Method for detecting acoustic feedback

Abstract

The invention relates to a method for detecting acoustic feedback in a vehicle audio system, the vehicle audio system comprising at least one microphone and one loudspeaker, the method comprising the steps of: calculating (201) an acoustic sensitivity indicator for each of a plurality of consecutive audio frames captured by the microphone; detecting (202) an acoustic feedback when the indicator monotonically increases within a predetermined length of time; and applying (204) the resolution strategy when the acoustic feedback is detected.

Description

Field of the Invention

[0001] The present invention relates to the field of hands-free communication systems, and more particularly, to the detection and attenuation of acoustic feedback in such communication systems. Background Art

[0002] So-called "hands-free" communication or control devices are very common in modern vehicles. Specifically, these systems allow the driver to make and continue phone calls without removing their hands from the steering wheel. Such a system typically includes at least one microphone and one speaker, which are associated with a multimedia processing device configured to interpret voice commands generated by the vehicle's occupants and transmit audible notifications, such as command confirmations or incoming call notifications. In such a system, the microphone continuously captures and processes audio signals in order to detect voice commands therein or transmit voice signals to the called / calling party during a phone call. Consequently, the voice notifications or phone calls presented by the speaker are also captured by the microphone, which may generate feedback effects such as echoes or cause misinterpretation of voice commands.

[0003] To avoid this, acoustic echo cancellation techniques have been developed, in which the microphone gain is reduced when the speaker presents a signal. Such systems are not suitable for phone calls because the voice of the other party cannot be heard and a conversation cannot be held simultaneously, which limits the fluency of the conversation.

[0004] To improve the user experience, ECNR or AEC techniques are employed (ECNR stands for Echo Cancellation and Noise Reduction, and AEC stands for Acoustic Echo Cancellation), in which the correlation between the signal captured by the microphone and the signal presented by the speaker is sought in order to subtract the presented signal from the captured signal when appropriate.

[0005] However, echo cancellers are complex devices and cannot avoid malfunctioning. For example, an ECNR system may fail to converge due to a lack of system resources, resulting in the occurrence of echoes. Software or hardware malfunctions may also cause the sound captured by the microphone to be reproduced by the speaker within the range of the microphone's location. For example, such a malfunction may occur in a remote terminal, and then echoes may occur in the local terminal during audio communication between the two terminals. In other cases, misrouting of the audio stream in the terminal may cause echoes, such as when the signal captured by the microphone is directly reinjected into the signal generated by the speaker within the range of that microphone instead of being transmitted to the terminal of the called / calling party. In some cases, the feedback effect of the output on the input generates ripples, the amplitude of which gradually increases until it reaches the limit of the audio system. This effect is called acoustic feedback and is typically manifested as an unpleasant whistling sound, the volume of which continuously increases until it reaches the maximum power of the amplifier used.

[0006] In addition to the discomfort caused, the occurrence of this effect in a moving motor vehicle can also distract the driver and lead to safety problems.

[0007] Therefore, it is necessary to ensure that this effect does not occur in the vehicle being driven. Since software failures are inherently unpredictable, such a system for preventing acoustic feedback must operate continuously and use minimal resources, as the functions of the vehicle computer are limited.

[0008] Therefore, a simple technique is needed to allow the detection and stopping of acoustic feedback in a vehicle. Summary of the Invention

[0009] To this end, a method for detecting acoustic feedback in a vehicle audio system is provided, the vehicle audio system including at least one microphone and one speaker, the method comprising the following steps:

[0010] - calculating an acoustic acuity index for a plurality of consecutive audio frames captured by the microphone,

[0011] - detecting acoustic feedback when the index increases monotonically over a predetermined time length,

[0012] - applying a resolution strategy when acoustic feedback is detected.

[0013] The fundamental frequency of the sound generated by the signal emitted by the speaker and fed back via the microphone depends on various parameters (such as the acoustic properties of the generation location, the distance between the transmitter and the receiver, and the directivity of the receiver), and cannot be used alone to characterize acoustic feedback. Additionally, using echo cancellers and noise reducers in an audio system presents special detection problems because acoustic feedback is non-linear and in the form of ripples with increasing amplitude. The acuity index provides information about the ratio of the energy of the high-frequency signal to its total energy. When such a ratio stabilizes at a value higher than a specific threshold, it may be characteristic of acoustic feedback.

[0014] The acoustic acuity index can be calculated for each captured signal frame (e.g., a 10-millisecond frame). Thus, when, within a certain number of consecutive frames, e.g., 40 frames (or within a certain time length, e.g., 400 ms), the index increases monotonically, acoustic feedback is detected and corrective measures are taken.

[0015] In a particular embodiment, the calculation of the acuity index includes calculating the energy spectral density of the signal captured by the microphone.

[0016] Determining the energy or power spectral density based on the captured signal enables the determination of the frequency band in which the energy of the signal is concentrated. Thus, the monotonicity of the acuity index can be estimated based on the frequencies at which acoustic feedback may occur.

[0017] According to a specific embodiment, the metric is calculated based on frequencies greater than 2 kHz.

[0018] In this way, the method makes it possible to eliminate some noise from the analysis of acoustic feedback. By focusing the analysis on high frequencies, this increases the relevance of the metric. Of course, other frequency values can be considered without modifying the present invention.

[0019] According to a specific embodiment, the solution strategy includes attenuating the signal captured by the microphone, the attenuation level being proportional to the length of time since the acoustic feedback was detected and the value of the acuity metric.

[0020] Therefore, it is proposed to attenuate the signal captured by the microphone in the case of acoustic feedback. The attenuation is variable such that when acoustic feedback is still detected after a certain time, the captured signal is further attenuated.

[0021] Conversely, when acoustic feedback is no longer detected after attenuation has been applied in a certain number of consecutive frames, the attenuation level is gradually reduced until the normal situation is restored.

[0022] The attenuation level is further adjusted according to the value of the acuity metric such that stronger attenuation is applied when the acuity metric is high.

[0023] According to a specific embodiment, the solution strategy is selected from the following strategies:

[0024] - Lower the amplifier volume,

[0025] - Lower the microphone gain,

[0026] - Apply a band - pass filter,

[0027] - Restart the audio system.

[0028] Lowering the output level of the amplifier or lowering the gain of the microphone limits the feedback of the output to the input and thus stops the acoustic feedback. The application of a band - pass filter configured to eliminate high frequencies can limit discomfort. Resetting the audio system (e.g., by restarting the device to restore a stable state) can stop the acoustic feedback (in the case where the acoustic feedback is caused by a software malfunction).

[0029] According to a specific embodiment, the method makes the calculation of the acoustic acuity metric take into account the influence produced by the absolute loudness of the captured signal.

[0030] Taking loudness into account in the calculation of the acuity metric can highlight the possible differences between different sounds, thus enabling better discrimination. Such a result is particularly suitable for determining whether a sound has the desired characteristics. In the present case, the reliability of acoustic feedback detection is thus improved, and when the howling caused by acoustic feedback is not uncomfortable for a person, the solution strategy is not triggered. In other words, it is proposed to apply the correction strategy only when the acoustic feedback becomes unpleasant for the user.

[0031] Furthermore, taking loudness into account makes it possible to process only the acoustic feedback that is unpleasant for the user, and thus only the loud acoustic feedback.

[0032] According to another aspect, the invention relates to a device for detecting acoustic feedback in a vehicle audio system, the vehicle audio system comprising at least one microphone and one loudspeaker, the device comprising a processor and a memory in which program instructions are stored, the program instructions being configured to, when executed by the processor, carry out the following steps:

[0033] - calculating an acoustic acuity metric for a plurality of consecutive audio frames captured by the microphone,

[0034] - detecting acoustic feedback when the metric increases monotonically over a predetermined length of time,

[0035] - applying a solution strategy when acoustic feedback is detected.

[0036] The invention also relates to a hands-free communication system comprising the device as described above, and to a vehicle comprising such a system.

[0037] Finally, the invention relates to an information medium comprising computer program instructions configured to carry out the steps of the detection method as described above when the instructions are executed by a processor.

[0038] For example, the information medium can be a non-transitory information medium such as a hard disk, a flash memory or an optical disc.

[0039] The information medium can be any entity or device capable of storing instructions. For example, the medium can include storage means such as ROM (read-only memory), RAM (random access memory), PROM (programmable read-only memory), EPROM (erasable programmable read-only memory), CD ROM or magnetic recording means (e.g., a hard disk).

[0040] On the other hand, the information medium can be a transmissible medium that can be transmitted via radio or by other means via a cable or an optical fiber, such as an electrical signal or an optical signal.

[0041] Alternatively, the information medium may be an integrated circuit containing the program, the circuit being adapted to execute or for executing the method under discussion.

[0042] Each of the above-mentioned embodiments or features can be added to the steps of the detection method independently or in combination with each other. These vehicles, communication systems, devices and information media have at least advantages similar to those conferred by the methods they are involved in. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Other features and advantages of the present invention will become more apparent by reading the following description. This description is purely illustrative and will be read with reference to the accompanying drawings, in which:

[0044] Figure 1 Figure 1 shows the architecture of a hands-free audio communication device that can be integrated into a vehicle's communication system,

[0045] Figure 2 Figure 2 is a flowchart showing the main steps of a method for detecting acoustic feedback according to a particular embodiment, and

[0046] Figure 3 Figure 3 shows the architecture of a device suitable for implementing the detection method according to a particular embodiment. DETAILED DESCRIPTION

[0047] Figure 1 The architecture of a so-called "hands-free" communication device 100 that can be integrated into a vehicle is shown in a simplified manner. The system 100 is configured to allow simultaneous acquisition Figure 1 of a first audio signal denoted "nearIN" and reproduction Figure 1 of a second audio signal denoted "nearOUT". For this purpose, the device 100 includes at least one microphone 101 and a speaker 102.

[0048] The signal captured by the microphone 101 is digitized by an analog-to-digital converter 104 and processed by an ECNR module 103, and then transmitted to another device, for example, an interpreter of voice commands or a device that enables speaking to the called / calling party, in the form of a signal "farOUT".

[0049] In parallel, the signal "farIN" transmitted by a voice notification system or a telecommunication device is processed by the ECNR module 103, and then converted to an analog signal by a converter 105 and presented by the speaker 102.

[0050] ​​​​​​In this so-called "hands-free" configuration, the microphone 101 captures not only the user's voice signal but also the signal presented by the speaker 102, and thus the device 100 is equipped with an ECNR module 103.

[0051] The ECNR module 103 is configured to find the correlation between the signal "nearOUT" and the signal "nearIN". More precisely, the module 103 searches for the signal "nearOUT" presented by the speaker 102 in the signal captured by the microphone 101 in order to cancel or reduce it. Simply put, echo cancellation is the identification of the signal when the signal initially emitted by the speaker 102 reappears in the signal captured by the microphone 101 with a certain delay. Once the echo is identified, it can be cleared by subtracting the echo from the captured signal, for example, by adding the signal "nearOUT" with the phase reversed to the signal "nearIN". Such an ECNR module is known in the art and its operation will not be discussed in detail.

[0052] Unfortunately, as already pointed out, such a device may not function properly, and therefore, establishing a second feedback loop may result in the generation of acoustic feedback. For example, such a failure may involve a component that feeds "farOUT" back to "nearOUT", and the ECNR component has no effect on this. Such acoustic feedback may not only be uncomfortable for the occupants of the vehicle, but most importantly, it will distract the driver, thus causing safety problems.

[0053] Now reference will be made to Figure 2 Describe a specific embodiment of a method for detecting acoustic feedback.

[0054] The method includes a first step 200 of capturing and digitizing the audio signal delivered by the microphone 101. The microphone 101 is placed, for example, in the passenger compartment of the vehicle to capture voice signals, such as voice commands or phone conversations.

[0055] In step 201, the digitized signal is processed to periodically calculate an acoustic acuity metric. For example, the metric is calculated on consecutive frames of 10 milliseconds to obtain a plurality of metric values, thereby allowing the study of its variation over time.

[0056] Acuity is a psychoacoustic parameter that quantifies the auditory sensation corresponding to the sensation of a sharp sound perceived as bright or sharp. It corresponds to the ratio of the amount of high-frequency energy to the amount of total energy and is measured in acum (1 acum corresponds to a narrowband noise of 1 kHz with a bandwidth less than 150 Hz and a sound level equal to 60 dB). The calculation of acuity is not standardized and can be determined in various ways.

[0057] In "Sharpness as an attribute of the timbre of steady sounds", Acustica 30(3), 159–172, von Bismark proposed in 1974 a calculation method based on the distribution of the variation of specific loudness with pitch. This method did not take into account the influence of absolute loudness on sharpness.

[0058] In "Sensory euphony as a function of auditory sensations", Acustica 58(5), 282–290, W. Aures proposed in 1985 a modified version of von Bismark's method in order to take into account the influence of loudness.

[0059] This algorithm normalizes the specific loudness spectrum by the total loudness and weights this spectrum as a function of frequency. The algorithm returns the frequency-weighted result as the specific sharpness relative to the critical band rate, and then integrates the specific sharpness to measure sharpness. High-frequency components in the signal usually result in higher sharpness measurements.

[0060] In a specific embodiment, the method proposed by Aures is used to calculate the sharpness metric, that is, the influence due to loudness is taken into account. Loudness expresses the sensation of the volume of a sound as perceived by humans. This parameter is defined in such a way that the amplitude of a sine signal with a frequency of 1 kHz and a pressure level of 40 dB is 1 sone.

[0061] In a specific embodiment, the metric is determined by calculating the energy spectrum of the captured signal. For example, it is proposed to calculate the power spectral density of the captured signal in order to obtain the frequency distribution of the power of the signal (depending on its constituent frequencies). Then, the metric is determined based on the power of the high-frequency (e.g., frequencies above 2 kHz) signal.

[0062] In step 202, the value of the metric calculated for the current digitized signal frame is compared with the value of the metric calculated for the previous signal frame to determine whether the value of the metric is increasing monotonically. For this purpose, a binary monotonicity indicator can be used, the value of which is initially set to "0". When the metric calculated for frame n is greater than or equal to the value of the metric calculated for frame n - 1, the monotonicity indicator is set to "1", while when it is observed that the metric calculated for frame n is less than the value of the metric calculated for frame n - 1, the monotonicity indicator is set to "0".

[0063] In step 203, the number of consecutive frames in which the monotonicity indicator is set to the value "1" is counted, and the counted number of frames is compared with a threshold to determine whether acoustic feedback is being generated. In a particular embodiment, the threshold is set to 40 frames. In other words, when the value of the acuity metric increases monotonically over a predetermined length of time (e.g., 400 milliseconds (40 frames, 10 ms per frame)), it is determined that acoustic feedback has been generated. The acoustic feedback thus detected is considered to remain unchanged as long as the value of the metric does not decrease.

[0064] Finally, the method includes step 204, in which a resolution strategy is implemented when acoustic feedback is detected.

[0065] According to a particular embodiment, the resolution strategy includes reducing the gain of the microphone or the output level of the speaker. For example, the gain and / or the output level are gradually reduced until the acuity metric drops below a specific threshold.

[0066] According to a particular embodiment, the resolution strategy includes resetting or restarting the audio system.

[0067] In a particular embodiment, the resolution strategy includes attenuating the presented audio signal as long as acoustic feedback is considered to be detected. For example, an attenuation of 26 dB can be applied. It is also contemplated to apply an attenuation value proportional to the value of the calculated acuity metric, such that when the acuity metric decreases, the attenuation also decreases.

[0068] According to a particular embodiment, when no acoustic feedback is detected within a determined length of time (e.g., 10 seconds), no attenuation is applied to the signal intended to be presented. Of course, various lengths of time can be contemplated without modifying the present invention.

[0069] Figure 3 A device 300 for detecting acoustic feedback according to a particular embodiment is shown.

[0070] Device 300 includes a storage space 302 such as a memory MEM, and a processing unit 301 equipped with a processor PROC, for example. The processing unit can be controlled by a program 303 (e.g., a computer program PGR) to implement the method for detecting acoustic feedback described with reference to Figure 2 and in particular to implement the following steps: calculating an acoustic acuity metric for each of a plurality of consecutive audio frames captured by a microphone; detecting acoustic feedback when the metric increases monotonically over a predetermined length of time; and applying a resolution strategy when acoustic feedback is detected.

[0071] At initialization, the instructions of the computer program 303 are loaded into RAM (Random Access Memory), for example, and then executed by the processor of the processing unit 301. The processor of the processing unit 301 implements the steps of the detection method according to the instructions of the computer program 303.

[0072] To this end, in addition to the memory and the processor, the device further includes a microphone 305 and a speaker 304, which are respectively coupled to an analog-to-digital converter 306 and a digital-to-analog converter (DAC) 307. These analog-to-digital converters are on the one hand intended for digitizing the signals captured by the microphone 304, and on the other hand intended for generating analog signals from digital audio signals. The converters 306 and 307 are connected to an ECNR module (ECNR 308), which is configured to estimate the delay and transfer function between the signal reproduced by the speaker 304 and the signal captured by the microphone 305, so as to subtract the presented signal from the captured signal.

[0073] The device 300 further includes a communication module 309, such as a wireless communication interface of the 2G, 3G, 4G, 5G, Wi-Fi or Bluetooth type, which is configured to receive digital audio signals transmitted by a device (for example, a voice synthesis device of a virtual assistant) or transmitted by the communication system of the called / calling party during a telephone call, and transmit the digitized voice signal processed by the module ECNR 308 to the called / calling party or a voice-activated control device.

[0074] The device 300 further includes a module 310 for calculating an acuity metric. The module 310 is configured by computer program instructions, for example, to obtain digitized audio signal frames (for example, frames of 10 milliseconds) captured by the microphone 305, apply an algorithm allowing the calculation of the power spectral density, and select frequencies above a threshold (for example, frequencies above 2 kHz). This acuity metric characterizes the power of the audio signal captured in the selected frequency band.

[0075] According to a specific embodiment, the module 310 is configured by computer program instructions to calculate the acuity metric according to the method proposed by Aures as described above.

[0076] The device further includes a module 311, which is configured to compare the values of the acuity metrics calculated for at least two consecutively captured frames (for example, frames of 10 milliseconds) and determine whether the values of the metric are increasing monotonically. The module 311 is further configured to determine that acoustic feedback is being generated when it is determined that the acuity metric increases monotonically within a predetermined time length (for example, 400 milliseconds, i.e., 40 frames, each frame being 10 milliseconds).

[0077] The device finally includes a module 312 for applying a solution strategy. The module 312 can be implemented by computer program instructions configured to: as long as feedback is generated, reduce the output level of the signal delivered to the speaker 304 and / or reduce the gain of the microphone 305; apply a band-pass filter, such as a low-pass filter that allows high frequencies (e.g., frequencies greater than 2 kHz) to be eliminated, or actually restart the audio system.

[0078] According to a particular embodiment, the device 300 is integrated into a hands-free communication system of a vehicle.

Claims

1. A method for detecting acoustic feedback in a vehicle audio system, the vehicle audio system including at least one microphone and one speaker, the method comprises the following steps: - calculating (201) an acoustic acuity index for each of a plurality of consecutive audio frames captured (200) by the microphone, - detecting (202) acoustic feedback when the index monotonically increases over a predetermined time length, - applying (204) a resolution strategy when acoustic feedback is detected.

2. The method according to claim 1, wherein, the calculation of the acuity index includes calculating the energy spectral density of the signal captured by the microphone.

3. The method according to claim 2, wherein, the index is calculated based on frequencies greater than 2 kHz.

4. The method according to any one of the preceding claims, wherein, the resolution strategy includes attenuating the signal captured by the microphone, the attenuation level being proportional to the time length since the acoustic feedback was detected and the value of the acuity index.

5. The method according to any one of claims 1 to 3, wherein, the resolution strategy is selected from the following strategies: - reducing the amplifier volume, - reducing the microphone gain, - applying a band-pass filter, - restarting the audio system.

6. The method according to any one of the preceding claims, wherein, the calculation of the acoustic acuity index takes into account the influence generated by the absolute loudness of the captured signal.

7. A device for detecting acoustic feedback in a vehicle audio system, the vehicle audio system including at least one microphone and one speaker, the device includes a processor and a memory storing program instructions therein, the program instructions being configured to implement the following steps when executed by the processor: - calculating the acoustic acuity index of a plurality of consecutive audio frames captured by the microphone, - detecting acoustic feedback when the index monotonically increases over a predetermined time length, - applying a resolution strategy when acoustic feedback is detected.

8. A hands-free communication system, including the device according to claim 7.

9. A vehicle, including the hands-free communication system according to claim 8.

Images