Method for locating specific sound source

Abstract

The present invention provides a method for locating a specific sound source. A sensor is used to collect sound wave signals at multiple positions on a preset path. Pre-processing is performed on the sound wave signals to obtain multiple signal features. The signal features are used as inputs of a deep learning model, and the deep learning model is used to perform signal recognition to obtain multiple specific audio signals at each position. An autocorrelation function operation is performed on these specific audio signals obtained at the same position to obtain multiple autocorrelation coefficients. A representative value is selected from these autocorrelation coefficients as a representative coefficient corresponding to each position. The specific sound source position is found according to the representative coefficient of each position.

Description

Technical Field

[0001] The present invention relates to a positioning method, and in particular to a positioning method for a specific sound source. Background Art

[0002] At present, most methods for detecting the point of occurrence of a specific sound source require the detection personnel to use handheld sensing equipment to gradually perform fixed-point sensing, and then use the detection personnel's experience to identify whether the sensed signal is a specific sound. Once it is confirmed that a specific sound is detected, the detection personnel will judge whether the location is the location where the specific sound is emitted based on their experience. However, the identification by the user is easily affected by factors such as subjective consciousness and external environment, resulting in poor identification accuracy. Summary of the invention

[0003] The present disclosure provides a method for locating a specific sound source, which can effectively improve the recognition accuracy.

[0004] The method for locating a specific sound source disclosed in the present invention includes: collecting sound wave signals at each of multiple positions on a preset path through a sensor; performing pre-processing on the sound wave signals to obtain multiple signal features; using the signal features as input to a deep learning model, and using the deep learning model to perform signal recognition to obtain multiple specific audio signals at each position; performing autocorrelation function operations on these specific audio signals obtained at the same position to obtain multiple autocorrelation coefficients; selecting representative values ​​from these autocorrelation coefficients as representative coefficients corresponding to each position; and finding the specific sound source position based on the representative coefficients of each position.

[0005] In one embodiment of the present disclosure, the above-mentioned step of performing pre-processing on the sound wave signal to obtain signal characteristics includes: performing Mel-cepstral calculation on the sound wave signal, and using the obtained Mel-cepstral coefficients, first-order differential Mel-cepstral coefficients, and second-order differential Mel-cepstral coefficients as signal characteristics.

[0006] In one embodiment of the present disclosure, the step of finding the specific sound source position according to the representative coefficients of each position includes: finding the maximum value among multiple representative coefficients of each position; and determining the position corresponding to the representative coefficient with the maximum value as the specific sound source position.

[0007] In one embodiment of the present disclosure, the step of finding the specific sound source position based on the representative coefficients of each position includes: calculating the difference between two representative coefficients of two adjacent positions, and when the difference is greater than a threshold, determining that the position corresponding to the larger of the two representative coefficients is the specific sound source position.

[0008] In one embodiment of the present disclosure, the deep learning model is a convolutional neural network.

[0009] In one embodiment of the present disclosure, the sensor obtains a plurality of sampling signals based on a sampling frequency and a sampling period, and uses the sampling signals as the sound wave signals.

[0010] In one embodiment of the present disclosure, the signal recognition using a deep learning model includes: using a deep learning model to determine whether the sound wave signal belongs to a specific audio signal; and when determining that the sound wave signal belongs to the specific audio signal, using the sampled signal as the specific audio signal.

[0011] In one embodiment of the present disclosure, the representative value is the maximum value of the autocorrelation coefficient.

[0012] Based on the above, the measured sound wave signal is identified in real time through the deep learning model, and the signal correlation analysis method is used to locate the location of the signal. By measuring the audio, specific events can be diagnosed in real time and the location of the event can be located to shorten the event processing time and improve the event processing efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] The accompanying drawings are included to provide a further understanding of the present invention and are incorporated in and constitute a part of this specification. The drawings illustrate embodiments of the present invention and together with the description serve to explain the principles of the present invention.

[0014] Figure 1A 4 is a block diagram of a device for locating a specific sound source according to an embodiment of the present disclosure.

[0015] Figure 1B is a block diagram of a device for locating a specific sound source according to another embodiment of the present disclosure.

[0016] Figure 2 is a flow chart of a method for locating a specific sound source according to an embodiment of the present disclosure. Figure 3 is a schematic diagram of detecting a specific sound source according to an embodiment of the present disclosure.

[0017] Reference numerals list

[0018] 1~9: Position

[0019] 100: Positioning device

[0020] 100A: Host

[0021] 110: Processor

[0022] 120: Storage device

[0023] 130: Sensor

[0024] 300A: Schematic diagram

[0025] 300B: Graph

[0026] 310: Preset path

[0027] D: Moving direction

[0028] S205-S230: Steps of the method for locating a specific sound source DETAILED DESCRIPTION

[0029] Reference will now be made in detail to exemplary embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Whenever possible, the same reference numerals are used in the drawings and the description to refer to the same or like parts.

[0030] Figure 1A 4 is a block diagram of a device for locating a specific sound source according to an embodiment of the present disclosure. Figure 1B is a block diagram of a device for locating a specific sound source according to another embodiment of the present disclosure.

[0031] exist Figure 1A In the embodiment, the positioning device 100 includes a processor 110, a storage device 120, and a sensor 130. The processor 110 is coupled to the storage device 120 and the sensor 130. The processor 110 is, for example, a central processing unit (CPU), a physical processing unit (PPU), a programmable microprocessor (Microprocessor), an embedded control chip, a digital signal processor (DSP), an application specific integrated circuit (ASIC), or other similar devices.

[0032] The storage device 120 is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk or other similar devices or a combination of these devices. The storage device 120 stores one or more code snippets, which, after being installed, will be executed by the processor 110 to implement the method for locating a specific sound source described below.

[0033] The sensor 130 is used to collect sound wave signals at each of the multiple positions on the preset path. In one embodiment, only one sensor 130 is required, but the present invention is not limited thereto. Multiple positions are pre-set on the preset path, and the positioning device 100 moves along the preset path from the initial position. Each time the positioning device 100 moves to a set position, the sensor 130 is used to collect the sound wave signals at the position, and then the processor 110 can find the position where the specific sound signal is emitted (hereinafter referred to as the specific sound source position) through the sound wave signals.

[0034] In another embodiment, if Figure 1B As shown, the positioning device 100 includes a host 100A and an independently arranged sensor 130. In this embodiment, the processor 110 and the storage device 120 are arranged in the same host 100A, and the sensor 130 is an independently arranged component. After collecting the sound wave signal, the sensor 130 transmits the sound wave signal to the host 100A through a wired or wireless transmission method. Here, one or more code snippets are stored in the storage device 120. After being installed, the above code snippets will be executed by the processor 110 to implement the positioning method of the specific sound source described later.

[0035] Figure 2 is a flow chart of a method for locating a specific sound source according to an embodiment of the present disclosure. Figure 3 FIG. 1 is a schematic diagram of detecting a specific sound source according to an embodiment of the present disclosure. Figure 3 In the figure, the upper schematic diagram 300A is used to represent that the sensor 130 collects acoustic wave signals at multiple positions 1 to 9 on the preset path 310, and the lower curve graph 300B represents the curve graph obtained based on the representative coefficients corresponding to positions 1 to 9.

[0036] Please refer to Figure 2 and Figure 3 In step S205, the sensor 130 collects acoustic wave signals at each of the plurality of positions 1 to 9 on the preset path 310. That is, when the sensor 130 moves to position 1, it stops to perform sampling to collect the acoustic wave signal at position 1. Then, the sensor 130 moves to position 2, stops to perform sampling to collect the acoustic wave signal at position 2, and so on to collect the acoustic wave signals at positions 1 to 9.

[0037] Next, in step S210, the acoustic signal is pre-processed to obtain a plurality of signal features. The processor 110 performs pre-processing on the acoustic signal collected at each position. In one embodiment, the processor 110 performs a Mel-Frequency Cepstrum (MFC) calculation on the acoustic signal, and uses the obtained Mel-Frequency Cepstrum coefficients, first-order differential Mel-Frequency Cepstrum coefficients, and second-order differential Mel-Frequency Cepstrum coefficients as signal features of the acoustic signal.

[0038] Then, in step S215, the processor 110 uses the signal features as input of a deep learning model and uses the deep learning model to perform signal recognition to obtain multiple specific audio signals at each location. The deep learning model is, for example, a convolutional neural network (CNN) model. The signal features captured in step S210 are used as input of the CNN model, and the output result of the CNN model is the recognition result.

[0039] The sensor 110 acquires a plurality of sampling signals based on a sampling frequency and a sampling period, and uses the sampling signals as the sound wave signal. For example, the sampling frequency of the sensor 130 is 8KHz, the sampling period is 1 second, and the number of samples is 185. This indicates that the sound wave signal includes 185 sampling signals. After performing steps S210 and S215, it is determined that the 185 sampling signals belong to specific audio signals, and these sampling signals are used as specific audio signals, and then the autocorrelation function operation (to be described in detail later) is performed.

[0040] The types of specific audio signals are, for example, water leakage sound, traffic sound, electrical sound, air leakage sound, or mechanical operation sound, but are not limited thereto. In one embodiment, during the training process, the signal features of water leakage sound, traffic sound, electrical sound, air leakage sound, mechanical operation sound, etc. calculated by Mel cepstrum are used to train a CNN model. Accordingly, after inputting the signal features into the CNN model, the CNN model can be used to identify the signal features as water leakage sound, traffic sound, electrical sound, air leakage sound, or mechanical operation sound.

[0041] For example, the signal features of the sound wave signals of electrical appliance sounds, traffic sounds, and water leakage sounds are input respectively, and the recognition results shown in Table 1 are obtained through CNN model recognition.

[0042] Table 1

[0043]

[0044] In Table 1, there are 53 sound wave signals whose actual category is traffic sound, and 53 of them are predicted as traffic sound by the CNN model. There are 66 sound wave signals whose actual category is electrical sound, and 66 of them are predicted as electrical sound by the CNN model. There are 66 sound wave signals whose actual category is water leakage sound, and 64 of them are predicted as water leakage sound by the CNN model, and 2 of them are misjudged as electrical sound. It can be seen that in this embodiment, the accuracy rate of using the CNN model to identify the signal characteristics of the sound wave signal can reach 98.9%.

[0045] A deep learning model is used to determine whether the sound wave signal belongs to a specific audio signal. When the sound wave signal is determined to belong to a specific audio signal, the sampled signal is used as the specific audio signal. In one embodiment, the main purpose is to detect the location of the water leakage. At this time, the sound wave signal identified as the water leakage sound is selected as the specific audio signal. Afterwards, in step S220, the processor 110 performs an autocorrelation function operation on multiple specific audio signals obtained in one position to obtain multiple autocorrelation coefficients. Afterwards, in step S225, the processor 110 selects a representative value from these autocorrelation coefficients as the representative coefficient corresponding to each position. The processor 110 uses the autocorrelation coefficients of multiple specific audio signals sampled at the same position to quantify the strength of the periodic signal. Specifically, in one embodiment, the representative value is a maximum value in the autocorrelation coefficient, that is, among the multiple autocorrelation coefficients of the multiple specific audio signals sampled at the same position, the maximum value is taken as the representative coefficient of the position, but it is not limited to this.

[0046] exist Figure 3 In the example of , the sound emitted from positions 1 to 9 set along the moving direction D on the preset path 300 belongs to a specific audio signal for explanation. First, when the sensor 130 moves to position 1 along the moving direction D, the processor 110 at position 1 can determine that the sound wave signal collected by the sensor 130 belongs to a specific audio signal through the CNN model, and then the autocorrelation function operation is performed on the multiple specific audio signals sampled at position 1, thereby obtaining multiple autocorrelation coefficients corresponding to position 1. Afterwards, the maximum value is taken from the multiple autocorrelation coefficients corresponding to position 1 as the representative coefficient corresponding to position 1. Next, the sensor 130 moves to position 2 along the moving direction D, and the processor 110 can determine that the sound wave signal collected by the sensor 130 belongs to a specific audio signal through the CNN model, and then the autocorrelation function operation is performed on the multiple specific audio signals sampled at position 2, thereby obtaining multiple autocorrelation coefficients corresponding to position 2. Afterwards, the maximum value is taken from the multiple autocorrelation coefficients corresponding to position 2 as the representative coefficient corresponding to position 2. Similarly, when gradually moving to positions 3 to 9, the representative coefficients corresponding to positions 3 to 9 can be calculated.

[0047] Then, in step S230, the processor 110 finds the specific sound source position according to the representative coefficients of each position. Specifically, the maximum value is found among the representative coefficients of the position, and the position corresponding to the representative coefficient with the maximum value is determined as the specific sound source position. Figure 3 For example, position 5 is determined to be the specific sound source position.

[0048] In another embodiment, the processor 110 may also calculate the difference between two representative coefficients of two adjacent positions, and when the difference is greater than a threshold, determine that the position corresponding to the larger of the two representative coefficients is the specific sound source position. Figure 3 The sound wave signals measured in the positions 1 to 9 shown are all determined to be specific audio signals. The representative coefficients of two adjacent positions are compared with each other until the difference between the two representative coefficients is greater than the threshold. The position with the largest representative coefficient is the specific sound source position.

[0049] The method for locating a specific sound source can be applied to locating water leakage sound, traffic sound, electrical appliance sound, air leakage sound or mechanical operation sound, and is not limited here. For example, in locating water leakage sound, audio features are used for identification. When a specific event (underground pipeline leakage) occurs, a specific sound is generated due to the pressure change of the material (liquid or gas) in the pipe.

[0050] In summary, the present disclosure only needs to use a single sensor to measure sound wave signals at multiple locations, and uses a deep learning model to identify the measured sound wave signals in real time, and uses the strong periodicity of the signal at the source of the specific audio to calculate the autocorrelation coefficient of the adjacent positions, thereby finding the specific sound source location. Accordingly, only a single sensor is needed to find the location where the specific audio is emitted (the specific sound source location). In addition, by measuring the sound wave signal, it is possible to diagnose in real time whether a specific event has occurred, and locate the location where the specific event has occurred, so as to shorten the event processing time and improve the event processing efficiency.

[0051] 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for locating a specific sound source, characterized in that: include: Collecting a plurality of acoustic wave signals at each of a plurality of locations on a preset path by means of a sensor; Performing pre-processing on each of the sound wave signals to obtain a plurality of signal features of each of the sound wave signals; Using the signal features as inputs of a deep learning model, and utilizing the deep learning model to perform signal recognition on each of the sound wave signals, so as to obtain a plurality of specific audio signals at each of the locations from the plurality of sound wave signals collected at each of the locations; Performing an autocorrelation function operation on the specific audio signal obtained at the same position to obtain a plurality of autocorrelation coefficients; Selecting a representative value from the autocorrelation coefficients as a representative coefficient corresponding to each of the positions; as well as Find the specific sound source position according to the representative coefficient of each position, The step of finding the specific sound source position according to the representative coefficient of each position includes: The difference between the two representative coefficients of two adjacent positions is calculated, and when the difference is greater than a threshold, the position corresponding to the larger one of the two representative coefficients is determined to be the specific sound source position.

2. The method for locating a specific sound source according to claim 1, characterized in that: The step of performing the pre-processing on each of the sound wave signals to obtain the signal characteristics of each of the sound wave signals comprises: Mel cepstrum calculation is performed on each of the sound wave signals, and the obtained Mel cepstrum coefficients, first-order differential Mel cepstrum coefficients, and second-order differential Mel cepstrum coefficients are used as the signal features.

3. The method for locating a specific sound source according to claim 1, characterized in that: The deep learning model is a convolutional neural network model.

4. The method for locating a specific sound source according to claim 1, characterized in that: The sensor obtains a plurality of sampling signals based on a sampling frequency and a sampling period, and uses the sampling signals as the plurality of sound wave signals.

5. The method for locating a specific sound source according to claim 1, characterized in that: The steps of using the deep learning model for signal identification include: Using the deep learning model to determine whether each of the sound wave signals belongs to a specific audio signal; and Each of the sound wave signals determined to belong to the specific audio is used as the specific audio signal.

6. The method for locating a specific sound source according to claim 1, characterized in that: The representative value is the maximum value among the autocorrelation coefficients.

Images