handle the data transfer between these two buffers, thereby reducing the CPU load and improving data processing speed.

[0085] After converting the target speech data and standard speech data into text data respectively, semantic analysis technology can be used to match the content of the first text data and the second text data.

[0086] S2072. Perform semantic analysis on the first text data and the second text data, and determine whether the target speech data matches the standard speech data by judging whether the semantic analysis results are consistent.

[0087] When the semantics of the first text data and the second text data are the same, it means that the information conveyed by the target speech data and the standard speech data is consistent or similar. Therefore, it can be determined that the target speech data matches the standard speech data.

[0088] When the semantics of the first text data and the second text data are different, it indicates that the information conveyed by the target speech data and the standard speech data are different. Therefore, it can be determined that the target speech data and the standard speech data do not match.

[0089] S208. Identify sub-voices as valid tones; Once a sound is identified as valid, it means that the sub-sound contains the information the user wants to know.

[0090] After step S208 is completed, proceed to step S212: amplify the volume of the sub-sound and send it to the user.

[0091] S209. Determine whether the target voiceprint information matches the standard voiceprint information. If yes, proceed to step S210. Identify the sub-voice as an undetermined sound. If no, proceed to step S211. Identify the sub-voice as noise.

[0092] When the voice data is not what the user wants to hear, the system checks whether the voiceprint information matches the standard voiceprint information. If they match, it means the sound source is something the user is interested in, and therefore it is classified as pending audio, allowing the user to decide whether to receive the sound later. If it is identified as noise, it means the sound does not contain any information the user wants to know.

[0093] By further enhancing the comparison between target voiceprint information and standard voiceprint information, the error tolerance of voice recognition can be increased, and interactivity can be improved, making the voice recognition results more user-friendly.

[0094] S210. Identify sub-voices as undetermined tones.

[0095] When a sound is identified as pending, it means that the first requirement of the sound part matches the filter whitelist, which may contain information that the user needs to obtain. Therefore, it is identified as pending and the user needs to further confirm the identification type.

[0096] After step S210 is completed, proceed to step S214. When a sub-sound is identified as an undetermined tone, prompt that a sub-sound exists and receive a determination instruction.

[0097] S211. Identify sub-sounds as noise.

[0098] If it is identified as noise, it means that the sub-sound does not contain the information the user wants to know.

[0099] After step S211 is completed, step S213 is executed. When a sub-sound is identified as noise, the sub-sound is filtered.

[0100] S212. Amplify the volume of the sub-sound and send it to the user; In this implementation, the volume is amplified to further improve the recognizability of the sub-sounds, making it easier for users to receive the information in the sub-sounds.

[0101] For example, if the sub-tone is the arrival notification tone, then the volume of the arrival notification tone is increased in the ambient sound to highlight it.

[0102] In addition, the effective tone can also be processed by looping the effective tone within a preset time to further help users obtain information.

[0103] After step S212 is completed, proceed to step S215. Play the valid tones in sequence according to their priority.

[0104] S213. Filter sub-sound; By filtering sub-sounds to reduce ambient noise, users can more fully enjoy the music and receive useful information from the outside world.

[0105] In this embodiment, noise can also be processed by reducing the volume of the noise to achieve a certain degree of balance and management of multiple sub-sounds.

[0106] S214. When a sub-sound is identified as an undetermined tone, a prompt is made indicating the existence of a sub-sound, and a determination instruction is received.

[0107] The judgment command is used to change the sub-sound to a valid tone or noise. In actual use, after the headphones issue a prompt, the user can generate the judgment command by pressing the corresponding button.

[0108] For example, if the voiceprint information of object B is pre-stored in the filtering whitelist, but the target speech data in the currently acquired voice of object B is inconsistent with the standard speech data, the headset will issue a prompt informing the user that object B is speaking and asking if they want to listen to its content. The user can choose yes or no based on their needs. If yes is selected, object B's voice is recognized as valid sound, and the volume is increased accordingly. If no is selected, object B's voice is recognized as noise and automatically filtered.

[0109] S215. Play the valid tones sequentially according to their priority.

[0110] When at least two valid sounds are identified, they are played sequentially according to priority. For example, the first valid sound is the subway arrival announcement, and the second valid sound is the voice of friend A. The standard sound corresponding to the first valid sound has the highest priority, so the second valid sound can be delayed. The delayed playback can be played immediately after the first valid sound finishes, or it can be delayed for 5 seconds or 10 seconds; the specific delay is not limited here.

[0111] In another possible embodiment, two valid tones with the same priority can be mixed and played to help the user receive more information at the same time.

[0112] In addition, when the user does not set a priority, the playback can be performed by default according to the time order in which the valid tones are emitted.

[0113] The implementation principle of this application is as follows: First, a filtering whitelist is set, and sounds that meet the filtering whitelist criteria are considered valid sounds. After acquiring ambient sounds, sounds from different sound sources in the ambient sounds are extracted, and the extracted sounds are identified separately. Recognition is performed from two dimensions: voiceprint data and speech data. Under preset recognition rules, by comparing the voiceprint data and speech data of the target sound with the data in the filtering whitelist, the sound judgment can be more accurate, thus making the recognition results more in line with the user's needs and solving the problem in existing technologies where headphones cannot accurately identify which are valid sounds and noise in transparency mode.

[0114] This application also discloses a system for headphones to identify noise, which can solve the problem in the prior art that headphones cannot accurately identify which are valid sounds and noise in transparency mode.

[0115] The specific implementation technical solutions for the above system are as follows: A system for self-identifying noise in headphones, comprising: The creation module is used to create a filter whitelist, which includes voiceprint data and voice data. The acquisition module is used to acquire ambient sound and extract information of at least one set of sub-sounds from the ambient sound, wherein the information of different sub-sounds is emitted by different sound sources; The extraction module is used to extract feature data of the sub-voice, which includes target voiceprint data and target speech data; The recognition module is used to match feature data with data in the filtered whitelist and determine the recognition type of the sub-sound based on the matching results. The recognition types include valid sound, noise, and undetermined sound.

[0116] The sound processing module is used to send valid sounds to the user based on the recognition type.

[0117] In this embodiment, the creation module may create a filter whitelist and send the whitelist data to the recognition module; the acquisition module may acquire ambient sound and extract information of at least one set of sub-sounds from the ambient sound, and send the sub-sound information to the extraction module; the extraction module may extract feature data of the sub-sounds and send the feature data to the recognition module; the recognition module may match the feature data with the whitelist data, determine the recognition type of the sub-sound based on the matching result, and send the recognition type to the sound processing module; the sound processing module may send the valid sound to the user according to the recognition type.

[0118] In a preferred embodiment, the present invention can be further configured such that the creation module includes a data acquisition unit, a processing unit, and an update unit, wherein: The acquisition unit is used to acquire the target sound and use the target sound as the standard sound. The processing unit is used to extract standard speech data and standard voiceprint data from the target sound; The update unit is used to update the standard voice data and standard voiceprint data to the filter whitelist.

[0119] In this embodiment, the acquisition unit may acquire the target sound and send it to the processing unit; the processing unit may extract standard speech data and standard voiceprint data from the target sound and send them to the update unit; the update unit may update the standard speech data and standard voiceprint data to the filtering whitelist.

[0120] In a preferred embodiment, the present invention can be further configured such that the identification module includes a first judgment unit, a first identification unit, a second judgment unit, a second identification unit, and a third identification unit, wherein: The first judgment unit is used to determine whether the target speech data matches the standard speech data; The first recognition unit is used to recognize the sub-voice as a valid voice when the first judgment unit determines that the target speech data matches the standard speech data. The second judgment unit is used to determine whether the target voiceprint information matches the standard voiceprint information when the first judgment unit determines that the target voice data does not match the standard voice data. The second identification unit is used to identify the sub-sound as noise when the second judgment unit determines that the target voiceprint information does not match the standard voiceprint information. The third recognition unit is used to recognize the sub-voice as an undetermined sound when the second judgment unit determines that the target voiceprint information matches the standard voiceprint information.

[0121] In this embodiment, the first judgment unit may determine whether the target speech data matches the standard speech data and send the judgment result to the first recognition unit and the second judgment unit; when the first judgment unit determines that the target speech data matches the standard speech data, the first recognition unit recognizes the sub-sound as a valid sound; when the first judgment unit determines that the target speech data does not match the standard speech data, the second judgment unit determines whether the target voiceprint information matches the standard voiceprint information and sends the judgment result to the second recognition unit and the third recognition unit; when the second judgment unit determines that the target voiceprint information does not match the standard voiceprint information, the second recognition unit recognizes the sub-sound as noise; when the second judgment unit determines that the target voiceprint information matches the standard voiceprint information, the third recognition unit recognizes the sub-sound as an undetermined sound.

[0122] In a preferred embodiment, the present invention can be further configured such that the first determining unit includes: The conversion subunit is used to convert the target speech data into first text data and the standard speech data into second text data; The analysis subunit is used to perform semantic analysis on the first text data and the second text data, and to determine whether the target speech data matches the standard speech data by judging whether the semantic analysis results are consistent.

[0123] In this embodiment, the conversion subunit may convert the target speech data into first text data and the standard speech data into second text data, and then send the first text data and the second text data to the analysis subunit; the analysis subunit performs semantic analysis on the first text data and the second text data, and determines whether the target speech data and the standard speech data match by judging whether the semantic analysis results are consistent.

[0124] In a preferred embodiment, the present invention can be further configured such that the sound processing module includes an amplification / transmission unit, a filtering unit, and a determination unit, wherein: An amplification / transmission unit is used to amplify the volume of a sub-sound and send it to the user when the sub-sound is identified as a valid tone; The filtering unit is used to filter out sub-sounds when they are identified as noise. The determination unit is used to indicate the existence of a sub-sound when it is identified as a pending tone, and to receive a determination instruction, which is used to change the sub-sound to a valid tone or noise.

[0125] In this embodiment, the amplification / transmission unit may amplify the volume of the sub-sound when it is identified as a valid tone; the filtering unit may filter the sub-sound when it is identified as noise; and the determination unit may indicate the existence of the sub-sound when it is identified as an undetermined tone, and receive a determination instruction. The determination instruction is used to change the sub-sound to a valid tone or noise, and the determination instruction is sent to the amplification / transmission unit or the filtering unit.

[0126] In a preferred embodiment, the present invention may be further configured such that the system also includes a setting module.

[0127] The setting module is used to receive priority setting instructions and assign corresponding priorities to different standard sounds according to the priority setting instructions.

[0128] In this embodiment, the setting module may also receive a priority setting instruction, assign a priority to different standard sounds according to the priority setting instruction, and send the priority information to the sound processing module.

[0129] In a preferred embodiment, the present invention may be further configured such that the sound processing module also includes a playback unit. The playback unit is used to play the two valid tones sequentially according to their priority when at least two valid tones are identified.

[0130] In this embodiment, when at least two valid tones are identified, the playback unit may play multiple valid tones sequentially according to the priority information sent by the setting module.

[0131] It should be noted that the above embodiments of the apparatus are only illustrated by the division of the above functional modules. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and the specific implementation process can be found in the method embodiments, which will not be repeated here.

[0132] This application also provides a device for headphone self-identification of noise. The device for headphone self-identification of noise may include: at least one processor, at least one network interface, a user interface, a memory, and at least one communication bus.

[0133] The processor is used to invoke a method storing headphone self-identification noise in memory, which, when executed by one or more processors, causes the headphone self-identification noise device to perform one or more of the methods described in the above embodiments.

[0134] This application also provides a computer-readable storage medium storing a computer program thereon. When the computer program is executed by a processor, it implements the headphone self-identification noise method of the above embodiments. To avoid repetition, it will not be described again here.

[0135] This specific embodiment is merely an explanation of the present invention and is not intended to limit the invention. After reading this specification, those skilled in the art can make modifications to this embodiment without contributing any inventive step, but such modifications are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. A method for headphone self-identification of noise, characterized in that, include: Creating a filtering whitelist includes: acquiring a target sound and using the target sound as a standard sound; extracting standard speech data and standard voiceprint data from the target sound; and updating the filtering whitelist with the standard speech data and standard voiceprint data as standard sound information. Acquire ambient sound and extract information of at least one set of sub-sounds from the ambient sound, wherein the information of different sub-sounds is emitted by different sound sources; Extract feature data from the sub-voice, the feature data including target voiceprint data and target speech data; The identification type is determined by matching the feature data with the data in the filtering whitelist and determining the identification type of the sub-sound based on the matching result. The recognition types include valid sounds, noise, and undetermined sounds. The judgment process specifically includes: determining whether the target speech data matches the standard speech data; if yes, then the sub-sound is recognized as a valid sound; if no, then determining whether the target voiceprint data matches the standard voiceprint data; if they match, then the sub-sound is recognized as an undetermined sound; otherwise, the sub-sound is recognized as noise. Determining whether the target speech data matches the standard speech data includes: converting the target speech data into first text data and converting the standard speech data into second text data; performing semantic analysis on the first text data and the second text data, and determining whether the target speech data matches the standard speech data by judging whether the semantic analysis results are consistent; The valid tone is sent to the user according to the identification type. The sending process specifically includes: when the sub-tone is identified as a valid tone, the volume of the sub-tone is amplified and sent to the user; when the sub-tone is identified as noise, the sub-tone is filtered; when the sub-tone is identified as an undetermined tone, the presence of the sub-tone is indicated and a determination instruction is received. The determination instruction is used to change the sub-tone to a valid tone or noise.

2. The method for headphone self-identification of noise according to claim 1, characterized in that, The method further includes: Receive a priority setting instruction, and assign a priority to different standard sound tags according to the priority setting instruction.

3. The method for headphone self-identification of noise according to claim 2, characterized in that, Sending the valid sound to the user according to the recognition type further includes: When at least two valid tones are identified, the multiple valid tones are played sequentially according to the priority.

4. A system for self-identifying noise in headphones, characterized in that, include: A creation module is used to create a filter whitelist, wherein the data of the filter whitelist includes at least one standard sound, and the standard sound includes standard voice data and standard voiceprint data; The creation module is used to acquire the target sound and use the target sound as the standard sound, extract the standard speech data and standard voiceprint data from the target sound as standard sound information, and update it to the filtering whitelist; An acquisition module is used to acquire ambient sound and extract information of at least one sub-sound from the ambient sound, wherein different sub-sounds are emitted by different sound sources; The extraction module is used to extract feature data of the sub-sound, the feature data including target voiceprint data and target speech data; The recognition module is used to match the feature data with the data in the filtering whitelist to determine the recognition type of the sub-voice. The recognition type includes valid sound, noise, and undetermined sound. The judgment process specifically includes: judging whether the target speech data matches the standard speech data. If they match, the sub-voice is recognized as a valid sound. If they do not match, judging whether the target voiceprint data matches the standard voiceprint data. If they match, the sub-voice is recognized as an undetermined sound. Otherwise, the sub-voice is recognized as noise. When judging whether the target speech data matches the standard speech data, the target speech data and the standard speech data are converted into first text data and second text data, respectively. Semantic analysis is performed on the first text data and the second text data. The matching is determined by judging whether the semantic analysis results are consistent. The sound processing module is used to send valid sounds to the user according to the recognition type; when the sub-sound is recognized as a valid sound, the volume of the sub-sound is amplified and sent to the user; when the sub-sound is recognized as noise, the sub-sound is filtered; when the sub-sound is recognized as an undetermined sound, the existence of the sub-sound is indicated and a determination instruction is received, the determination instruction being used to change the sub-sound to a valid sound or noise.

Images