Audio signal processing method

Abstract

The present invention relates to an audio signal processing method (100), comprising the following steps: providing (110) an audio signal; providing (120) a filtering function; applying (130) the filtering function to the audio signal in the time domain to provide a filtered audio signal; and outputting the filtered audio signal (140); wherein the provision of the filtering function depends on a single filtering function, wherein; and wherein the single filtering function includes at least two high-pass filtering functions or at least two low-pass filtering functions, and a band-pass filtering function is obtained based on the at least two high-pass filtering functions or at least two low-pass filtering functions.

Description

Technical Field

[0001] This invention relates to an audio signal processing method and a corresponding apparatus. Background Technology

[0002] Various audio signal processing methods suitable for different application scenarios are known in the prior art. For example, (audio) signal processing methods are described in DE 102022 111 300 A1 and US 6 141 672 A.

[0003] Existing methods are primarily used in the fields of communication technology and hearing aids to suppress interference noise. This typically involves first acquiring an unfiltered audio signal, then transferring it to the frequency domain, and finally filtering the audio signal within the frequency domain. When converting the acquired time signal to the frequency domain, a short-time Fourier transform (also known as a short-time Fourier transform or STFT) is usually employed. The advantage of frequency domain signal processing lies in its ability to selectively suppress or enhance specific frequencies. For example, an audio signal can be attenuated in the frequency domain where interference noise is expected, while the signal can be enhanced or at least kept attenuated in the frequency domain where speech information is expected.

[0004] One drawback of frequency-domain audio signal processing is that audio data must be processed frame by frame, resulting in higher latency. Furthermore, the audio signal (acquired as a time-correlated signal) must first be converted to the frequency domain, filtered there, and then converted back to the time domain, increasing computational complexity. This resulting latency can lead to unnatural perception and even disorientation for the user, as the perceived acoustic signal is delayed relative to visual perception (e.g., in a conversation between two people, one of whom is wearing a hearing aid). The superposition of passive sound with delayed active reproduction sound can produce a comb filtering effect, leading to spectral distortion. Moreover, if the processing latency is too high, the system may not be able to react quickly enough to impulse noise, thus failing to prevent, for example, hearing loss.

[0005] Many existing audio codecs are equipped with dedicated processors for recording, processing, and playing audio signals with low latency. Signal conversion and processing are typically performed per sample point, so at relatively high sampling rates, latency can be in the microsecond range. However, such processors usually have only basic components and limited resources to combine into an audio signal processing topology. Therefore, for more complex audio applications such as interference and noise suppression, existing methods often employ the well-established framework based on the aforementioned short-time Fourier transform. An alternative is the filter bank approach, where filter banks are typically designed based on linear-phase FIR filters. However, known approaches in the prior art result in significant latency, often reaching several milliseconds, which is generally undesirable in practice. Summary of the Invention

[0006] In view of the above-mentioned drawbacks of known methods in the prior art, the purpose of this invention is to provide an audio signal processing method that can achieve efficient and low-latency audio signal processing.

[0007] To achieve the above objectives, the present invention proposes an audio signal processing method, which includes the following steps: - Provide audio signal ; - Provide filtering functions; - Applying the filtering function to the audio signal in the time domain To provide filtered audio signals ;as well as - Output filtered audio signal ;in - The provision of the filter function depends on A single filter function, wherein ;and - Wherein a single filter function includes at least two high-pass filter functions or at least two low-pass filter functions, and a band-pass filter function is obtained based on the at least two high-pass filter functions or at least two low-pass filter functions.

[0008] Unlike most methods known in the prior art, the method of the present invention filters the audio signal in the time domain. Furthermore, the method of the present invention can efficiently provide a filtering function using at least two high-pass filters or at least two low-pass filters. As described below, according to... When providing a single filter function, different implementation variations can be used within the scope of this invention. Although the general form of the invention can be described based on two single filter functions, it is obvious that more than two single filter functions can also be used. In particular, multiple bandpass filter functions can be provided using four, eight, or 16 single filter functions (high-pass or low-pass filters). Based on these bandpass filter functions, a filter function (also known as the total filter function) can then be calculated. As described below, each bandpass filter function can be used to provide a bandpass-filtered audio signal, which is then multiplied by a weighted function depending on the specific application scenario (e.g., for suppressing specific interfering noise, such as road noise, coffee shop noise, or channel noise).

[0009] In the method of the present invention, it is preferably designed to provide the high-pass filter function online (i.e., in real time). This is achieved through the efficient implementation of the method of the present invention, particularly by efficiently providing the band-pass filter function using at least two high-pass filter functions or at least two low-pass filter functions.

[0010] In the method of the present invention, it is preferably designed such that the provision of the filter function includes subtracting the first single filter function from the second single filter function, wherein the first single filter function and the second single filter function are respectively implemented as filter functions of a high-pass filter or filter functions of a low-pass filter.

[0011] The subtraction of a single filter function can be performed either directly or implicitly by introducing a modified weighting factor, as will be explained in detail below. Thus, by subtracting two high-pass filter functions or two low-pass filter functions, a band-pass filter function can be provided with relatively little computation.

[0012] In some embodiments of the method of the present invention, it can be designed as follows: - Using two high-pass filters (specifically two adjacent high-pass filters), the high-pass filter transfer function of these two high-pass filters Designed such that its amplitude response is at the first limiting frequency The above have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, particularly preferably a maximum deviation of 3% or a maximum deviation of 1%, or - Using two low-pass filters (specifically two adjacent low-pass filters), the low-pass filter transfer function of these two low-pass filters. Designed such that its amplitude response is at the second limiting frequency The following have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, and particularly preferably a maximum deviation of 3% or 1%.

[0013] In other words, two high-pass filters can be used, designed so that their amplitude responses are close to each other at high frequencies, or two low-pass filters can be used, designed so that their amplitude responses are close to each other at low frequencies. Within the scope of this invention, the first and second threshold frequencies are also referred to as stopband frequencies. The amplitude response deviation (expressed as a percentage) of the two high-pass filters or the two low-pass filters can be defined as follows: , in Indicates the deviation in amplitude response. This indicates the amplitude response of the first single filter (high-pass or low-pass filter) above or below the corresponding stopband frequency. This represents the amplitude response of the second single filter (high-pass or low-pass filter) above or below the corresponding stopband frequency. In a preferred embodiment, the deviation of the amplitude response above or below the corresponding stopband frequency... Accordingly, it is always less than the aforementioned 10%, 5%, 3% or 1%.

[0014] By using a single filter function whose amplitude response is at least one side close to each other, destructive interference can be achieved, thus enabling efficient bandpass filtering. This will be explained in detail below with reference to the accompanying drawings.

[0015] High-pass and low-pass filter functions are typically given in the z-domain.

[0016] In some embodiments of the method of the present invention, it can be designed as follows: - Using two high-pass filters, the high-pass filter transfer function of these two high-pass filters They were designed such that their phase response was at the first limiting frequency. The above have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, particularly preferably a maximum deviation of 3% or a maximum deviation of 1%, or - Using two low-pass filters, the low-pass filter transfer functions of these two low-pass filters. Designed such that its phase response is at the second limiting frequency The following have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, and particularly preferably a maximum deviation of 3% or 1%.

[0017] In other words, two high-pass filters can be used, designed so that their phase responses are close to each other at high frequencies, or two low-pass filters can be used, designed so that their phase responses are close to each other at low frequencies. The phase response deviation (expressed as a percentage) between the two high-pass filters or the two low-pass filters can be defined similarly to the amplitude response deviation described above: , in This indicates the deviation in phase response. This indicates the phase response of the first single filter (high-pass or low-pass filter) above or below the corresponding stopband frequency. This indicates the phase response of the second single filter (high-pass or low-pass filter) above or below the corresponding stopband frequency. In a preferred embodiment, the deviation of the phase response above or below the corresponding stopband frequency... The phase is correspondingly always less than the aforementioned 10%, 5%, 3% or 1%.

[0018] Destructive interference can be achieved by using a single filter function whose phase response is at least one side close to each other and performing a subtraction operation, thereby providing the desired bandpass filter. Destructive interference can also be achieved by modifying the phase function (particularly by a 180° phase rotation) and performing an addition operation, which, according to the definition of the invention, is also considered subtraction.

[0019] Furthermore, in the method of the present invention, a single filter can be implemented as a finite impulse response (FIR) filter or an infinite impulse response (IIR) filter.

[0020] The advantage of using FIR filters lies in their relatively simple design and implementation. Furthermore, because FIR filters do not use feedback loops, they are always stable.

[0021] The advantage of using IIR filters is their higher efficiency. Furthermore, compared to FIR filters, IIR filters can achieve a steeper threshold frequency.

[0022] In the method of this invention, it is preferable that the single filter has nonlinear phase characteristics. Preliminary studies have shown that by using a single filter with nonlinear phase characteristics, the delay can be further reduced.

[0023] In some preferred embodiments of the method of the present invention, the filtered audio signal can be designed as follows: Determine according to the following formula: in Let represent the weighting factors, and . Determine according to the following formula: , in Describe the high-pass filter respectively Filtered audio signal And high-pass filter Each has a limiting frequency ,in By implementing the filtering function described above, a high-pass filter can be used to efficiently process audio signals, thereby achieving low latency.

[0024] In other preferred embodiments of the method according to the invention, the audio signal can be designed as follows: Determine according to the following formula: in Let represent the weighting factors, and . Determine according to the following formula: , in Describe the low-pass filter respectively Filtered audio signal And the low-pass filter Each has a limiting frequency ,in By implementing the filtering function described above, a low-pass filter can be used to efficiently process audio signals, thereby achieving low latency.

[0025] As mentioned earlier, subtraction in a single filter function can also be implicitly implemented using a modified weighting factor. Therefore, it is preferable to design the filtered audio signal as follows: Determine according to the following formula: in Describe the high-pass filter respectively Filtered audio signal The high-pass filter Each has a limiting frequency ,in ,and These represent the modified gain factors, which are determined as follows: .

[0026] Therefore, the filtering function can be implemented with high-pass filters in a particularly efficient manner, thereby achieving low latency.

[0027] Alternatively, it can be preferably designed as: filtered audio signal Determine according to the following formula: in Describe the low-pass filter respectively Filtered audio signal And the low-pass filter Each has a limiting frequency ,in ,and These represent the modified gain factors, which are determined as follows: .

[0028] Therefore, a low-pass filter can be used to provide a particularly efficient way to implement the filtering function, thereby achieving low latency.

[0029] In some embodiments of the method of the present invention, at least one single filter function can be designed as a delta function (also known as the Dirac function or unit impulse function). In this case, it is preferable to set... , , or There are several single filter functions, at least one of which is designed as a delta function. This is equivalent to short-circuiting a high-pass or low-pass filter. Therefore, the overall delay caused by the high-pass or low-pass filter can be reduced when needed. Simultaneously, the complexity of the filter can be reduced when necessary. In the case of using a high-pass filter, it is preferable to make the first single filter function ( The last filter function () is designed as a delta function. When using a low-pass filter, it is preferable to design the last filter function () as a delta function. It is designed as a delta function.

[0030] According to some embodiments of the method of the present invention, it can be designed as follows: - Weighting factor Compared with audio signals filtered by high-pass or low-pass filters Calculated signal Multiplication or modified weighting factor Audio signals filtered by high-pass or low-pass filters The multiplication is based on the first frequency In progress; - at the first frequency Weighting factors According to the filtered audio signal Calculated signal Summing the products, or applying the modified weighting factors. Audio signals filtered by high-pass or low-pass filters Summing the products; and - at the second frequency Calculate weighting factors Or the modified weighting factor ;in - Second frequency Below the first frequency .

[0031] Preliminary studies indicate that the second frequency used in calculating the weighting factor or the modified weighting factor ( Alternatively, the frequency can be selected to be lower than the (modified) weighting factor and the filtered audio signal. Multiplication and the (modified) weighting factor with the filtered audio signal The first frequency used to sum the products ( This allows for the processing of audio signals more efficiently without significantly affecting the quality of the filtered audio signal, or frequency, without substantially impacting its quality. This further reduces latency.

[0032] For example, a first frequency can be set. And the second frequency You can also set the first frequency. And the second frequency The first and second frequencies can be modified according to the application scenario. To achieve efficient sampling rate conversion, and The ratio between Advantageously, the ratio should be an integer. The ratio can be chosen according to requirements, advantageously between 2 and 64, preferably between 4 and 32, and particularly preferably between 8 and 16. In particular, as shown in the example above, the ratio can be selected... = 12. At this time, the second frequency Preferably, the kHz values ​​are ≤ 48 kHz, ≤ 24 kHz, or ≤ 16 kHz. Preliminary studies have shown that by selecting the above parameter values, efficient processing of audio signals can be achieved while ensuring good quality of the filtered audio signal.

[0033] Furthermore, the method of the present invention can be designed to apply the filtering function to the audio signal in the time domain. This includes applying convolution operations to audio signals. and the impulse response of high-pass or low-pass filters. .

[0034] To achieve the aforementioned objective, an audio signal processing apparatus is further proposed, comprising: - Used for acquiring audio signals The acquisition unit; - Used to calculate filter functions and process audio signals The computational unit is designed to apply a filtering function to the audio signal in the time domain. And calculate the filtered audio signal. ; - Used to output filtered audio signals The output unit; Its features are, - The computing unit is designed to be based on Each single filter function is used to obtain the filter function, where ;as well as - Obtain the bandpass filter function from at least two single filter functions, including two high-pass filter functions or two low-pass filter functions.

[0035] The device of the present invention is capable of processing audio signals with particularly high efficiency in a low-latency manner.

[0036] The apparatus of the present invention is preferably configured such that the computing unit is designed to calculate the bandpass filter function by subtracting the first single filter function from the second single filter function, wherein the first single filter function and the second single filter function are respectively implemented as the filter function of a high-pass filter or the filter function of a low-pass filter.

[0037] Furthermore, the apparatus of the present invention may also be designed to have the characteristics described above in conjunction with the method of the present invention, and the computing unit of the apparatus of the present invention is designed to perform the method steps described in conjunction with the method.

[0038] In particular, the high-pass or low-pass filter used in the method of the present invention can be implemented as a nonlinear phase filter.

[0039] In this invention, it is preferable to implement the high-pass filter or low-pass filter as a recursive minimum-phase filter.

[0040] Alternatively, it can be designed as: a first high-pass filter ( ) or the last low-pass filter ( The ) is replaced by a short circuit. In this way, the computational complexity can be reduced, so as to achieve lower latency.

[0041] Furthermore, it is preferable to design the filter to calculate a weighting factor or a modified weighting factor based on at least one sensor signal, particularly a microphone signal. This allows for a response to dynamic interference signals and enables the filter function to match the acquired interference signal. Attached Figure Description

[0042] The present invention will now be described in detail with reference to the accompanying drawings. Wherein: Figure 1 This is a flowchart of an embodiment of the present invention. Figure 2 This is a signal flow graph according to an embodiment of the present invention. Figure 3 This is an exemplary amplitude response of a high-pass filter according to an embodiment of the present invention. Figure 4This is an exemplary phase response of a high-pass filter according to an embodiment of the present invention. Figure 5 As an exemplary amplitude response of the bandpass filter obtained according to an embodiment of the present invention, Figure 6 This is a signal flow graph of an embodiment of the present invention based on a low-pass filter. Figure 7 This is a signal flow graph of an embodiment of the invention using the modified weighting factor. Figure 8 This is a signal flow graph from an embodiment of the present invention, wherein the first high-pass filter is replaced by a short circuit. Figure 9 This is a schematic diagram of one embodiment of the device of the present invention. Figure 10 This is a schematic diagram of another embodiment of the device of the present invention, which includes multiple sensors, a preprocessing unit, and a speaker. Figure 11 This is a schematic diagram of another embodiment of the device of the present invention, which is in the form of an in-ear headphone, including multiple sensors, a speaker, and a computing unit for calculating and applying a filtering function. Detailed Implementation

[0043] Figure 1 A flowchart illustrating an embodiment of the method 100 of the present invention is shown. In a first method step 110, an audio signal is provided. This can be achieved by using a microphone or other sensor. Alternatively, the audio signal can be designed to be provided by an external source that has previously acquired the audio signal. In a second method step 120, a filtering function is provided. The provision of this filtering function depends on... A single filter function, wherein These individual filter functions can be implemented as high-pass or low-pass filter functions. A band-pass filter function is calculated based on these individual filter functions. This can be achieved, in particular, by subtracting two high-pass or two low-pass filter functions. In the third method step 130, the provided filter function is applied to the audio signal in the time domain. This provides a filtered audio signal. In the fourth method step 140, the filtered audio signal is output. The filtered audio signal can be output through a speaker or other output unit, or it can be output to an external device via signal transmission.

[0044] Figure 2 An exemplary embodiment of the method of the present invention is shown. In this embodiment, the input signal is... The signal is transmitted to multiple high-pass filters 10 via a parallel circuit, among which... Figure 2 The first high-pass filter 10a, the second high-pass filter 10b, and the third high-pass filter 10b are shown. A high-pass filter 10c. With limiting frequency ,in The specific setting of the limiting frequency can be designed differently for different application scenarios. A high-pass filtered signal is output at the output of high-pass filter 10. Subsequently, adjacent high-pass filtered signals are subtracted to generate a bandpass signal. .

[0045] Next, time-varying weighting factors are used. The output signal is obtained by weighting these bandpass signals and then summing them. .

[0046] Weighting factor The weighting factor can preferably take values ​​from 0 to 1. In some embodiments of the invention, it can also be designed such that the weighting factor can take values ​​greater than 1 or less than 0. The weighting factor is also calculated according to the specific application. Methods known in the prior art can be used to calculate the weighting factor. In order to generate the corresponding bandpass signal from each high-pass filter by subtraction, each high-pass filter preferably has specific amplitude and phase characteristics. These preferred characteristics are exemplarily illustrated in... Figure 3 and Figure 4 In China. Especially from Figure 3 As can be seen, the amplitude responses 20 of each high-pass filter are close to each other at high frequencies. Figure 3 The diagram shows the first amplitude response 20a, the second amplitude response 20b, the third amplitude response 20c, and the fourth amplitude response 20d. For example... Figure 3 As shown, the amplitude response 20 is above a certain frequency (especially at the upper stopband frequency). The above are relatively close to each other.

[0047] In addition, from Figure 4 It can also be seen that the phase responses 20 are similar at high frequencies. Figure 4 The diagram shows the first phase response 21a, the second phase response 21b, the third phase response 21c, and the fourth phase response 21d. The bandpass transfer function can be defined as the transfer function of two high-pass filters in... The difference in the domain is as follows: .

[0048] Due to the high-pass characteristic of high-pass filters, band-pass filters also have high-pass characteristics, and their corresponding lower stopband frequency is... In particular, adjacent high-pass filters can be designed to have their amplitude response and / or phase response above the stopband frequency. The above are all close to each other, causing the subtraction operation to lead to destructive interference. In summary, the first... The passband of each bandpass filter is Since the bandpass filter is defined by the highpass filter, the lower cutoff frequency of the subsequent bandpass filter corresponds to the upper cutoff frequency of the bandpass filter, i.e. .

[0049] The threshold frequency of the high-pass filter and the corresponding passband of the band-pass filter can be advantageously evenly distributed across a psychoacoustic-driven frequency scale (e.g., the Barker scale). This allows the input signal to be processed within a frequency band that simulates human hearing. However, the choice of the threshold frequency depends on the specific application scenario. This invention is not limited to the specific selection of the threshold frequency.

[0050] A key advantage of the topology of this invention is that when the unity gain is When, the overall transfer function simplifies to In other words, the total transfer function is entirely defined by the first high-pass filter. Therefore, the total transfer function is smooth within the passband and has a small group delay.

[0051] For example, a high-pass filter can be designed using an optimization method that employs a cost function based on the amplitude response of the bandpass filters in the passband and stopband, and the sum of the costs of all bandpass filters. The filter can be implemented as an FIR or IIR filter. These filters can advantageously be minimum-phase and have a nonlinear phase response. A filter is considered optimal if its zeros (i.e., the zeros of the numerator polynomial of its filter transfer function) lie within the unit circle, or have... If the amplitude of the filter is zero, then the filter can be called minimum-phase. This definition applies to FIR and IIR filters that are not implemented as all-pole filters (i.e., filters whose transfer function only contains denominator coefficients and possible gain factors). By definition, an all-pole filter is minimum-phase.

[0052] A high-pass filter can also be advantageously optimized to improve the overall transfer function. By following the expected amplitude and phase responses, the filter bank can implicitly perform frequency weighting or equalization, for example.

[0053] Figure 5 Showing based on Figure 3 and Figure 4 The resulting amplitude response 22 of the exemplary bandpass filter for the high-pass filter is shown. Overall, Figure 5 The diagram shows the first amplitude response 22a, the second amplitude response 22b, the third amplitude response 22c, and the fourth amplitude response 22d of the corresponding bandpass filter. At this time, the overall transfer function... Corresponding to Figure 3 and Figure 4 The high-pass filter is represented by a solid line in the image.

[0054] like Figure 6 As shown in the example, a low-pass filter 11 (also known as...) can also be used. The method of the present invention is implemented by replacing the high-pass filter. Figure 6 Exemplary examples show a first low-pass filter 11a, a second low-pass filter 11b, and a third low-pass filter 11c. A low-pass filter 11c. The aforementioned basic principle, based on filter characteristics and the bandpass characteristics formed by destructive interference generated through subtraction, still holds, but... Figure 2 Compared to the implementation shown, some adjustments are needed. Specifically, the subtraction method needs to be adjusted so that... .

[0055] In this case, when the unity gain is When, the overall transfer function simplifies to .

[0056] Figure 7 Another embodiment of the method of the present invention is shown, wherein... Figure 2 Compared to the illustrated embodiment, the high-pass signal has been removed. Instead of explicit subtraction, a modified weighting factor was used. This implicitly yields the difference during the weighting process. For a high-pass filter-based configuration, the modified weighting factor can be calculated as follows: .

[0057] This rule can be determined by rearranging the equations of the overall transfer function; there are no restrictions here. The generality within the domain ignores the time-varying nature of the weighting factor: Figure 8 Another embodiment of the method of the present invention is shown, wherein Figure 2 The first high-pass filter 10a shown is replaced by a short circuit. Accordingly, Figure 6 The last low-pass filter 11c in the illustrated configuration can also be replaced by a short circuit. Therefore, the transfer function of the unity-gain structure is... Therefore, there is no inherent delay. It should be noted that for different weighting factors... The overall transfer function will not simplify accordingly, which may lead to an increase in delay. However, in these cases, the input-to-output delay remains very small due to the filter structure.

[0058] Figure 9An embodiment of the device 50 of the present invention is shown, the device including a acquisition unit 30 implemented as a microphone, a filter bank 31, an output filter 32, and an output unit 33 implemented as a speaker.

[0059] Figure 10 Another embodiment of the device 50 of the present invention is shown, which includes two sensors 30, two input filters 34, a filter bank 31, and a speaker 33.

[0060] These configurations are found, for example, in modern headphones, where a microphone records ambient sound, and a pre-processing / post-processing unit (also known as input and output filters) filters the microphone signal to achieve, for example, active noise cancellation (for reducing the volume of ambient sound) or an ambient mode (for natural communication with the surrounding environment), and then the filtered audio signal is played back through the headphones' speakers. The method of the present invention can be applied, for example, to scenarios where microphone noise or wind noise is removed from the microphone signal, speech optimization is performed, or the microphone signal is compressed to reduce the level of a loud signal before playback.

[0061] at last, Figure 11 An in-ear headphone 60 is exemplarily shown, equipped with multiple external microphones 40, an internal microphone 41, a vibration sensor 42, a speaker 33, and a computing unit 43. Furthermore, Figure 12 also shows an earplug 44, an ear canal 45, and a tympanic membrane 46. The computing unit 43 is designed to perform the method steps of the method of the present invention. In particular, the computing unit 43 is designed to provide a filtering function and apply it to the audio signal in the time domain. During this process, different preprocessing and post-processing steps, such as filtering, enhancement, compression, or limiting, can be selectively performed on the sensor signal. These steps can be implemented according to methods described in the prior art. Subsequently, the single-channel signal processed based on the method of the present invention is played back by the speaker 33. Multiple instances of the method of the present invention can also be used to process different sensor signals, which are then combined and sent to an output unit.

[0062] Weighting factors can be calculated using any sensor signal. According to the spirit of the invention, it is preferable to incorporate the bandpass signal into the calculation of the relevant weighting factor. In particular, the weighting factor... It can be a bandpass signal The function of the weighting factor can also depend on other bandpass signals. While the filter bank, weighting factor application, and summation can be handled at a first frequency, it is preferable to design the calculation of the weighting factor at a second frequency. In this case, the second frequency is preferably lower than the first frequency. For this purpose, the bandpass signal can be converted using a sampling rate converter. High-pass / low-pass signals or even the input signal The signal is then transferred to a second processing step. This second processing step can then simulate a portion or even the entire filter bank to provide a bandpass signal at a second frequency.

[0063] Since weighting factors typically change slowly, the second sampling rate is advantageously smaller than the first sampling rate to save computational complexity. However, when designing the sampling rate, the Nyquist frequency should be considered to ensure that the signal is transmitted with the appropriate bandwidth without information loss. The weighting factors calculated at the second sampling rate do not necessarily need to be matched to the first sampling rate through sampling rate conversion.

[0064] For example, weighting factors can be calculated to allow speech components in each frequency band to pass while suppressing all interfering noise, such as wind noise, ambient noise, or microphone noise, thereby improving speech quality or speech intelligibility. Furthermore, weighting factors can be calculated to suppress loud signal components in each frequency band, for example, to protect the hearing of users in headphone ambient mode or hearing protection devices with communication capabilities.

[0065] Similarly, the method of the present invention can also be applied to other audio sources, such as music, telephone calls, computer games, and movies. The method of the present invention can be used to calibrate or personalize active noise cancellation in music playback systems, ambient modes, and headphones. Furthermore, the method can be used, for example, for hearing loss compensation in hearing aid applications. The method can also be used for the calibration of audio devices (including frequency-related calibration).

[0066] Explanation of reference numerals in the attached figures 10 High-pass filter 10a First High-Pass Filter 10b Second High-Pass Filter 10c One high-pass filter 11 Low-pass filter 11a First Low-Pass Filter 11b Second Low-Pass Filter 11c One low-pass filter 20. Amplitude response of a high-pass filter 20a First Amplitude Response 20b Second Amplitude Response 20c Third Amplitude Response 20d Fourth Amplitude Response 21. Phase Response of a High-Pass Filter 21a First Phase Response 21b Second Phase Response 21c Third Phase Response 21d fourth phase response 22. Amplitude response of the obtained bandpass filter 22a First Amplitude Response 22b Second Amplitude Response 22c Third Amplitude Response 22d Fourth Amplitude Response 30 acquisition units 31 Filter Bank 32 Output Filter 33 Output Unit 34 Input Filter 40 External Microphone 41 Internal microphone 42 Vibration Sensor 43 Computing Units 44 Earplugs 45 Ear canal 46. ​​Tympanic membrane 50. The apparatus of the present invention 60 In-ear headphones 100 The method of the present invention 110 First Method Steps 120 Second Method Steps 130 Third Method Steps 140 Fourth Method Steps

Claims

1. An audio signal processing method (100), comprising the following steps: - Provide (110) audio signal ; - Provides (120) filter functions; - Apply the filtering function (130) to the audio signal in the time domain. To provide filtered audio signals ;as well as - Output the filtered audio signal described in (140) ;in - The provision of the filtering function depends on A single filter function, wherein ;and - The single filtering function includes at least two high-pass filtering functions or at least two low-pass filtering functions, and a band-pass filtering function is obtained based on the at least two high-pass filtering functions or the at least two low-pass filtering functions.

2. The method (100) according to claim 1, characterized in that, The provision of the filter function (120) includes subtracting the first single filter function from the second single filter function, wherein the first single filter function and the second single filter function are respectively implemented as filter functions of a high-pass filter or filter functions of a low-pass filter.

3. The method (100) according to claim 2, characterized in that, - Using two high-pass filters, the high-pass filter transfer function of the two high-pass filters. Designed such that its amplitude response is at the first limiting frequency The above have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, particularly preferably a maximum deviation of 3% or a maximum deviation of 1%, or - Using two low-pass filters, the low-pass filter transfer functions of the two low-pass filters. Designed such that its amplitude response is at the second limiting frequency The following have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, and particularly preferably a maximum deviation of 3% or 1%.

4. The method (100) according to any one of claims 2 or 3, characterized in that, - Using two high-pass filters, the high-pass filter transfer function of the two high-pass filters. Designed such that its phase response is at the first limiting frequency The above have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, particularly preferably a maximum deviation of 3% or a maximum deviation of 1%, or - Using two low-pass filters, the low-pass filter transfer functions of the two low-pass filters. Designed such that its phase response is at the second limiting frequency The following have a maximum deviation of 10% from each other, preferably a maximum deviation of 5%, and particularly preferably a maximum deviation of 3% or 1%.

5. The method (100) according to any one of claims 1 to 4, characterized in that, The single filter is implemented as a finite impulse response (FIR) filter or an infinite impulse response (IIR) filter.

6. The method (100) according to any one of claims 1 to 5, characterized in that, The single filter has nonlinear phase characteristics.

7. The method (100) according to any one of claims 1 to 6, characterized in that, The filtered audio signal Determine according to the following formula: in Let represent the weighting factors, and . Determine according to the following formula: ， in Describe the high-pass filter respectively Filtered audio signal And the high-pass filter Each has a limiting frequency ,in .

8. The method (100) according to any one of claims 1 to 6, characterized in that, The filtered audio signal Determine according to the following formula: in Let represent the weighting factors, and . Determine according to the following formula: ， in Describe the low-pass filter respectively Filtered audio signal The low-pass filter mentioned above Each has a limiting frequency ,in .

9. The method (100) according to any one of claims 1 to 6, characterized in that, The filtered audio signal Determine according to the following formula: in Describe the high-pass filter respectively Filtered audio signal The high-pass filter mentioned above Each has a limiting frequency ,in ,and These represent the modified gain factors, which are determined according to the following formula: 。 10. The method (100) according to any one of claims 1 to 6, characterized in that, The filtered audio signal Determine according to the following formula: in Describe the low-pass filter respectively Filtered audio signal The low-pass filter mentioned above Each has a limiting frequency ,in ,and These represent the modified gain factors, which are determined according to the following formula: 。 11. The method (100) according to any one of claims 1 to 10, characterized in that, At least one single filter function is designed as a delta function.

12. The method (100) according to any one of claims 8 to 11, characterized in that, - The weighting factor With the audio signal filtered by the high-pass filter or the low-pass filter Calculated signal The multiplication or the modified weighting factor The audio signal filtered by the high-pass filter or low-pass filter The multiplication is based on the first frequency Performed; - at the first frequency For the weighting factor According to the filtered audio signal Calculated signal Summing the products, or applying the modified weighting factors. The audio signal filtered by the high-pass filter or low-pass filter Summing the products; as well as - at the second frequency Calculate the weighting factor Or the modified weighting factor ;in - Second frequency Below the first frequency .

13. The method (100) according to any one of claims 1 to 12, characterized in that, The filtering function is applied to the audio signal in the time domain. This includes applying convolution operations to the audio signal. and the impulse response of high-pass or low-pass filters. .

14. An audio signal processing apparatus (50), comprising: - Used for acquiring audio signals The acquisition unit (30); - Used to calculate the filtering function and process the audio signal The computational unit (43); where, The computing unit (43) is designed to apply the filtering function to the audio signal in the time domain. And calculate the filtered audio signal. ; - Used to output the filtered audio signal 'output unit (33); Its features are, - The computing unit (43) is designed to be based on The filter function is obtained by a single filter function, wherein ;as well as - Obtain the bandpass filter function from at least two single filter functions, including two high-pass filter functions or two low-pass filter functions.

15. The apparatus (50) according to claim 14, characterized in that, The computing unit (43) is designed to calculate a bandpass filter function by subtracting a first single filter function from a second single filter function, wherein the first single filter function and the second single filter function are respectively implemented as a filter function of a high-pass filter or a filter function of a low-pass filter.

Images