DE102021208922A1 - Method and system for generating noises in an interior based on extracted and classified real noise sources and for specific target noises acoustically transparent vehicle comprising such a system - Google Patents

Abstract

Method for generating noises in an interior based on extracted and classified real noise sources (30), comprising the steps of: detecting a sound field in an exterior of a vehicle (10) by measuring sound pressures using a plurality of first sound transducers (1 ) (V1); Angular scanning of the sound field by means of beamforming signal processing, with sound pressures measured at an angle to be scanned in relation to one of the first sound transducers (1) being extracted from the sound field and thus the noise sources (30) being isolated from one another (V2); Classification of isolated noise sources (30) (V3); Further processing of signals obtained from the respective isolated noise sources (30) depending on the classification, the signals being filtered using head-related or outer-ear transfer functions (V4); Playback of the filtered signals in the interior by means of second sound transducers (2) that can be arranged in the interior and generation of corresponding noises in the interior, the noises being perceived as coming from the respective directions of the real noise sources (30) based on the head-related or outer ear transfer functions (V5).

Description

The invention relates to a method and a system for generating noises in an interior based on extracted and classified real noise sources. Furthermore, the invention relates to a vehicle that is acoustically transparent for specific target noises, comprising a system according to the invention.

Acoustic cameras are known from the prior art as industrial measurement systems for localizing acoustic emissions using microphone arrays, see for example URL https://www.acoustic-camera.com/de/produkte.html.

Furthermore, vehicle-external microphones are known for acoustic signal perception, for example perception of siren noises from emergency vehicles. The state of the art is, for example, in DE 10 2019 219 525 B3 , DE10 2019 213 695 B3 and DE 10 2019 206 331 A1 disclosed. The signals or noises can be classified, e.g. into accident noises, e.g. by means of artificial neural networks, as in DE 10 2018 200 878 B3 disclosed.

Methods for locating noise sources are also known, for example transformation of the power phase with controlled reaction, also called steered-response power phase transform, abbreviated SRP-PHAT, see for example URL https://en.wikipedia.org/wiki/Steered-Response_Power_Phase_Transform .

Furthermore, 3D audio effects are known that manipulate the sound generated by stereo loudspeakers, surround sound loudspeakers, loudspeaker arrays or headphones using head-related or outer-ear transfer functions, also called head-related transfer functions, abbreviated HRTF, see for example URL https:/ /en.wikipedia.org/wiki/3D_audio_effect.

In the automotive sector, it is problematic that the damping of sound from outside to inside caused by the closed design of vehicles, which is desired in particular in the premium segment in order to create a quieter atmosphere in the vehicle, is not selective. All ambient noise is equally shielded. That's not always desirable. Certain signals are important for the driver and/or helpful for the smooth flow of road traffic, such as bicycle bells, calls for help, emergency vehicle sirens, for example a siren. In particular, if the driver of the vehicle is distracted, for example by the navigation system, multimedia system, telephone call, stress, and overlooks a cyclist or pedestrian, he cannot hear them either because of the vehicle insulation.

As a cyclist or pedestrian, you have a better feeling for your surroundings on the road because you can, for example, perceive and locate approaching vehicles through their rolling noise, engine noise and wind noise. The same applies to all other noises such as conversations, footsteps, bicycle bells, which pedestrians can hear better than drivers.

For example, reverse parking in front of a wall is much easier in a convertible with an open top, since you can better hear the reflections of the sound of your own vehicle on the wall and thus better assess the distance.

Permanently opening the window to circumvent the damping and thus better acoustic perception of the vehicle's surroundings is not an option, since on the one hand the air conditioning would be obsolete and on the other hand all noises from the outside could penetrate into the interior, with which one unselected all noises of the exposed to the outside world, not just the relevant noises.

The object of the invention was how to provide a vehicle that is acoustically transparent to specific target noises.

The objects of claims 1, 9 and 10 each solve this problem.

• • Erfassung eines Schallfeldes in einem Außenraum eines Fahrzeuges durch Messen von Schalldrücken mittels mehreren an dem Fahrzeug anordbaren ersten Schallwandlern;
• • Winkelabtastung des Schallfeldes mittels einer beamforming-Signalverarbeitung, wobei jeweils in einem mit Bezug auf einen der ersten Schallwandler abzutastenden Winkel gemessene Schalldrücke aus dem Schallfeld extrahiert werden und damit die Geräuschquellen voneinander isoliert werden;
• • Klassifizierung der isolierten Geräuschquellen;
• • Weiterverarbeitung von aus den jeweiligen isolierten Geräuschquellen erhaltenen Signale in Abhängigkeit der Klassifizierung, wobei die Signale mittels kopfbezogenen oder Außenohr-Übertragungsfunktionen gefiltert werden;
• • Wiedergabe der gefilterten Signale in dem Innenraum mittels in dem Innenraum anordbaren zweiten Schallwandlern und Erzeugen von entsprechenden Geräuschen in dem Innenraum, wobei basierend auf den kopfbezogenen oder Außenohr-Übertragungsfunktionen die Geräusche als aus jeweiligen Richtungen der realen Geräuschquellen kommend wahrgenommen werden

According to one aspect, the invention provides a method for generating noise in an interior based on extracted and classified real noise sources. The procedure includes the steps

• • Detection of a sound field in an exterior space of a vehicle by measuring sound pressures using a plurality of first sound transducers that can be arranged on the vehicle;
• • Angular scanning of the sound field by means of beamforming signal processing, with sound pressures measured in an angle to be scanned in relation to one of the first sound transducers being extracted from the sound field and thus the noise sources being isolated from one another;
• • classification of isolated noise sources;
• • Further processing of signals obtained from the respective isolated noise sources in Dependence of classification, where the signals are filtered using head-related or outer-ear transfer functions;
• • Playback of the filtered signals in the interior by means of second sound transducers that can be arranged in the interior and generation of corresponding noises in the interior, the noises being perceived as coming from the respective directions of the real noise sources based on the head-related or outer-ear transfer functions

• • mehrere erste Schallwandler, ausgeführt ein Schallfeld in einem Außenraum des Fahrzeuges zu erfassen;
• • ein Hardwaremodul, ausgeführt Signale der ersten Schallwandler zu verarbeiten und gefilterte Signale zu erhalten gemäß des erfindungsgemäßen Verfahrens;
• • mehrere zweite Schallwandler, ausgeführt die gefilterten Signale als aus jeweiligen Richtungen der realen Geräuschquellen wahrnehmbar kommende Geräusche in dem Innenraum wiederzugeben.

According to one aspect, the invention provides a system for a vehicle to generate noise in an interior based on extracted and classified real noise sources. The system includes

• • a plurality of first sound transducers designed to capture a sound field in an exterior space of the vehicle;
• • a hardware module designed to process signals from the first sound transducer and to obtain filtered signals according to the method according to the invention;
• • Several second sound converters designed to reproduce the filtered signals as noises coming from the respective directions of the real noise sources in the interior.

In one aspect, the invention provides a vehicle that is acoustically transparent to specific target sounds. The vehicle includes a system according to the invention. At least a first group of at least two of the first sound transducers is arranged on or in a first bumper, for example the front bumper, of the vehicle. A second group of at least two of the first sound transducers is arranged on or in a second bumper, for example the rear bumper, of the vehicle. At least two loudspeakers or a pair of headphones are arranged in the interior. For example, the first group captures the sound field in the front area. The second group captures the sound field in the rear area, for example. The first sound converters can be arranged, for example, on or in the areas of license plate attachments. According to a further aspect of the invention, the first sound transducers are arranged on other body parts.

Alternatively, three of the first sound transducers can be arranged on a vehicle roof. The sound hits the first sound transducer equally from all directions. The distance between the first sound transducers can be in the range from 2 cm to 10 cm, preferably 4.5 cm. In this way, runtime differences can be easily detected.

Advantageous refinements of the invention result from the definitions, the dependent claims, the drawings and the description of preferred exemplary embodiments.

The first sound transducer is an acoustic sensor, for example. The acoustic sensor is an airborne sound pick-up in the form of a sensor that detects mechanical vibrations, for example caused by sound pressure from airborne sound waves, and converts them into a processable signal, for example an electrical signal such as an electrical voltage. The acoustic sensor includes an analog and/or digital signal output. According to one aspect, the forming takes place in two stages. In a first acoustic-mechanical conversion stage, the airborne sound is converted into the movement of an object according to a specific reception principle. In the second mechanical-electrical conversion stage, the movement of the object is converted into an electrical signal according to a specific converter principle. Examples of acoustic sensors are an arrangement of a magnet and an electric coil, microphones, accelerometers, piezo sensors or strain gauges. A micro-electro-mechanical system, MEMS for short, comprising an arrangement of semiconductor elements that record vibrations, can also be used as an acoustic sensor. In one aspect, the first sound transducer is a microphone. According to a further aspect, the first sound transducer is a MEMS microphone.

The second sound transducer converts an electrical signal into mechanical vibrations that can be perceived as sound. The second sound transducer is a loudspeaker or headphones, for example.

erhalten durch Abtastung des Empfängersignals r_i(t) und Mittelung seiner gewichteten und verzögerten Versionen, wobei N die Anzahl der ersten Schallwandler, ω_i die Gewichte, T die Abtastperiode und n_iT die gesteuerte Verzögerung für den i-ten der ersten Schallwandler sind.Beamforming is a signal processing technique for spatially selecting propagation waves, such as acoustic waves. According to one aspect of the invention, beamforming is performed with discrete time-weighted delay and sum, the so-called discrete-time weighted delay-and-sum beamforming. The time-discrete output signal $$bƒ\left(nT\right)=\frac{1}{N}{\displaystyle \sum _{i=0}^{N-1}{\omega }_i{\mathrm{right}}_i\left(nT-n_{i}T\right)}$$

obtained by sampling the receiver signal r _i (t) and averaging its weighted and delayed versions, where N is the number of first transducers, ω _{i is} the weights, T is the sampling period and n _i T is the controlled delay for the ith of the first transducers .

The isolated noises of the individual noise sources are classified. According to one aspect, the isolated noise sources are classified in a data-driven manner using trained or trainable machine learning models for acoustic event classification. According to one aspect of the invention, the machine learning models include artificial neural networks, for example convolutional networks, LSTM units, or, in particular in connection with speech recognition, transformers, for example bidirectional encoder representations of transformers, also called bidirectional encoder representations from transformers, BERT for short.

A head-related or outer ear transfer function, also head-related transfer functions, describes the direction and frequency dependency of the attenuation of the human ear. Head-related or outer-ear transfer function filtering means that a target signal is manipulated by frequency-dependent filtering in such a way that the corresponding sound is perceived by the human ear as if the sound is coming from a specific direction, even though the second sound transducers are picking up the sound play, be in a different place. For example:

Die resultierende Übertragungsfunktion $T=\frac{L\ast F}{H}$

entspricht dann dem resultierenden Filter. Eine Übertragungsfunktion H(f) bei einer Frequenz f ist allgemein gleich dem Quotienten aus Ausgangssignal und Eingangssignal bei dieser Frequenz.Let x ₁ (t) be the signal driving a real loudspeaker in an outdoor space, simulating one of the real noise sources, and let y ₁ (t) be the signal received by a microphone in a listener's eardrum. Then: y ₁ (t) = L _* F _* M _* x ₁ (t), where L is the response function of the outdoor speaker, F is the head-related or outer-ear response function, and M is the response function of the microphone. Let x ₂ (t) be the target signal that drives one of the second sound transducers, for example headphones or a loudspeaker in the interior, and let y ₂ (t) be the response signal from the microphone. Then: y ₂ (t) = H _* M _* x ₂ (t), where H is the headphone-eardrum transfer function. The task of the filter is to choose x ₂ (t) such that y ₂ (t) = y ₁ (t). That means: $x_2\left(t\right)=\frac{L\ast f}{H}{x}_1\left(t\right).$

The resulting transfer function $T=\frac{L\ast f}{H}$

then corresponds to the resulting filter. A transfer function H(f) at a frequency f is generally equal to the ratio of the output signal to the input signal at that frequency.

As a result, the second acoustic transducers in the interior are controlled in such a way that the signal that has been filtered or isolated sounds to the listener as if it is coming from the direction from which the actual signal is coming. It therefore sounds to the vehicle driver as if the vehicle were transparent and he can hear the target sound directly.

The vehicle is, for example, a car or truck, or a people-transport system, such as a shuttle or robotic taxi. The vehicle includes, for example, technical equipment for self-driving, ie driverless or fully automated, autonomous driving. The ADAS, i.e. advanced driver assistance system, or the AD, i.e. autonomous driving, Domain ECU, i.e. electronic control unit, as a central part of the driving system, uses environment detection sensors; which, for example, include the driving system, perceives a vehicle environment, derives a trajectory planning from this and determines corresponding control signals that are provided to vehicle actuators in order to regulate the longitudinal and/or lateral guidance of the driving system. The acoustic sensor of the arrangement according to the invention is an example of an environment detection sensor. Other environment detection sensors are, for example, optical sensors, such as a camera or lidar, or radar sensors. According to one aspect of the invention, the signals from the acoustic sensor are merged with signals from other environment detection sensors in order to localize and/or classify objects in road traffic. According to one aspect, the hardware module of the system according to the invention is integrated into the AD ECU of the vehicle.

According to one aspect of the invention, the first sound transducers measure sound pressures from a specific angle range, individually or in groups. When the first sound transducers are arranged in groups, for example a microphone array, transit time differences, also time delays of arrival, are evaluated within respective groups of the first sound transducers. This improves the localization of noise sources.

According to a further aspect, the angular scanning of the sound field is carried out by means of a steered-response power phase transform, or SRP PHAT for short. With this algorithm, every possible angle is sampled and the sound is extracted from the currently sampled angle. At the angles at which a real noise source is present, this results in a maximum in the SRP PHAT spectrum amplitude.

Dabei entspricht t_m(x) der Zeitverzögerung aufgrund der Schallausbreitung von der Geräuschquelle am Ort x bis zum Ort des m-ten Mikrofons. Durch diskrete Fouriertransformation erhält man eine Darstellung der Leistung P₀(x) mit gesteuerter Reaktion im Frequenzraum.In detail: Let s _m (n) be the time-discrete output signal of a microphone m with m ∈ {1,... , M}. The power P ₀ (x) with controlled response at location x = [x,y,z] ^T can be expressed as P ₀ (x) $P_0\left(x\right)\triangleq {\displaystyle {\sum }_{n\in Z}|{\displaystyle {\sum }_{m=1}^M{s}_m\left(n-{\tau }_m\left(x\right)\right){|}^2.}}$

Here, t _m (x) corresponds to the time delay due to sound propagation from the noise source at location x to the location of the mth microphone. A representation of the power P ₀ (x) with controlled response in the frequency domain is obtained by discrete Fourier transformation.

These maxima can be extracted using a maximum determination method. All maxima are then treated separately from one another. Each maximum corresponds to a different noise source. By focusing on the one maximum, other noises that do not come from that direction are dampened.

According to a further aspect, the isolated noise sources are classified into relevant classes, including noises from emergency vehicles, signaling noises from road users and/or accident noises, and into irrelevant classes, for example using the machine learning models. The signals obtained from the noises classified into the relevant classes are then further processed. Emergency vehicle sounds include siren, yelp, wail, rumbler, and similar sounds. Signaling noises include noises from endangered road users, so-called vulnerable road users, noises from bicycle bells, horn noises, calls for help, speech. Accident noise includes brake noise, tire noise and vehicle crash noise. Classes that are not relevant include, for example, engine noise during normal driving, conversations or dialogues between passers-by, church bells and others. In this way, the external noises are selectively put through.

According to a further aspect, speech is detected. When a corresponding isolated noise source is classified into a relevant class comprising signaling noises, the signals obtained are reproduced in an accelerated manner. According to a further aspect, mood recognition is carried out, or the content of what is spoken is analyzed by speech-to-text or speech-to-intent. Then the importance of the language for the listener is determined. Certain conversations outside the vehicle, for example a dialogue between two passers-by, should not be put through to the driver, firstly because it would disturb the driver and secondly in order not to violate data protection and confidentiality of the word. If the vehicle is spoken to directly, i.e. without reflection or inflection of the spoken word, which can be recognized acoustically, or in the case of an obvious call for help, the speech is put through directly according to another aspect so that the listener can be made aware of the situation. In order to compensate for the delay between received speech or other noises and the classification into a class with a noise that is "important" for the driver, the signal is reproduced at an accelerated rate. For example, 1s time is required for the classification. The signal can then be played back at 1.5x speed. After 3s, the signal from outside is again in sync with the signal from inside and can now be put through at normal playback speed without delay.

According to a further aspect, the filtered signals are reproduced in a superimposed manner. By simply superimposing the signals, several signals can also be played back simultaneously. Each signal can therefore be evaluated and reproduced individually and the signals do not have to be sorted according to importance. In one aspect, for less critical signals such as a bicycle bell or a horn, the signal is played from its direction and an audio source in the cabin, such as the radio in the vehicle, is played from the corresponding opposite direction, so that the driver can continue to hear his entertainment and less disturbed by ambient noise.

According to a further aspect, the filtered signals will be reproduced in the interior of the vehicle. The acoustically transparent vehicle is thus realized. In the case of the interior of the vehicle, at least two of the second sound converters, for example at least two loudspeakers, are provided according to one aspect of the invention. According to a further aspect, further loudspeakers are used in the interior. This improves the hearing impression. The sound can be better focused on the driver if more loudspeakers are installed, especially if there are several occupants in the vehicle.

According to a further aspect of the invention, sound fields in the exterior of vehicles driving in an automated manner are recorded and the filtered signals are reproduced in the interior of an institution that is higher-ranking with regard to the vehicles. This is particularly advantageous in the case of automated vehicles, for example a corresponding fleet of vehicles belonging to an operator. A hearing impression of the surroundings of the vehicle is thus obtained, for example, in a central location of the operator from which the vehicles are organized. In this way, n channels of the first sound transducers can also be reduced to two channels, for example headphones with two loudspeakers.

According to a further aspect, when filtered signals are reproduced in the interior of the vehicle, vehicle control is reduced Volumes of acoustic outputs of a navigation and/or infotainment system and/or engine torques. This allows the vehicle to react immediately to outside noise.

• 1 ein Ausführungsbeispiel einer erfindungsgemäßen Anordnung von ersten Schallwandlern,
• 2 eine schematische Darstellung des erfindungsgemäßen Verfahrens,
• 3 ein Ausführungsbeispiel einer kopfbezogenen oder Außenohr-Übertragungsfunktion und
• 4 ein Ausführungsbeispiel eines erfindungsgemäßen Fahrzeuges.

The invention is illustrated in the following exemplary embodiments. Show it:

• 1 an embodiment of an inventive arrangement of first sound transducers,
• 2 a schematic representation of the method according to the invention,
• 3 an embodiment of a head-related or outer ear transfer function and
• 4 an embodiment of a vehicle according to the invention.

In the figures, the same reference symbols denote the same or functionally similar reference parts. For the sake of clarity, only the relevant reference parts are highlighted in the individual figures.

1 shows a car as an example of an acoustically transparent vehicle 10 according to the invention. The first sound transducers 1 are, for example, microphones.

In vehicle 10, for example, a first microphone array comprising two of the first sound transducers 1 is arranged on or in a front bumper of vehicle 10, for example in the area of a license plate holder. The first microphone array captures sound paths, also called beams, 4 that come from a first angular area, for example from a front area.

In vehicle 10, for example, a second microphone array comprising two of the first sound transducers 1 is arranged on or in a rear bumper of vehicle 10, for example in the area of a license plate holder. The second microphone array captures sound paths, also called beams, 4 that come from a second angular area, for example from a rear area.

The sound paths or beams 4 can represent, for example, the output signals of a time-discrete beamformer algorithm or the SRP PHAT spectrum amplitudes. In 1 only 16 beams are shown for clarity. Typically, several beams are evaluated, for example 72 to 360.

A real noise source 30 is located in the front area of the vehicle. The beam from the direction of the noise source has the greatest amplitude.

2 shows the steps of the method according to the invention.

In a method step V1, a sound field in an exterior space of the vehicle 10 is detected by measuring sound pressures using a plurality of first sound transducers 1 that can be arranged on the vehicle 10. For example, the sound field of the in 1 noise source 30 shown detected with the first and second microphone arrays shown there.

In a method step V2, the sound field is scanned at an angle by means of beamforming signal processing. For example, the sound field is sampled with the SRP PHAT algorithm.

In a method step V3, the isolated noise sources 30 are classified, for example by means of a machine learning model trained for acoustic event detection, for example an artificial neural network. For example, relevant classes including noises from emergency vehicles, signaling noises from road users, including speech, including pedestrians screaming for help, and/or accident noises, and irrelevant classes are classified.

The signals obtained from the noises classified into the relevant classes are further processed in a method step V4 and filtered using head-related or outer-ear transfer functions.

3 Figure 12 shows the head-related or outer-ear transfer function X _R (f) for the right ear and the head-related or outer-ear transfer function X _L (f) for the left ear in the form of frequency responses to a noise source located anterior-upward relative to the head. The head-related or outer-ear transfer functions X _R (f), X _L (f) can be obtained, for example, by measuring a head-related impulse response at the ear drum for a sound pulse S(t) at the location of the noise source. The head-related or outer-ear transfer functions X _R (f),X _L (f) are then the Fourier transforms of the head-related impulse responses.

In a method step V5, the filtered signals are reproduced in the interior by means of second signals that can be arranged in the interior Sound transducers 2 and the generation of corresponding noises in the interior. Based on the head-related or outer ear transfer functions, the noises are perceived as coming from respective directions of the real noise sources 30, such as 4 in the example of the interior of the vehicle 10 shows. This creates the acoustically transparent vehicle 10. The signals filtered from speech are reproduced in an accelerated manner, for example.

4 shows the vehicle 10 according to the invention and the system according to the invention integrated in the vehicle 10 . The first sound transducers 1, as in 1 shown are not shown again for the sake of clarity. For example, two of the second sound transducers 2 are arranged in the interior of the vehicle 10 . The second sound transducers 2 include loudspeakers, for example.

The noise output in the interior of the vehicle takes place at the locations of the second sound transducers 2. Due to the method used according to the invention, however, an auditory perception of the noises arises from the location of the perceived noise source 3 shown. In this example, the perceived noise source 3 corresponds to the location of the in 1 shown real noise source 30.

Reference List

1

first transducer

2

second transducer

3

perceived noise source

4

sound path

10

vehicle

30

source of noise

V1-V5

process steps

QUOTES INCLUDED IN DESCRIPTION

This list of documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Patent Literature Cited

• DE 102019219525 B3 [0003]
• DE 102019213695 B3 [0003]
• DE 102019206331 A1 [0003]
• DE 102018200878 B3 [0003]

Claims (11)

Method for generating noises in an interior based on extracted and classified real noise sources (30) comprising the steps • Detection of a sound field in the exterior of a vehicle (10) by measuring sound pressures by means of a plurality of first sound transducers (1) (V1) that can be arranged on the vehicle (10); • Angular scanning of the sound field by means of beamforming signal processing, with sound pressures measured in an angle to be scanned in relation to one of the first sound transducers (1) being extracted from the sound field and thus the noise sources (30) being isolated from one another (V2); • Classification of isolated noise sources (30) (V3); • Further processing of signals obtained from the respective isolated noise sources (30) depending on the classification, the signals being filtered using head-related or outer-ear transfer functions (V4); • Playback of the filtered signals in the interior by means of second sound transducers (2) that can be arranged in the interior and generation of corresponding noises in the interior, the noises being perceived as coming from the respective directions of the real noise sources (30) based on the head-related or outer ear transfer functions become (V5). procedure after claim 1 , wherein the first sound transducers (1) measure sound pressures from a specific angular range individually or in groups, and when arranged in groups, transit time differences within respective groups of the first sound transducers are evaluated. A method according to any one of the preceding claims, wherein the angular sampling of the sound field is performed by power phase transformation with controlled response. Method according to one of the preceding claims, wherein the isolated noise sources are classified into relevant classes comprising noises from emergency vehicles, signaling noises from road users and/or accident noises and into irrelevant classes and the signals obtained from the noises classified into the relevant classes are further processed. Method according to one of the preceding claims, in which speech is detected and when a corresponding isolated noise source (30) is classified into a relevant class comprising signaling noise, the signals obtained are reproduced in an accelerated manner. Method according to one of the preceding claims, in which the filtered signals are reproduced in a superimposed manner. A method according to any one of the preceding claims, wherein the filtered signals are displayed in the interior of the vehicle (10). Method according to one of the preceding claims, in which sound fields in the exterior of automated vehicles (10) are detected and the filtered signals are reproduced in the interior of an institution that is higher-ranking with regard to the vehicles (10). System for a vehicle (10) for generating noise in an interior based on extracted and classified real noise sources (30) comprising • a plurality of first sound transducers (1) designed to detect a sound field in an exterior space of the vehicle (10); • a hardware module designed to process signals from the first sound transducers (1) and to obtain filtered signals according to the method according to any one of the preceding claims; • a plurality of second sound transducers (2), designed to reproduce the filtered signals as noises coming from the respective directions of the real noise sources (30) in the interior. Acoustically transparent vehicle (1) for specific target noises, comprising a system claim 9 , wherein at least a first group of at least two of the first sound transducers (1) are arranged on or in a first bumper and a second group of at least two of the first sound transducers (1) are arranged on or in a second bumper of the vehicle (10) and in which Interior at least two speakers or headphones are arranged. Vehicle (10) after claim 10 , wherein when filtered signals are reproduced in the interior of the vehicle (10), a vehicle controller reduces the volume of acoustic outputs from a navigation and/or infotainment system and/or engine torques.

Images