Computer-implemented method for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor

Abstract

A computer-implemented method for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor (1) is proposed, wherein at least one correlation function is applied to the input signal and the correlation function is applied, in particular iteratively, to the result data of the correlation in order to increase the signal-to-noise ratio of the input signal.

Description

[0001] The invention relates to a computer-implemented method for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor. The invention further relates to a computing device, a system, and a vehicle. The invention also relates to a computer program for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor.

[0002] Various methods for increasing the signal-to-noise ratio of a noisy acoustic input signal are known in the prior art. These methods are only useful if the input signal clearly contains areas with and without the desired signal. However, if the desired signal is masked to such an extent that it is not detectable in the input signal, these methods are essentially inapplicable. Such masking frequently occurs, for example, with real signals from distant sirens that are recorded by acoustic sensors mounted externally on a fast-moving vehicle. The prior art is disclosed in US 2022 / 0363261A1 and CN 108731886A.

[0003] A computer-implemented method is proposed for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor. At least one correlation function is applied to the input signal. The correlation function is then iteratively applied to the resulting correlation data to improve the signal-to-noise ratio of the input signal.

[0004] The method is designed for processing noisy acoustic input signals detected by an acoustic sensor, preferably mounted externally on a vehicle. Alternatively, the acoustic sensor could be located separately from the vehicle, particularly on infrastructure elements such as traffic lights, road signs, or the like. The method can be executed by a computing device, such as a control unit, within the vehicle. "Designed" is understood to mean specifically programmed, specially equipped, and / or specially designed. The phrase "designed" means that an object performs a function in at least one operating state.

[0005] The acoustic sensor preferably comprises at least one microphone for capturing the input signal. In particular, the acoustic sensor comprises at least one interface for providing the input signal to the computing device that executes the method. The interface can be wired or wireless. Preferably, the acoustic sensor comprises at least one analog-to-digital converter for providing the input signal in digital form.

[0006] The input signal is affected by noise from wind, tire noise, internal vehicle noise, etc. A noise component of the input signal masks the useful signal contained within it, particularly in the broadband range. The useful signal is designed as a siren signal. The useful signal can also be designed as a warning signal, a traffic control signal, or any other signal that a specialist would consider appropriate.

[0007] Preferably, by applying the correlation function to the input signal and to the resulting correlation data, the signal-to-noise ratio of the input signal is increased to such an extent that the desired signal clearly stands out from the noise components and can be filtered out of the input signal. The signal-to-noise ratio of the input signal is, in particular, a ratio of the signal strength of the desired signal to the signal strength of the noise components. Specifically, the signal-to-noise ratio of the input signal is increased more frequently the correlation function is applied.

[0008] An iterative application of the correlation function corresponds, in particular, to multiple applications of the correlation function, especially to the result data of the preceding correlation. Preferably, the correlation function is applied to result data at least twice. In particular, the correlation function is applied to the input signal, then the correlation function is applied to the result data of the correlation of the result data of the input signal, and then the correlation function is applied to the result data of the correlation of the result data of the input signal.

[0009] Depending on the useful signal filtered by the method, various vehicle functions can be controlled. For example, a notification, particularly regarding an emergency vehicle, can be issued to the driver. Driving functions, such as braking and pulling over to the side of the road, can be controlled at least semi-autonomously.

[0010] The inventive design of the computer-implemented method advantageously allows a useful signal to be filtered from a noisy acoustic input signal. Advantageously, even heavily masked useful signals can be detected. Advantageously, a high level of traffic safety can be achieved.

[0011] Furthermore, it is proposed that the signal be a siren. Preferably, an emergency vehicle, such as an ambulance, emergency medical vehicle, fire engine, police car, or the like, emits the siren. Alternatively or additionally, it is conceivable that the siren is emitted by a stationary warning system, such as a disaster warning system. The siren can be, in particular, a siren, a yelp signal, a wail signal, or any other siren signal that would be considered appropriate by a person skilled in the art. It is advantageous that heavily masked siren signals can be detected.

[0012] Furthermore, it is proposed that the correlation function be designed as an autocorrelation function or as a cross-correlation function. This can advantageously enable a particularly effective correlation.

[0013] It is further proposed that the correlation function be applied block-wise to a specific number of samples of a signal and that the correlation of the entire signal be compiled from the results of the blocks. Depending on the process step, the signal is either the input signal or the result data from the preceding correlation. The input signal is preferably continuous-time. Advantageously, this allows for efficient correlation.

[0014] Furthermore, it is proposed that bandpass filtering of the input signal be performed before applying the correlation function. Preferably, the corner frequencies of the bandpass filter are chosen such that the desired signal is preserved after filtering. Advantageously, the filtering can be further optimized.

[0015] Furthermore, it is proposed that at least one noise reduction method, based on the principle of spectral subtraction of masking components, be applied to the final correlation results. The final correlation results correspond, in particular, to the results of a last iteration of the correlation. Preferably, the noise reduction method is used to determine the spectral properties of the noise or masking in the final results. Preferably, the determined properties are then spectrally minimized, especially by subtraction, using the noise reduction method. Advantageously, a particularly high signal-to-noise ratio can be achieved.

[0016] Furthermore, a computing device is proposed. The computing device comprises at least one interface for receiving at least one noisy acoustic input signal. The computing device comprises at least one computing module designed to execute a method according to the invention. The computing module can be configured, in particular, as a microprocessor, as an integrated circuit, especially as an FPGA (Field Programmable Gate Array), or as an application-specific integrated circuit (ASIC), or the like. Preferably, the computing device has at least one interface for receiving the measured values ​​from the acoustic sensor. Preferably, the computing device is connected to the acoustic sensor, in particular to the interface of the acoustic sensor, via the interface for data transmission purposes. In particular, the interface is designed to provide the measured values ​​to the computing module.The interface can be wired or wireless. A computing device can be advantageously provided to ensure a high level of traffic safety.

[0017] Furthermore, a system is proposed. The system comprises at least one computing device according to the invention. The system includes at least one acoustic sensor designed to detect at least one noisy acoustic input signal and provide it to the computing device. The computing device is preferably designed for use with the acoustic sensor. In particular, the acoustic sensor can include the computing device. For example, the computing device can be configured as a signal processor for the acoustic sensor. Alternatively, it is conceivable that the computing device is configured separately from the acoustic sensor. For example, a control unit (ECU), particularly an electronic one, of a vehicle can include or form the computing device. The acoustic sensor is preferably connected to the computing device, at least via data transmission. Advantageously, a system can be provided that enables a high level of road safety.

[0018] Furthermore, a vehicle is proposed. The vehicle comprises at least one computing device or at least one system according to the invention. Preferably, the vehicle is designed as an automated vehicle. An "automated vehicle" is understood to mean, in particular, a vehicle with one of the automation levels 1 to 5 of the SAE J3016 standard. Specifically, the automated vehicle has technical equipment required for these automation levels. This technical equipment includes, in particular, environmental sensing sensors, such as at least one acoustic sensor, radar sensors, lidar sensors and / or cameras, control units, or the like. Preferably, the vehicle is designed as a land vehicle.The vehicle can be configured, in particular, as a passenger car, preferably as a passenger transport vehicle, especially an autonomous shuttle, as a truck, as a construction vehicle, as an agricultural vehicle, or as any other vehicle that would be considered suitable by a person skilled in the art. Alternatively, the vehicle can also be configured as an aircraft, for example, as a drone, as an airplane, as a helicopter, as a vertical take-off and landing aircraft, or the like, or as a watercraft, especially as a ship, as a boat, or the like. Preferably, the vehicle can have a plurality of acoustic sensors, in particular distributed over an outer surface of the vehicle. Preferably, the computing device, in particular the computing module, is designed, in addition to carrying out the method, to control functions of the vehicle, for example, at least semi-autonomous driving functions, especially depending on the filtered useful signal.Alternatively or additionally, it is conceivable that the vehicle has at least one further computing device designed to control the vehicle's functions, particularly depending on the filtered useful signal. Specifically, this computing device can provide the filtered useful signal to the other computing device. This can advantageously result in a particularly safer vehicle.

[0019] Furthermore, a computer program is proposed for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor. The computer program comprises execution instructions which, when executed by a computing device according to the invention, cause it to execute a method according to the invention. Advantageously, a computer program can be provided that enables a high level of traffic safety.

[0020] The invention is illustrated by an exemplary embodiment in the following figures. They show: Fig. 1 a vehicle according to the invention in a schematic representation, Fig. 2 a system according to the invention of the vehicle according to the invention made of Fig. 1 in a schematic representation, Fig. 3 a flowchart of a process in a schematic representation, Fig. 4 a frequency spectrum of a useful signal in a schematic representation and Fig. 5 a time segment of an input signal in a schematic representation.

[0021] Figure 1Figure 1 shows a vehicle 6 in a schematic representation. In the present embodiment, the vehicle 6 is designed as an example of a land vehicle, in particular a passenger car. The vehicle 6 is also designed as an example of an automated vehicle. The vehicle 6 comprises at least one computing device 2. The vehicle 6 comprises at least one acoustic sensor 1. The acoustic sensor 1 is arranged on an outer surface of the vehicle 6. The acoustic sensor 1 and the computing device 2 form a system 5. The acoustic sensor 1 is designed to detect at least one noisy acoustic input signal and provide it to the computing device 2.

[0022] Figure 2 System 5 of vehicle 6 shows Figure 1in a schematic representation. In the present embodiment, the computing device 2 is designed separately from the acoustic sensor 1. The computing device 2 forms at least part of a control unit, in particular an electronic one, of the vehicle 6. The acoustic sensor 1 is connected to the computing device 2 via data transmission. The acoustic sensor 1 includes at least one interface 7 to provide the input signal to the computing device 2. The acoustic sensor 1 includes at least one microphone 8 to detect the input signal.

[0023] The computing device 2 comprises at least one interface 3 for receiving the at least one noise-laden acoustic input signal. The computing device 2 is connected via interface 3 to the acoustic sensor 1, in particular to interface 7 of the acoustic sensor 1, for data transmission purposes. The computing device 2 comprises at least one computing module 4, which is designed to process at least part of the Figure 3 to execute the described procedure, in particular a computer-implemented part of the procedure.

[0024] Figure 3 Figure 1 shows a flowchart of a process in a schematic representation. In a first process step 9, the at least one noisy acoustic input signal is received by the acoustic sensor 1. In a second process step 10, in particular before applying a correlation function, bandpass filtering of the input signal is performed.

[0025] In a third process step 11, at least one correlation function is applied to the input signal. In a fourth process step 12, the correlation function is applied, in particular iteratively, to the result data of the correlation in order to increase the signal-to-noise ratio of the input signal. The fourth process step 12 can be repeated, in particular multiple times, with the correlation function being applied, in particular, to the result data of the preceding correlation.

[0026] The signal is designed as a siren signal. In the present embodiment, the siren signal is designed, for example, as a siren signal, as shown in Figure 4The correlation function is shown in the present embodiment as an example of an autocorrelation function. In an alternative embodiment, the correlation function can also be implemented as a cross-correlation function. In the third and fourth process steps 11 and 12, the correlation function is applied block by block to a specific number of samples of a signal. The correlation of the entire signal is then compiled from the results of the blocks. In the third process step 11, the signal is the input signal. In the fourth process step 12, the signal is the result data from the preceding correlation.

[0027] In a fifth process step 13, at least one noise reduction method, based on the principle of spectral subtraction of masking components, is applied to the final correlation data. The second process step 10, the third process step 11, the fourth process step 12, and the fifth process step 13 constitute a computer-implemented method for filtering at least one useful signal from the noisy acoustic input signal received by the acoustic sensor 1. The computing module 4 is designed to execute at least the second, third, fourth, and fifth process steps 10-13.

[0028] Depending on the useful signal filtered by the method, functions of the vehicle 6 can be controlled. The computing device 2, in particular the computing module 4, can be designed to control the functions of the vehicle 6 in addition to executing the computer-implemented method.

[0029] A computer program product for filtering the at least one useful signal from the at least one noisy acoustic input signal received by the acoustic sensor 1 comprises execution instructions which, when the program is executed by the computing device 2, cause it to execute the computer-implemented method, in particular the second to fifth method steps 10-13.

[0030] Figure 4 Figure 14 shows a schematic representation of the frequency spectrum 14 of the desired signal. Time is plotted on the abscissa 15. Frequency is plotted on the ordinate 16. Horizontal bars 17 represent the desired signal. The desired signal is exemplified as a siren signal. The desired signal is composed of two sine tones spaced one fourth apart, each offset by one second, until a total signal length is reached.

[0031] Figure 5Figure 18 shows a time segment 18 of an input signal in a schematic representation. Time is plotted on an abscissa 19 axis. Signal strength is plotted on an ordinate 20 axis. A first curve 21 represents the input signal as received by the acoustic sensor 1. A second curve 22 represents the input signal after applying the correlation function to the input signal. A third curve 23 represents the input signal after applying the correlation function to the resulting data of the correlation of the input signal. A fourth curve 24 represents the input signal after applying the correlation function to the resulting data of the correlation of the input signal.

[0032] The input signal received by acoustic sensor 1 is so noisy that the desired signal cannot be distinguished from the noise. With each iteration of applying the correlation function, the signal-to-noise ratio of the input signal is further increased, so that the desired signal can be almost completely freed from the noise and filtered out of the input signal. Reference sign

[0033] 1 Acoustic sensor 2 Computing device 3 Interface 4 Computing module 5 System 6 Vehicle 7 Interface 8 Microphone 9 Process step 10 Process step 11 Process step 12 Process step 13 Process step 14 Frequency spectrum 15 Abscissa axis 16 Ordinate axis 17 Bar 18 Time segment 19 Abscissa axis 20 Ordinate axis 21 Curve 22 Curve 23 Curve 24 Curve

Claims

1. Computer-implemented method for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor (1) arranged on a vehicle, wherein the useful signal is designed as a siren signal, wherein the input signal is noisy due to wind noise, tire rolling noise or vehicle internal noise and a noise component of the input signal masks the useful signal contained in the input signal, wherein at least one correlation function is applied to the input signal and the correlation function is iteratively applied to result data of the correlation in order to increase a signal-to-noise ratio of the input signal.

2. Computer-implemented method according to claim 1, wherein the correlation function is configured as an autocorrelation function or as a cross-correlation function.

3. Computer-implemented method according to one of the preceding claims, wherein the correlation function is applied block-wise to a specific number of samples of a signal and wherein the correlation of the entire signal is composed over results of the blocks.

4. Computer-implemented method according to one of the preceding claims, wherein bandpass filtering of the input signal is performed before applying the correlation function.

5. Computer-implemented method according to one of the preceding claims, wherein at least one noise suppression method operating on the principle of spectral subtraction of masking components is applied to the final result data of the correlation.

6. Computing device comprising at least one interface (3) for receiving at least one noisy acoustic input signal, and at least one computing module (4) designed to perform a method according to one of the preceding claims.

7. System comprising at least one computing device (2) according to claim 6 and at least one acoustic sensor (1) which is provided to detect at least one noisy acoustic input signal and to provide it to the computing device (2).

8. Vehicle comprising at least one computing device (2) according to claim 6 or at least one system (5) according to claim 7.

9. Computer program product for filtering at least one useful signal from at least one noisy acoustic input signal received by at least one acoustic sensor (1), comprising execution instructions which, when the program is executed by a computing device (2) according to claim 6, cause it to execute a method according to one of claims 1 to 5.

Images