Incoming call reminder method, device, electronic device and storage medium
The matching situation between the incoming audio signal and the sampled signal is determined through the audio matching model in the smart device, which solves the problem of incoming calls leakage caused by the power limitation of the mobile phone speaker, and improves the success rate and user experience of answering calls.
Patent Information
- Application Number
- CN202210987158.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2022-08-17
- Publication Date
- 2024-11-08
- Estimated Expiration
- 2042-08-17
AI Technical Summary
In the prior art, the power increase of mobile phone speakers is limited, and the problem of incoming calls cannot be effectively solved, especially when the volume is low, the distance is long or the user's hearing is poor.
Using the audio matching model in the intelligent device, the signal pair is composed by collecting sample signals and pre-set incoming audio signals, and the sample signal pairs obtained by training are matched to determine whether the signal matches, and incoming calls are prompted in the matching situation to improve the user Success rate for answering calls.
Through intelligent device-assisted call prompts, the probability of incoming calls is reduced, prediction efficiency and accuracy are improved, user experience is improved, and audio matching is reliable when the terminal is far away, the volume is low or the user's hearing is poor. accuracy.
Smart Images

Figure CN115580679B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a method, device, electronic device and storage medium for prompting incoming calls. Background Art
[0002] With the development of communication technology, mobile phones have become an essential communication tool for everyone. When users use their mobile phones, voice calls are what they pay most attention to. However, when the volume of the mobile phone is low, the user is far away from the phone, or the user has poor hearing, calls may be missed.
[0003] In the prior art, the power of the internal speaker of the mobile phone is generally increased to provide call notification, which increases the power consumption of the mobile phone. At the same time, due to the limited size of the mobile phone body, the power of the speaker is limited and the speaker power cannot be increased to a better state. Summary of the invention
[0004] The present invention provides a method, device, electronic device and storage medium for prompting incoming calls, which are used to solve the defect of missed incoming calls in the prior art, and to assist in prompting incoming calls with the assistance of intelligent devices, thereby improving the success rate of users answering calls.
[0005] The present invention provides a method for prompting incoming calls, which is applied to a smart device and includes:
[0006] Collecting sampling signals;
[0007] The sampled signal and a preset incoming call audio signal are combined into a signal pair, and the signal pair is input into an audio matching model, and an audio matching result is outputted to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different;
[0008] When the audio matching result indicates that the sampled signal matches the incoming call audio signal, an incoming call prompt is performed.
[0009] According to the incoming call prompt method provided by the present invention,
[0010] In a case where the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal;
[0011] In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal;
[0012] In the case that the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively.
[0013] According to the incoming call prompting method provided by the present invention, the sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
[0014] According to the incoming call prompt method provided by the present invention, the sampling signal and the preset incoming call audio signal form a signal pair, and the signal pair is input into an audio matching model, and an audio matching result used to characterize whether the sampling signal and the incoming call audio signal match is output, including:
[0015] When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model;
[0016] When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
[0017] According to the incoming call prompt method provided by the present invention, the sampling signal and the preset incoming call audio signal form a signal pair, and the signal pair is input into an audio matching model, and an audio matching result used to characterize whether the sampling signal and the incoming call audio signal match is output, including:
[0018] When the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
[0019] According to the incoming call prompt method provided by the present invention, the method further includes:
[0020] The sampling signal and the incoming call audio signal are sequentially combined into a plurality of signal pairs according to the order of the incoming call frequencies corresponding to the incoming call audio signals from high to low.
[0021] The present invention also provides an incoming call reminder device, comprising:
[0022] An acquisition module, used for acquiring sampling signals;
[0023] A matching module, used to form a signal pair from the sampled signal and a preset incoming call audio signal, and input the signal pair into an audio matching model, and output an audio matching result used to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, the sample signal pairs including sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different;
[0024] The output module is used to provide an incoming call prompt when the audio matching result indicates that the sampling signal matches the incoming call audio signal.
[0025] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned incoming call prompt methods is implemented.
[0026] The present invention also provides a non-transitory computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the incoming call prompt method as described in any one of the above methods is implemented.
[0027] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the incoming call prompt method as described above is implemented.
[0028] The incoming call prompt method, device, electronic device and storage medium provided by the present invention apply the audio matching model in the smart device to determine whether the collected sampling signal and the pre-set incoming call audio signal match, thereby ensuring the rationality and accuracy of the audio feature matching between the sampling signal and the incoming call audio signal during the matching process, and using the smart device to assist in the incoming call prompt, thereby improving the success rate of users answering the phone and enhancing the user experience.
[0029] In addition, since the sample incoming call signals used to train the audio matching model form sample signal pairs with the first sample far-field signals and the second far-field signals of multiple test locations, this can diversify the samples and ensure the reliability and accuracy of audio matching when the terminal is far away, the volume is low, or the user has poor hearing. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0031] Figure 1 It is a flowchart of the incoming call prompt method provided by the present invention;
[0032] Figure 2 It is a schematic diagram of the structure of a training data set of the audio matching model provided by the present invention;
[0033] Figure 3 It is a structural schematic diagram of the incoming call reminder device provided by the present invention;
[0034] Figure 4 It is a structural schematic diagram of the electronic device provided by the present invention. DETAILED DESCRIPTION
[0035] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0036] In recent years, mobile phones have become an essential communication tool for everyone. When users use their mobile phones, calls may be missed if the volume of the phone is low, the user is far away from the phone, or the user has poor hearing.
[0037] At present, in order to improve the success rate of users answering calls, the power of the internal speaker of the mobile phone hardware is generally increased to provide call reminders, which increases the power consumption of the mobile phone. At the same time, due to the limited size of the mobile phone body, the power of the speaker is limited and the speaker power cannot be increased to an optimal state. Therefore, increasing the power of the speaker cannot solve the problem of missed calls.
[0038] In view of the above problems, the present invention provides a method for prompting incoming calls, which is applied to smart devices. Figure 1 FIG. 1 is a flow chart of the incoming call prompt method provided by the present invention, such as Figure 1 As shown, the method includes:
[0039] Step 110: Collect sampling signals.
[0040] Specifically, the sampling signal is an audio signal that needs to be input into the audio matching model together with a preset incoming call audio signal for matching prediction. The sampling signal is collected in real time by the smart device during the process of playing the incoming call audio signal after a call comes in to the user's mobile phone.
[0041] Optionally, the above-mentioned smart device can be a smart device that includes sound collecting devices such as pickups and microphones. Smart devices with sound collecting devices include but are not limited to: smart speakers, monitoring equipment, smart screens, smart robots, and tablet computers. For the sake of ease of explanation, the smart device takes a smart speaker as an example.
[0042] Step 120: Combining the sampled signal and a preset incoming call audio signal into a signal pair, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other;
[0043] Among them, the audio matching model is obtained by training based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include a first sample far-field signal collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test positions, and a second sample far-field signal collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test positions and the smart device are different.
[0044] Specifically, in order to determine whether the sampled signal is the incoming call ringtone of the user terminal, the input of the audio matching model is a signal pair consisting of the sampled signal and the incoming call audio signal pre-set in the smart device, and the output is an audio matching result of whether the sampled signal and the incoming call audio signal match, that is, the signal pair is input into the audio matching model, and the audio features of the sampled signal and the incoming call audio signal in the signal pair are extracted respectively. Based on the similarity of the audio features of the two signals, it is predicted whether the sampled signal is the pre-set incoming call audio signal, that is, whether the sampled signal is the incoming call ringtone of the user terminal. If the sampled signal is determined to be the incoming call ringtone of the user terminal, the incoming call reminder is performed through the smart device to avoid missed calls. If the sampled signal is determined not to be the incoming call ringtone of the user terminal, the incoming call reminder is not performed. It should be noted that the incoming call audio signal pre-set in the smart device is the same as the incoming call ringtone of the user terminal, such as: the incoming call audio signal pre-set in the smart device and the incoming call ringtone of the user terminal are the same song.
[0045] Step 130: When the audio matching result indicates that the sampled signal matches the incoming call audio signal, an incoming call prompt is issued.
[0046] Specifically, after the signal pair is input into the audio matching model, the output result includes whether the sampled signal matches the preset incoming call audio signal, or whether the sampled signal does not match the preset incoming call audio signal. If the sampled signal matches the preset incoming call audio signal, the smart device generates an incoming call prompt instruction to prompt the incoming call. The incoming call prompt method can be a voice broadcast method to remind the user to answer the call, or a light flashing method or a combination of the two. If the sampled signal does not match the preset incoming call audio signal, no incoming call prompt is performed, and new sampled signals are continuously collected for prediction.
[0047] Through the above steps, the present invention can install the audio matching model in the smart device, actively collect the sampling signal, predict whether the input sampling signal matches the incoming call audio signal through the audio matching model, and use the smart device to assist in incoming call prompts to reduce the probability of missed calls. At the same time, relying on the powerful computing power of the audio matching model, the prediction efficiency and accuracy are improved, thereby improving the user experience.
[0048] Optionally, before executing step 120, a training data set may be pre-constructed to train the initially constructed audio matching model. Figure 2 : is a schematic diagram of the training data set structure of the audio matching model provided by the present invention, such as Figure 2 As shown, the training data set includes multiple groups of sample signal pairs, each group of sample signal pairs includes a sample incoming call audio signal and a sample far-field signal, the sample far-field signal includes a first far-field signal and a second far-field signal, wherein the first far-field signal is obtained in the process of multiple terminals playing the sample incoming call audio signal at multiple test positions respectively, and the second far-field signal is obtained in the process of multiple terminals playing non-sample incoming call audio signals at multiple test positions respectively, the sample incoming call audio signal and the first far-field signal constitute a matching group, and the sample incoming call audio signal and the second far-field signal constitute a non-matching group, that is, the matching group indicates that the original signal played and the collected signal belong to the same song or ringtone, and the non-matching group indicates that the original signal played and the collected signal do not belong to the same song or ringtone, wherein the sample incoming call audio signal can be a sound source file of various songs and ringtones, and the format of the sound source file can be MP3 (Moving Picture Experts Group Audio Layer III, dynamic effects expert compression standard audio layer 3), WAV (WaveForm, waveform file), WMA (Windows Media Audio), MOV (QuickTime packaging format), FLAC (Free Lossless Audio Codec, lossless audio compression coding) and other formats, the present invention is not limited to this; the non-sample incoming call audio signal can be scene noise, such as multi-person chat noise, environmental noise, etc., and the second far-field signal can be collected during the playing of the scene noise.
[0049] It should be noted that in the training data set, the ratio of the number of sample signal pairs in the matching group and the non-matching group is as close to 1:1 as possible, so that the sample ratio in the training data set is balanced, and the audio matching result is avoided as much as possible due to the excessively high ratio of one of the sample signal pairs. The deviation is further ensured to further ensure the rationality and accuracy of the audio matching model prediction.
[0050] When constructing a training data set, it is taken into account that there is a certain distance between the user terminal and the user. The farther the distance, the lower the intensity of the incoming call ringtone heard by the user, and the greater the probability of missing the call. To address the above problem, in an embodiment of the present invention, when collecting sample far-field signals, the terminal that plays the sample incoming call audio signal and the non-sample incoming call audio signal is set at a test position at different distances from the smart device, thereby obtaining rich training data, so that the audio matching model can predict incoming call ringtones of different intensities, thereby further improving the prediction accuracy of the audio matching model. For example, taking the location of the smart device as the origin, the terminal is sequentially set at test positions 1m, 2m, 3m, 4m, 5m, or even farther away from the smart device to play in sequence, so that the intensities of the sample far-field signals collected sequentially by the smart device are different, thereby diversifying the training data.
[0051] Optionally, considering that the hardware parameters of different models of terminals will affect the audio features such as sampling frequency, sound quality, sound effect, sound intensity, bass fullness, etc. when playing the sample incoming call audio signal, multiple models of terminals can be used to play the sample incoming call audio signal and the non-sample incoming call audio signal. The power of the speakers in different models of terminals is different, so that the audio features of the collected sample far-field signals are different, and the training data is further enriched, so that the trained audio matching model can recognize the incoming call ringtones of various models of terminals, and expand the scope of use of the method provided in the embodiment of the present invention. In addition, the operating system of the terminal includes but is not limited to: Android system (Android), iOS system, Hongmeng system (Harmony), and the embodiment of the present invention does not limit this.
[0052] Optionally, in the training data set, a method for determining a sample signal pair includes:
[0053] In a case where the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal;
[0054] In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal;
[0055] In the case that the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively.
[0056] Specifically, since the ringing duration after an incoming call on a user terminal is different, in order to collect the sample signal and predict the audio matching result in a short time, and promptly remind the user to answer the call, there is no need to wait until the complete incoming call ringtone is collected before making a prediction. In the present invention, when constructing a training data set, a preset duration is set, and the collection duration of the sample incoming call audio signal is compared with the preset duration. If the collection duration is greater than the preset duration, the sample incoming call audio signal and the sample far-field signal are segmented respectively, and the segmented sample incoming call audio signal and the sample far-field signal are combined as a sample signal pair, wherein the playback duration of the segmented sample incoming call audio signal and the sample far-field signal is the same; if the collection duration is equal to the preset duration, then it is not matched. The sample signal pairs are processed and can be trained directly; if the collection time is less than the preset time, random noise is added to the sample incoming call audio signal and the sample far-field signal respectively until the playback time of the sample incoming call audio signal and the sample far-field signal is equal to the preset time, that is, considering that the playback time of the sample signal pair is short, the prediction error of the audio matching model is large. Therefore, in order to reduce the prediction error of the audio matching model, random noise is added before or after or on both sides of the sample incoming call audio signal and the sample far-field signal to provide a sample signal pair of sufficient length for the audio matching model to make predictions. Through the above method, the present invention can meet the real-time requirements in practical applications and further improve the user experience.
[0057] For example, the preset duration is set to 3s. If the collection duration of the sample incoming call audio signal and the sample far-field signal is 10s, and the collection duration is greater than the preset duration, the sample incoming call audio signal and the sample far-field signal are respectively segmented at intervals of 3s to obtain three sample signal pairs with a playback duration of 3s and one sample signal pair with a playback duration of 1s. Since the playback duration of the sample signal pair is 1s, which is less than the preset duration, 2s of random noise can be added before or after the segmented sample incoming call audio signal and the sample far-field signal, respectively, or 1s of random noise can be added to both sides of the sample signal pair; if the collection duration of the sample incoming call audio signal and the sample far-field signal is 3s, and the collection duration is equal to the preset duration, the sample signal pair will not be processed and can be directly used for training.
[0058] Optionally, the sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
[0059] Specifically, taking into account that users have different settings for the volume of terminal ringtones, in order to make the audio matching results of the audio matching model more accurate, in the present invention, the volume of the sample incoming call audio signal in the training data set can be preset to a fixed value, such as: the volume of the sample incoming call audio signal is set to 100%, and the volume can be adjusted step by step to obtain sample incoming call audio signals of multiple different volume levels, such as: the sample incoming call audio signal is divided into 5 levels, the volume of the first level is 100%, the volume of the second level is 90%, the volume of the third level is 80%, the volume of the fourth level is 70%, and the volume of the fifth level is 60%. The sample incoming call audio signal and non-sample incoming call audio signal of each volume level are respectively set to be played at different test positions and collected by smart devices, so that the training data is richer and the prediction accuracy of the audio matching model is further improved.
[0060] After constructing the sample signal pair, the category attributes of the sample incoming call audio signal and the sample far-field signal in the sample signal pair are annotated to obtain the sample label of the sample signal pair. If the sample signal pair belongs to the matching group, the sample label (label) of the sample signal pair is 1; if the sample signal pair belongs to the non-matching group, the sample label (label) of the sample signal pair is 0. Based on the sample signal pairs and the corresponding sample labels, a training data set is constructed to perform supervised training on the initial audio matching model.
[0061] After constructing the training data set, the training method of the audio matching model includes the following steps: constructing an initial audio matching model; then inputting the sample signal pairs and the corresponding sample labels into the initial audio matching model for iterative training, confirming the similarity of the two signal features after feature extraction, and then obtaining the loss function based on the difference between the predicted value and the output value, and updating the parameters of the initial audio matching model through back propagation, and continuing to input the next sample signal pair for training until the loss function is minimized, thereby obtaining a trained audio matching model.
[0062] The extracted features include but are not limited to: spectral features, harmonic features, Mel-scale Frequency Cepstral Coefficients (MFCC); the loss function can be the degree of difference between the sample label distribution and the output audio matching prediction result distribution. In addition, the initial audio matching model can be a traditional neural network, such as: Convolutional Neural Networks (CNN), Deep Neural Networks (DNN), Long Short-Term Memory (LSTM), or a hybrid neural network, such as: a combination of a neural network and a genetic algorithm, a combination of a neural network and an expert system, which is not specifically limited in the embodiments of the present invention.
[0063] Optionally, training may be performed by determining the similarity of corresponding audio features of a sample incoming call audio signal and a sample far-field signal in a sample signal pair, and the similarity measurement index may be the distance between the two audio features at the same time.
[0064] Optionally, for different numbers of signal pairs, the matching method of the audio matching model includes:
[0065] When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model;
[0066] When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
[0067] Specifically, when there is only one terminal, there is only one incoming call audio signal preset in the smart device, and the sampled signal and the preset incoming call audio signal form a signal pair, and the signal pair is directly input into the audio matching model for prediction; in addition, considering that a user has multiple terminals, or there are multiple users and multiple terminals in a home application scenario, there are multiple incoming call audio signals preset in the smart device. If one of the terminals has an incoming call ringtone, the smart device collects and obtains the sampled signal and forms multiple signal pairs with multiple preset incoming call audio signals. The smart device can create multiple threads and call the audio matching model for parallel prediction respectively, thereby improving prediction efficiency and improving user experience.
[0068] Optionally, when the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
[0069] Specifically, considering that a user has multiple terminals, or there are multiple users and multiple terminals in a home application scenario, multiple signal pairs are formed, and the multiple signal pairs are sorted and input into the audio matching model in sequence for prediction. That is, by traversing the sorted signal pairs, audio matching results are obtained in sequence. When an audio matching result that matches the sampled signal and the incoming call audio signal is obtained, subsequent signal pairs are stopped from being input into the audio matching model, and an incoming call prompt is given based on the audio matching result to remind the user to answer the call in time and avoid missing the call.
[0070] Optionally, in the case of multiple signal pairs, the method for sorting the signal pairs includes: sequentially grouping the sampled signals and the incoming call audio signals into multiple signal pairs in descending order of the incoming call frequencies corresponding to the incoming call audio signals.
[0071] Specifically, in the case of multiple signal pairs, the matching efficiency is low when prediction is performed through traversal. Therefore, in an embodiment of the present invention, based on historical data, the incoming call frequencies of the incoming call audio signals are sorted from high to low, and then they are respectively formed into signal pairs with the sampled signals, which are input into the audio matching model in sequence for matching, thereby increasing the probability of matching the incoming call audio signal with the sampled signal, thereby improving the matching efficiency and enhancing the user experience.
[0072] For example, assume that there are three preset incoming call audio signals in the smart device, namely A signal, B signal, and C signal. Based on the historical data predicted by the audio matching model multiple times, the incoming call frequencies of the three incoming call audio signals are 3 times / day for A signal, 5 times / day for B signal, and 1 time / day for C signal. The three incoming call audio signals are ranked from high to low as follows: B signal, A signal, and C signal. That is, after the incoming call rings, the probability of the sampled signal matching the B signal is the highest, the probability of the sampled signal matching the A signal is the second highest, and the probability of the sampled signal matching the C signal is the lowest. The sampled signal is first paired with the B signal and input into the audio matching model for prediction. If the audio matching result is that the sampled signal matches the B signal first, it indicates that the terminal corresponding to the B signal has an incoming call, and a call reminder can be made based on the matching result. At the same time, the sampled signal is no longer paired with the A signal, and the incoming call frequency and ranking results of the three incoming call audio signals are updated. If the sampled signal does not match the B signal first, the sampled signal is paired with the A signal and input into the audio matching model for prediction.
[0073] The incoming call prompt method provided by the present invention applies the audio matching model in the smart device to determine whether the collected sampling signal and the pre-set incoming call audio signal match. It relies on the powerful computing power of the audio matching model to ensure the rationality and accuracy of the audio feature matching between the sampling signal and the incoming call audio signal during the matching process, and uses the smart device to assist in the incoming call prompt, thereby improving the success rate of users answering the phone and enhancing the user experience.
[0074] In addition, since the sample incoming call signals used to train the audio matching model form sample signal pairs with the first sample far-field signals and the second far-field signals of multiple test locations, this can diversify the samples and ensure the reliability and accuracy of audio matching when the terminal is far away, the volume is low, or the user has poor hearing.
[0075] The incoming call prompting device provided by the present invention is described below. The incoming call prompting device described below and the incoming call prompting method described above can be referred to each other.
[0076] Figure 3 Schematic diagram of the structure of the incoming call reminder device provided by the present invention. Figure 3 The incoming call prompt device 200 is applied to a smart device, and the incoming call prompt device 200 includes: a collection module 201, a matching module 202 and an output module 203, wherein:
[0077] The acquisition module 201 is used to acquire the sampling signal;
[0078] A matching module 202 is used to form a signal pair with the sampled signal and a preset incoming call audio signal, and input the signal pair into an audio matching model, and output an audio matching result for characterizing whether the sampled signal and the incoming call audio signal match; the audio matching model is obtained by training based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different;
[0079] The output module 203 is used to provide an incoming call prompt when the audio matching result indicates that the sampled signal matches the incoming call audio signal.
[0080] The incoming call prompt device provided by the present invention applies the audio matching model in the smart device to determine whether the collected sampling signal and the pre-set incoming call audio signal match. It relies on the powerful computing power of the audio matching model to ensure the rationality and accuracy of the audio feature matching between the sampling signal and the incoming call audio signal during the matching process, and uses the smart device to assist in the incoming call prompt, thereby improving the success rate of users answering the phone and enhancing the user experience.
[0081] In addition, since the sample incoming call signals used to train the audio matching model form sample signal pairs with the first sample far-field signals and the second far-field signals of multiple test locations, this can diversify the samples and ensure the reliability and accuracy of audio matching when the terminal is far away, the volume is low, or the user has poor hearing.
[0082] Optionally, the sample signal pair is determined based on the following steps:
[0083] In a case where the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal;
[0084] In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal;
[0085] In the case that the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively.
[0086] Optionally, the sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
[0087] Optionally, the matching module 202 is specifically configured to:
[0088] When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model;
[0089] When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
[0090] Optionally, the matching module 202 is specifically configured to:
[0091] When the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
[0092] Optionally, the matching module 202 is specifically configured to:
[0093] The sampling signal and the incoming call audio signal are sequentially combined into a plurality of signal pairs according to the order of the incoming call frequencies corresponding to the incoming call audio signals from high to low.
[0094] Figure 4 An example of a physical structure diagram of an electronic device is shown in FIG. Figure 4 As shown, the electronic device 300 may include: a processor 310, a communication interface 320, a memory 330 and a communication bus 340, wherein the processor 310, the communication interface 320 and the memory 330 communicate with each other through the communication bus 340. The processor 310 may call the logic instructions in the memory 330 to execute the incoming call prompt method, which includes:
[0095] Collecting sampling signals;
[0096] The sampled signal and a preset incoming call audio signal are combined into a signal pair, and the signal pair is input into an audio matching model, and an audio matching result is outputted to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different;
[0097] When the audio matching result indicates that the sampled signal matches the incoming call audio signal, an incoming call prompt is performed.
[0098] Optionally, when the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal;
[0099] In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal;
[0100] In the case that the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively.
[0101] Optionally, the sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
[0102] Optionally, forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes:
[0103] When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model;
[0104] When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
[0105] Optionally, forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes:
[0106] When the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
[0107] Optionally, the method further comprises:
[0108] The sampling signal and the incoming call audio signal are sequentially combined into a plurality of signal pairs according to the order of the incoming call frequencies corresponding to the incoming call audio signals from high to low.
[0109] In addition, the logic instructions in the above-mentioned memory 330 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.
[0110] On the other hand, the present invention further provides a computer program product, the computer program product includes a computer program, the computer program can be stored in a non-transitory computer-readable storage medium, when the computer program is executed by a processor, the computer can execute the incoming call reminder method provided by the above methods, the method includes:
[0111] Collecting sampling signals;
[0112] The sampled signal and a preset incoming call audio signal are combined into a signal pair, and the signal pair is input into an audio matching model, and an audio matching result is outputted to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different;
[0113] When the audio matching result indicates that the sampled signal matches the incoming call audio signal, an incoming call prompt is performed.
[0114] Optionally, when the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal;
[0115] In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal;
[0116] In the case that the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively.
[0117] Optionally, the sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
[0118] Optionally, forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes:
[0119] When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model;
[0120] When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
[0121] Optionally, forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes:
[0122] When the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
[0123] Optionally, the method further comprises:
[0124] The sampling signal and the incoming call audio signal are sequentially combined into a plurality of signal pairs according to the order of the incoming call frequencies corresponding to the incoming call audio signals from high to low.
[0125] In another aspect, the present invention further provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the method for prompting incoming calls provided by the above methods is implemented, and the method includes:
[0126] Collecting sampling signals;
[0127] The sampled signal and a preset incoming call audio signal are combined into a signal pair, and the signal pair is input into an audio matching model, and an audio matching result is outputted to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different;
[0128] When the audio matching result indicates that the sampled signal matches the incoming call audio signal, an incoming call prompt is performed.
[0129] Optionally, when the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal;
[0130] In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal;
[0131] In the case that the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively.
[0132] Optionally, the sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
[0133] Optionally, forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes:
[0134] When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model;
[0135] When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
[0136] Optionally, forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes:
[0137] When the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
[0138] Optionally, the method further comprises:
[0139] The sampling signal and the incoming call audio signal are sequentially combined into a plurality of signal pairs according to the order of the incoming call frequencies corresponding to the incoming call audio signals from high to low.
[0140] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.
[0141] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.
[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.
Claims
1. A method for prompting incoming calls, characterized in that: Applied to smart devices, including: Collecting sampling signals; The sampled signal and a preset incoming call audio signal are combined into a signal pair, and the signal pair is input into an audio matching model, and an audio matching result is outputted to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, and the sample signal pairs include sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different; When the audio matching result indicates that the sampled signal matches the incoming call audio signal, providing an incoming call prompt; In a case where the acquisition time of the sample incoming call audio signal is greater than a preset time, the sample signal pair includes segments of the sample incoming call audio signal and the sample far-field signal; In a case where the acquisition time length of the sample incoming call audio signal is equal to the preset time length, the sample signal pair includes the sample incoming call audio signal and the sample far-field signal; In the case where the acquisition time of the sample incoming call audio signal is less than the preset time, the sample signal pair is obtained by adding random noise to the sample incoming call audio signal and the sample far-field signal respectively; The sample incoming call audio signal includes a sample incoming call audio signal with a preset volume or a plurality of sample incoming call audio signals with different volume levels.
2. The incoming call reminder method according to claim 1, characterized in that: The step of forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes: When the number of the signal pairs is equal to 1, inputting the signal pairs into the audio matching model to obtain an audio matching result output by the audio matching model; When the number of the signal pairs is greater than 1, the audio matching models corresponding to the number of the signal pairs are called by multi-threading, and the signal pairs are recognized in parallel based on each of the audio matching models to obtain the audio matching results output by each of the audio matching models.
3. The incoming call notification method according to claim 1, characterized in that: The step of forming a signal pair from the sampled signal and a preset incoming call audio signal, inputting the signal pair into an audio matching model, and outputting an audio matching result indicating whether the sampled signal and the incoming call audio signal match each other includes: When the number of signal pairs is greater than 1, starting from the first signal pair, the signal pairs are input into the audio matching model in sequence to obtain the audio matching results. When it is determined that the audio matching results are used to characterize the match between the sampled signal and the incoming call audio signal, stop inputting the signal pairs into the audio matching model.
4. The incoming call notification method according to claim 3, characterized in that: The method further comprises: The sampling signal and the incoming call audio signal are sequentially combined into a plurality of signal pairs according to the order of the incoming call frequencies corresponding to the incoming call audio signals from high to low.
5. A call reminder device for performing call reminder using the method according to any one of claims 1 to 4, characterized in that: Applied to smart devices, including: An acquisition module, used for acquiring sampling signals; A matching module, used to form a signal pair from the sampled signal and a preset incoming call audio signal, and input the signal pair into an audio matching model, and output an audio matching result used to characterize whether the sampled signal and the incoming call audio signal match; the audio matching model is trained based on sample signal pairs and sample labels corresponding to the sample signal pairs, the sample signal pairs including sample incoming call audio signals and sample far-field signals; the sample far-field signals include first sample far-field signals collected and obtained during the process of multiple terminals playing the sample incoming call audio signals at multiple test locations, respectively, and second sample far-field signals collected and obtained during the process of playing non-sample incoming call audio signals, and the distances between each of the test locations and the smart device are different; The output module is used to provide an incoming call prompt when the audio matching result indicates that the sampling signal matches the incoming call audio signal.
6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the incoming call prompt method as claimed in any one of claims 1 to 4 is implemented.
7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the incoming call prompt method according to any one of claims 1 to 4 is implemented.
