EP3375204B1 - Audio signal processing in a vehicle - Google Patents

Description

The present invention relates to a method for audio signal processing in a vehicle and a corresponding audio signal processing apparatus for a vehicle. In particular, the present invention relates to audio signal processing with echo cancellation, for example for speech processing.

In vehicles, such as passenger cars or trucks, speech dialogue systems are used to assist the driver or occupants. For example, speech dialogue systems serve to control electronic devices without the need for a haptic operator operation. The electronic devices may include, for example, a vehicle computer or a multimedia system of the vehicle. Spoken speech from the driver or occupant is received via a hands-free microphone and fed to speech recognition.

An application of microphones in the vehicle interior, for e.g. Voice control, telephony or vehicle interior communication can potentially be affected by acoustic coupling of loudspeaker outputs of the vehicle sound system. This can lead to recognition errors in the case of speech recognition, to far-end echo in the case of hands-free telephony, and to feedback in the case of vehicle interior communication. Depending on the application, the consequences are disturbed communication, increased distraction or even disturbing noises and echoes.

If, for example, during the speech dialogue in the vehicle, simultaneous and continuous audio signals are reproduced via the vehicle sound system, a part of the audio signals passes as acoustic feedback from the loudspeakers into the hands-free microphone and thus disturbs the speech recognition. The audio signals reproduced via the vehicle sound system can comprise, for example, music, traffic news, radio broadcasts, outputs of a navigation system or the (artificial) language of a speech dialogue system. The disturbance of the speech recognition can lead to recognition errors which can make the dialogue inefficient and an increased distraction from the driving task can cause. This can cause dissatisfaction or annoyance to the driver or occupant.

A simple solution to the aforementioned problem is to mute the audio playback of, for example, a radio during the voice dialogue or telephone call in the vehicle. The muting of audio playback is often perceived by vehicle users as disturbing and unnecessary. In addition, important information can be missed by, for example, a navigation system. In addition, it may be that a vehicle user feels in queries of the speech dialogue system and simultaneously muted audio playback urged to respond as quickly as possible to the queries of the speech dialogue system.

Alternatively, during audio dialogue, the audio playback volume may be temporarily reduced. For the speech recognizer, the extent of the interference by the audio playback is then lower, but generally still so large that it can not be dispensed with a further cleanup of the microphone signal.

To a limited extent, o.g. Couplings are also reduced by constructive-acoustic measures. For example, microphones with a suitable directional characteristic can be used, microphones and loudspeakers inside the vehicle can be suitably arranged relative to one another, or acoustic conditions in the vehicle can be suitably exploited.

However, since this is generally not enough, signal processing components are used to clean up the microphone signals. In the process, the signal components coupled into the microphones by the loudspeakers of the vehicle sound system are estimated and removed from the microphone signals. Such methods are referred to as echo cancellation or echo cancellation.

In this connection, the publication of Jacob Benesty et al: "A Hybrid Mono / Stereo Acoustic Echo Canceler" (APPLICATIONS OF SIGNAL PROCESSING TO AUDIO AND ACOUSTICS, 1997. 1997 IEEE ASSP WORKSHOP ON NEW PALTZ, NY, USA OCTOBER 19-22, 1997, NEW YORK, NY) USA, IEEE, US, October 19, 1997 (1997-10-19), page 4pp, XP010248238, ISBN: 978-0-7803-3908-8 ) a hybrid mono / stereo echo canceller based on a psychoacoustic principle. Two signals (left and right) are low-pass filtered to obtain the frequencies below f _c to realize the stereo effect. wherein the transition frequency f _{c is} on the order of 1 kHz. The sum of the left and right channels is high-pass filtered to obtain the frequencies above f _c as a monophonic signal. The left (or right) speaker signal is the sum of the low frequencies of the left (or right) channel and high frequencies of both channels. Two echo cancellers are needed: a high frequency mono echo canceller and a low frequency stereo echo canceller.

The US 2012/232890 A1 relates to a speech recognition system in which an n-channel system sound is converted by means of a monophonic unit into a single-channel acoustic signal. The single-channel acoustic signal is output to a weighting unit and to an echo canceller. The n-channel system sound is output through multiple speakers in, for example, a vehicle. A microphone is used to receive voice input from a user as well as multichannel system sound output from the speakers and to the echo canceller.

A common type of echo cancellation is linear echo cancellation.

• der Frequenz- und Phasengang des vor den Lautsprecher geschalteten Verstärkers,
• der Frequenz- und Phasengang des Lautsprechers,
• die räumliche Abstrahlcharakteristik des Lautsprechers,
• der akustische Übertragungspfad vom Lautsprecher zum Mikrofon durch den Fahrzeuginnenraum, einschließlich Reflexionen, Beugung, Streuung, Absorption, usw.,
• die räumliche Empfangscharakteristik des Mikrofons, sowie
• der Frequenz- und Phasengang des Mikrofons.

In linear echo cancellation, it is assumed that the microphones, loudspeakers and their respective amplifiers are linear transducers and that, thus, the loudspeaker sound components coupled to a specific microphone superimpose linearly in the microphone signal. Furthermore, it is assumed that these loudspeaker sound components result in a linear convolution of the respective loudspeaker source signal with a respective impulse response. Each of these impulse responses refers to a particular pair of microphone-speakers and characterizes the entire electro-acoustic transmission path from the source sound source to the microphone signal. Thus, among other things, the following quantities are mapped in such an impulse response:

• the frequency and phase response of the amplifier connected in front of the loudspeaker,
• the frequency and phase response of the loudspeaker,
• the spatial radiation characteristic of the loudspeaker,
• the acoustic transmission path from the speaker to the microphone through the vehicle interior, including reflections, diffraction, scattering, absorption, etc.,
• the spatial reception characteristics of the microphone, as well
• the frequency and phase response of the microphone.

This impulse response is therefore also referred to as LEM impulse response (Loudspeaker Enclosure Microphone). It is generally due to changes in the Vehicle interior geometry (occupants and their movements, moving parts, loading, etc.) as well as the electro-acoustic properties of microphones and loudspeakers (depending on temperature, air pressure, humidity, age, etc.) are time-variable.

A linear echo cancellation algorithm adaptively estimates the LEM impulse response for each possible microphone-speaker pair. Based on the LEM impulse response, the injected loudspeaker sound components in each microphone signal are then calculated and subtracted therefrom. The rate of adaptation and effective echo cancellation are limited and generally in competition with each other.

Various improved techniques for echo cancellation or echo cancellation are known in the art, for example to simplify echo cancellation and thus reduce the required computational power. For example, the disclosed EP 1936939 A1 an echo cancellation in which the microphone signal is divided into subband signals and sub-sampled. A reference audio signal is output through a speaker. The reference audio signal is also sub-sampled and sub-sampled subband signals of the reference audio signal are stored. Furthermore, echoes in the microphone subband signals are estimated and the estimated echoes are subtracted from the microphone subband signals to obtain improved microphone subband signals.

In echo cancellation, however, a frequently present multichannel nature of the audio signal to be output is problematic. The multi-channel audio signal may be, for example, a stereo signal or a surround signal in the vehicle.

In the case of multiple audio source signals from multiple speakers, in addition to the increased computational complexity of the algorithms, the following problem is added: Due to correlations between the different audio source signals, the estimation problem is mathematically underdetermined. As a consequence, in the event of a sudden appearance of audio source signals, the effectiveness of echo cancellation may be greatly reduced. It may even occur that the LEM estimate diverges, for example when there are changes in the surround sound image. This can occur, for example, when so-called phantom sound sources appear, disappear or shift in the surround panorama.

There are several approaches to dealing with this, but they either lead to audible distortions or are very computationally intensive (watermarking, Kalman filter solutions).

Furthermore, in this context from the DE 102008027848 A1 For example, an echo canceller is known which cooperates with a sound output device with a multi-channel audio unit. The sound output device outputs output sound signals as analog signals of multiple channels through a plurality of speakers. A microphone detects an outside sound and generates an input sound signal as an analog signal. The outside sound echoes the output sound signals. The echo canceller has an echo cancellation function for removing the echo from the input tone signal. For this purpose, the echo canceller receives the output sound signals from the sound output device. Such a solution for the compensation of multi-channel acoustic echo sources is technically very complex and requires a high computing power. Furthermore, for channel numbers greater than two, there are no explicit solutions.

Another possibility is an improved separation of speech signals and general interference signals. The general interfering signals may also include multichannel audio playback. This is for example in the DE 102009051508 A1 considered. To reduce interference in speech recognition, a microphone array is used instead of a single microphone. Through the microphone array a multi-channel speech signal is recorded, which is passed instead of a simple speech signal to an echo cancellation unit. The multichannel speech signal recorded by the microphone array is post-processed before being input to the echo canceling unit in a unit downstream of the microphone array for processing the microphone signals by delayed summation of the signals. As a result, the signals of the authorized speakers are separated and all other speaker signals and noise are reduced. In addition, the echo canceling unit evaluates the propagation time of the various channels of the multi-channel speech signal and removes all portions of the signal that do not emanate from the authorized speaker's location according to their propagation time. However, using a microphone array or multiple microphones increases costs, requires more installation space, and requires powerful computational resources.

It is therefore an object of the present invention to enable reliable voice input in a vehicle while reproducing a multi-channel audio signal. Additional costs or expenses for example, additional microphones or powerful signal processing units should be avoided.

According to the present invention, this object is achieved by a method for audio signal processing in a vehicle according to claim 1 and an audio signal processing apparatus for a vehicle according to claim 7. The dependent claims define embodiments of the invention.

In accordance with the present invention, a method for audio signal processing in a vehicle is provided. In the method, a mono audio signal is generated based on a multi-channel audio source signal. The multi-channel audio source signal is, for example, a stereo signal or a surround signal which is to be output in the vehicle via a plurality of loudspeakers of the vehicle. The mono audio signal is limited to a frequency range between a predetermined lower frequency and a predetermined upper frequency. The mono audio signal can be limited, for example, with a bandpass filter to the frequency range between the predetermined lower frequency and the predetermined upper frequency. Limiting the mono audio signal to the frequency domain produces a limited mono audio signal.

The limited mono audio signal is output through the plurality of speakers in the vehicle. Now, when a voice audio signal from a vehicle occupant or a driver of the vehicle is received via a microphone, this voice audio signal contains the limited mono audio signal output via the plurality of loudspeakers. An influence of this limited mono audio signal output via the several loudspeakers on the speech audio signal received via the microphone is compensated by means of the limited mono audio signal. For example, an echo compensation can be performed, which takes into account only the mono audio signal. A complex echo cancellation taking into account a multi-channel audio signal is therefore not required. Instead, only a single-channel echo cancellation is required, which can be realized with relatively low processing power.

The echo cancellation taking into account only one echo signal (mono audio signal) is very reliable, even if the mono audio signal is output through several different speakers, as with a mono audio signal no changes in the multi-channel sound image can occur. Thus, the interfering mono audio signal can be largely or completely removed from the voice audio signal.

The predetermined lower frequency has a value in the range of 100 Hz to 300 Hz, and the predetermined upper frequency has a value in the range of 4 kHz to 8 kHz. A speech recognizer, which is used, for example, for voice control or voice input in a vehicle, in many cases evaluates audio signals in a limited frequency range of, for example, 100 Hz to 8 kHz in order to recognize the voice input from a user. Therefore, echo cancellation is required only in this limited frequency range. Therefore, the predetermined lower frequency is preferably 100 Hz and the predetermined upper frequency is 8 kHz. As a result, an undisturbed speech signal can be provided to the speech recognizer in the limited frequency range relevant for him.

In order nevertheless to maintain an effect of a multi-channel audio reproduction, in one embodiment of the method additionally a plurality of limited channel-specific audio signals are generated as a function of the multi-channel audio source signal. A channel-specific audio signal relates, for example, to an audio signal which is determined by the multi-channel audio signal source specifically for a loudspeaker associated with the respective channel. For a stereo source signal, this may include, for example, an audio signal for the right speaker or an audio signal for the left speaker. A respective limited channel-specific audio signal of the plurality of limited channel-specific audio signals is therefore assigned to a respective audio signal of the multi-channel audio source signal. A respective limited channel-specific audio signal is limited to a frequency range which only includes frequencies below the predetermined lower frequency and frequencies above the predetermined upper frequency. A respective limited channel-specific audio signal is formed by a corresponding frequency limitation from the respectively assigned audio signal of the multi-channel audio source signal. In other words, the audio signals of the multi-channel audio signal are respectively limited or filtered so that they only include frequencies below the predetermined lower frequency and / or frequencies above the predetermined upper frequency. The plurality of limited channel-specific audio signals are output through the plurality of speakers in the vehicle, so that the effect of multi-channel audio reproduction such as stereo reproduction or surround reproduction can be achieved. In summary, an audio reproduction in the vehicle is modified so that in the frequency range between the predetermined lower frequency and the predetermined upper frequency, the multi-channel audio source signal is reproduced in one channel (mono) and in the remaining frequency range multi-channel.

For example, the mono audio signal and the plurality of limited channel-specific audio signals may be generated from the multi-channel audio source signal according to the following embodiment. In this embodiment, the multi-channel audio source signal is divided into a center signal component which is the same on all channels and a respective side signal component per audio channel of the multi-channel audio source signal. The limited mono audio signal is generated from the center signal component and the plurality of limited channel-specific audio signals are generated from the respective side signal components. The mid-signal component, for example, can be used directly as a mono audio signal or suitably scaled used as a mono audio signal. Likewise, the side signal components may be used directly as the limited channel specific audio signals or in suitably scaled form. In particular, in the case of a stereo signal, the center signal component can be formed, for example, from the sum of the right and left audio source signals. The side signal components may be coded together in a difference signal from the difference between the right and left audio source signals and further processed. As a result, in particular when processing a stereo source signal, the center signal component and the side signal components can be generated and processed in a simple manner.

In another embodiment, the center signal component is formed by averaging respective samples of the audio channels of the multi-channel audio source signal. The respective side signal components are formed by subtracting the center signal component from the respective audio signals of the multi-channel audio source signal. This generation of the center signal component and the side signal components is possible for audio source signals with an arbitrary number of channels. In addition, an implementation in, for example, a digital signal processor can be realized in a simple manner.

In a further embodiment of the method, the speech audio signal received via the microphone is limited to a frequency range between the predetermined lower frequency and the predetermined upper frequency. The echo cancellation is applied to the thus limited voice audio signal using the limited mono audio signal. Thus, the influence of the limited mono audio signal outputted through the plural speakers to the limited voice audio signal is compensated. Since the speech recognizer operates generally only in the frequency range between the predetermined lower frequency and the predetermined upper frequency, echo cancellation in a speech audio signal limited thereto is sufficient. Furthermore, spurious signals outside this frequency range are already eliminated before the echo cancellation and have therefore, does not affect echo cancellation and speech recognition, which makes both echo cancellation and speech recognition more reliable.

In some cases, the reproduction of an audio signal is more important to some occupants of the vehicle than to others. For example, audio outputs of a navigation system are more important to the driver than to the remaining occupants, whereas audio outputs of a video displayed in the rear of the vehicle are more important for rear passengers than for the driver and front passenger. Thus, according to one embodiment, a plurality of weighting factors associated with the respective loudspeakers may be generated in response to the multi-channel audio source signal. The limited mono audio signal is weighted for each loudspeaker with the weighting factor assigned to the respective loudspeaker. Thereby, a center of gravity of the audio output in the vehicle can be appropriately shifted.

As long as the weighting factors are essentially static, the weighted output will not affect the quality of the echo cancellation. When the weight is changed, the echo cancellation may adjust to the new weight in a relatively short time, for example within a few seconds or minutes. In the above example of the audio outputs of the navigation system, in a vehicle having, for example, four speakers instead of an equally distributed output of the mono audio signal via the four speakers, the following weighting may be used. For example, the speaker in the driver's area may output 70% of the mono audio signal, and the remaining three speakers may output only 10% of the mono audio signal, for example.

According to the present invention, there is further provided an audio signal processing apparatus for a vehicle. The audio signal processing apparatus is capable of generating a mono audio signal based on a multi-channel audio source signal. For this purpose, the audio signal processing device may have, for example, a summation device. The audio signal processing apparatus is further capable of limiting the mono audio signal to a frequency range between a predetermined lower frequency and a predetermined upper frequency. This can be realized for example with a bandpass filter. The predetermined lower frequency has a value in the range of 100 Hz to 300 Hz, and the predetermined upper frequency has a value in the range of 4 kHz to 8 kHz. The limited mono audio signal is output through several speakers in the vehicle. Furthermore, the limited mono audio signal is output to a compensation device, for example to a Echo canceller. The compensation device serves to compensate for an influence of the limited mono audio signal output via the plurality of loudspeakers on a voice audio signal received in the vehicle via a microphone by means of the limited mono audio signal. The audio signal processing device is therefore suitable for carrying out the method described above and its embodiments and therefore also comprises the advantages described above.

• Figur 1 zeigt schematisch ein Fahrzeug mit einer Audiosignalverarbeitungsvorrichtung gemäß einer Ausführungsform der vorliegenden Erfindung.
• Figur 2 zeigt schematisch ein Audiowiedergabesystem und ein Spracherkennungssystem in Verbindung mit einer Audiosignalverarbeitungsvorrichtung gemäß einer Ausführungsform der vorliegenden Erfindung.
• Figur 3 zeigt schematisch ein Verfahren für eine Audiosignalverarbeitung in einem Fahrzeug gemäß einer Ausführungsform der vorliegenden Erfindung.

The present invention will be described below in detail with reference to the accompanying drawings.

• FIG. 1 schematically shows a vehicle with an audio signal processing apparatus according to an embodiment of the present invention.
• FIG. 2 schematically shows an audio playback system and a speech recognition system in connection with an audio signal processing apparatus according to an embodiment of the present invention.
• FIG. 3 schematically shows a method for audio signal processing in a vehicle according to an embodiment of the present invention.

The present invention will be described below with reference to FIGS Figures 1-3 will be described in detail. In FIG. 1 First, the environment of an audio signal processing device 15 according to the invention in a vehicle 10 will be described. In FIG. 2 Details of the audio signal processing device 15 will be described in conjunction with other components of the vehicle 10. FIG. 3 Finally, Fig. 12 schematically shows the operation of the audio signal processing apparatus 15. Like reference numerals in the figures refer to the same or similar components.

FIG. 1 shows a vehicle 10 in a plan view. The vehicle 10 includes a voice recognition system 11. Using the voice recognition system 11, voice commands or instructions from occupants of the vehicle 10 may be captured, processed, and executed. For example, configuration settings of the vehicle 10 or a multimedia system of the vehicle 10 may be changed via appropriate instructions. For example, an audio signal source, such as CD or radio, can be selected. Furthermore, for example, a specific radio station can be selected or a title of a CD. Furthermore, with appropriate instructions a telephone connection to a desired participants are set up or a navigation destination can be set in a navigation system of the vehicle 10. For this purpose, for example, corresponding commands or instructions are received by a driver 12 of the vehicle 10 via a microphone 13. A spoken command of the driver 12 is forwarded by the microphone 13 as a speech audio signal to an audio signal processing device 15. The operation of the audio signal processing device 15 will be described later with reference to FIG FIG. 2 will be described in detail. After the processing of the speech audio signal in the audio signal processing device 15, the processed speech audio signal is supplied to the speech recognition system 11. The speech recognition system 11 evaluates the speech audio signal and detects contained commands and instructions and executes them. The speech recognition system can be coupled with a so-called dialogue system, which can lead a dialogue with the driver via questions and answers.

The vehicle 10 further includes an audio signal source 14. The audio signal source 14 may include, for example, a broadcast receiver, a media player such as a CD player or an MP3 player, or a navigation system of the vehicle 10. The audio signal source 14 outputs a multi-channel audio source signal. The multi-channel audio source signal is supplied to and processed by the audio signal processing device 15, as described below with reference to FIG FIG. 2 will be described. The processed multi-channel audio source signal is output from the audio signal processing device 15 to an amplifier 16. The amplifier 16 amplifies the individual signals of the processed multi-channel audio source signal so that they can be reproduced via speakers 17-20 in an interior of the vehicle 10.

In the in FIG. 1 As shown, the vehicle 10 includes four loudspeakers 17-20. In other embodiments of the invention, the vehicle 10 may include any number of speakers, for example two, three or more than four. In the example shown the FIG. 1 are the speakers 17-20 assigned to the seats of the vehicle 10. Thus, the loudspeaker 17 is associated with a driver's seat of the driver 12, the loudspeaker 18 with a passenger seat, the loudspeaker 19 with a rear right seat and the loudspeaker 20 with a rear left seat.

In operation of the vehicle 10, the driver 12 may voice instructions or commands to the speech recognition system 11. This will be in FIG. 1 represented by the dashed arrow between the driver 12 and the microphone 13. While the driver 12 is issuing commands and instructions, multi-channel audio source signals may be supplied from the audio signal source 14 be output through the speakers 17-20. The outputs of the speakers 17-20 also reach the microphone 13, as indicated by the corresponding dashed arrows between the speakers 17-20 and the microphone 13 in the FIG. 1 is shown. However, the outputs from the speakers 17-20 may interfere with speech intelligibility, such that the speech recognition system 11 may not or insufficiently recognize the commands and instructions from the driver 12.

FIG. 2 shows details of the audio signal processing device 15 and the speech recognition system 11, which help to reduce or compensate for the influence of the outputs from the loudspeakers 17-20 on the speech signal of the driver 12. For ease of illustration, the audio signal source 14 in the example of FIG FIG. 2 however, it is clear that the audio signal processing apparatus 15 described below can process any number of channels of a multi-channel audio signal source in the same way.

Before the operation of the audio signal processing device 15 will be described, the in FIG. 2 shown components of the audio signal processing device 15 described. In the FIG. 2 shown components of the audio signal processing device 15 may not necessarily be designed as concrete components or assemblies, but may be partially or wholly reproduced programmatically and implemented by a suitable controller, such as a microprocessor or a digital signal processor.

The audio signal processing device 15 includes inputs via which the multi-channel audio source signal is received by the audio signal source 14. A two-channel stereo audio source signal includes, for example, a left channel L and a right channel R, which are supplied to the audio signal processing device 15. With a first signal converter 21, a middle signal component M and for each channel a side signal component S are generated from the two- or multi-channel audio source signal. Especially for a stereo signal, instead of two side signal components, a common side signal component can be formed as the difference between the left channel L and the right channel R. Since, regardless of the number of side signal components, all side signal components are treated the same in the following FIG. 2 only one path for the side signal components S is shown. This one path can therefore comprise only one side signal component in the case of a stereo signal or several side signal components in the multi-channel case.

The middle signal component M may comprise, for example, a sum signal from all the supplied channels. In a stereo signal, therefore, the center signal component M may comprise the sum signal from the left channel L and the right channel R (M = R + L). A respective side signal component S may, for example, comprise a difference signal between the respective audio signal of the respective channel of the multi-channel audio source signal and the mid-signal component. Especially in the case of a stereo signal, the side signal component S may also comprise, for example, a difference signal from the right channel R and the left channel L (S = R-L).

The audio signal processing device 15 further comprises a first bandpass filter 23 and a notch filter or notch filter 22. The first bandpass filter 23 has a predetermined lower frequency and a predetermined upper frequency. The first bandpass filter 23 essentially passes only signals having a frequency between the predetermined lower frequency and the predetermined upper frequency. Signals with a frequency below the predetermined lower frequency and signals with a frequency above the predetermined upper frequency are substantially suppressed or at least greatly attenuated. In an analog embodiment of the first bandpass filter 23, the attenuation may be, for example, 70 dB or more, and in a digital embodiment of the first bandpass filter, the signal above the predetermined upper frequency and below the predetermined lower frequency can be completely suppressed. The notch filter 22 has a frequency response which is substantially inverse to the frequency response of the first bandpass filter 23. That is, the notch filter 22 essentially only passes signals having a frequency below the predetermined lower frequency or above the predetermined upper frequency. For example, the lower predetermined frequency may be 100 Hz, and the upper predetermined frequency may be 8 kHz, for example. Alternatively, the lower predetermined frequency may be selected in a range of 100 Hz to 300 Hz, and the upper predetermined frequency may be selected in a range of 4 kHz to 8 kHz. The larger the frequency range between the lower predetermined frequency and the upper predetermined frequency is selected, the more reliable the speech recognition function. However, reproduction of a multi-channel audio source signal is more affected the greater the frequency range between the lower predetermined frequency and the upper predetermined frequency is selected. In the event that several side signal components are generated, a corresponding notch filter 22 having the lower predetermined frequency and the upper predetermined frequency is to be provided for each of these plurality of side signal components.

By filtering the center signal component M with the bandpass filter 23, a filtered or frequency-limited mid-signal component Mb is generated. By filtering the side signal components S with the notch filters 22, filtered or frequency-limited side signal components Sb are generated. The filtered center signal component Mb and the filtered side signal component Sb are applied to a second signal converter 24, which generates filtered audio signals for the individual channels. The filtered audio signal for a respective single channel may be formed, for example, by summing the filtered center signal component Mb and the corresponding filtered channel-specific side signal component Sb. Specifically, in the case of a stereo audio source signal, Rb = Mb + Sb and Lb = Mb-Sb, for example. The filtered audio signals Lb, Rb are output from the audio signal processing device 15 and supplied to the amplifier 16 channel by channel.

The audio signal processing device 15 further comprises a second bandpass filter 26. The second bandpass filter 26 has the same filter characteristic as the first bandpass filter 23. The second bandpass filter 26 is coupled on the input side to the microphone 13 and on the output side with an echo canceller 25 of the speech recognition system 11. The echo canceller 25 of the speech recognition system 11, the filtered center signal component Mb is further supplied. Based on the filtered center signal component Mb, the echo canceller 25 performs echo cancellation on the filtered voice signal from the microphone 13. The speech signal processed by the echo canceler 25 is supplied to a speech recognizer 27 of the speech recognition system 11.

In addition, the audio signal processing device 15 comprises a weighting device 28, which is coupled to the multi-channel audio source signal and / or the audio signal source 14. Based on information from the multi-channel audio source signal or information from the audio signal source 14, the weighting device 28 provides weighting factors with which the filtered audio signals are weighted prior to being output from the second signal converter 24.

With reference to FIG. 3 The operation of the audio signal processing device 15 in the vehicle 10 will be described below. FIG. 3 shows a method 30 with method steps 31-37, which are carried out by the audio signal processing device 15 in conjunction with the speech recognition system 11. It is clear that in FIG. 3 illustrated processing steps can be performed with electronic tools, which include, for example, analog or digital circuits and processing devices. processing devices For example, they may include microprocessors or digital signal processors. Furthermore, the entire functionality of the audio signal processing device 15 can be integrated into, for example, an existing electronic device, for example a digital signal processor of the speech recognition system 11.

In step 31, a multi-channel audio source signal, for example a stereo signal or a surround signal, is received by the audio signal source 14 at the audio signal processing device 15. In steps 32 and 33, a frequency limited mono audio signal and frequency limited channel specific audio signals are generated by the first signal converter 21 and the filters 22 and 23. The frequency-limited center signal component Mb described above may be, for example, the frequency-limited mono audio signal. The frequency-limited side signal components Sb described above may be, for example, the frequency-limited channel-specific audio signals. However, the frequency-limited mono audio signal and the frequency-limited channel-specific audio signals can also be formed in any other way from the multi-channel audio source signal, for example in a digital signal processor.

In step 34, the limited mono audio signal is output through all the loudspeakers 17-20 and the limited channel specific audio signals are outputted over the loudspeaker associated with the respective channel. The mono audio signal is limited to a relevant frequency range for speech recognition, for example to a frequency range of 100 Hz to 8 kHz. The channel-specific audio signals are limited to a frequency range outside the relevant frequency range for speech recognition, that is, for example, to frequencies below 100 Hz and above 8 kHz. Due to the reduction of the multichannel nature of the audio reproduction within the frequency range relevant to the speech recognizer 27, only the single-channel mono audio signal is present as the disturbing signal for speech recognition. For the vehicle occupant, however, a feeling of spatial perception remains in the sound perception since the multichannel nature is retained for frequencies outside the range relevant for speech recognition.

When outputting the limited mono audio signal through the speakers 17-20, an audio center of gravity in the vehicle can be changed. For example, the weighting device 28 may determine an audio centroid for the multi-channel audio source signals or the current signal source based on the information supplied thereto, and the limited mono audio signal corresponding to that audio centroid on the audio source Distribute audio channels. For example, if a voice output of a navigation system represents the multi-channel audio signal source, the limited mono audio signal, for example, for the speaker 17 may be weighted more heavily than for the speakers 18-20, since this information is more relevant to the driver 12 than for the rest of the vehicle occupants. The weighting device 28 may consider further information of the vehicle 10, for example a current seat occupancy in the vehicle.

For speech recognition, a speech audio signal is received via the microphone 13 in step 35. In step 36, the received speech audio signal is frequency limited by the second bandpass filter 26. The echo canceller 25 is supplied with the limited mono audio signal and the limited voice audio signal. In step 37, the echo canceller 25 performs echo cancellation in the voice audio signal using the mono audio signal. Since both the speech audio signal and the mono audio signal are limited to the frequency range relevant to the speech recognition (eg 100Hz-8kHz), the echo cancellation can also be limited to this limited frequency range, whereby less noise occurs and the echo canceller 25 can be made simpler or less Computing power required. Furthermore, single-channel echo cancellation requires only a single audio reference signal, namely the mono audio signal, and only needs to estimate an acoustic impulse response. As a result, system resources are saved in the echo cancellation, which are available for example for the speech recognizer 27.

The thus adjusted speech audio signal is supplied to the speech recognizer 27 and processed there to extract corresponding commands and instructions from the spoken speech.

LIST OF REFERENCE NUMBERS

10

vehicle

11

Voice recognition system

12

passenger

13

microphone

14

Audio signal source

15

Audio signal processing device

16

amplifier

17-20

speaker

21

first signal converter

22

Notch filter

23

first bandpass filter

24

second signal converter

25

echo canceller

26

second bandpass filter

27

speech

28

weighting device

30

method

31-37

step

Claims (8)

1. A method (30) for an audio signal processing in a vehicle (10), comprising:

   - Generating a mono audio signal of a multichannel audio source signal,

   - limiting the mono audio signal to a frequency range between a prescribed lower frequency and a prescribed upper frequency,

   - outputting the limited mono audio signal via the multiple loudspeakers (17-20) in the vehicle (10), and

   - compensation of an influence of the limited mono audio signal output via the multiple loudspeakers on a voice audio signal received in the vehicle (10) via a microphone (13) by means of the limited mono audio signal,

   characterized in that the prescribed lower frequency has a value in the range of 100 Hz to 300 Hz and the prescribed upper frequency has a value in the range of 4kHz to 8kHz.
2. A method according to Claim 1, further characterized by:

   - Generating multiple limited channel-specific audio signals depending on the multichannel audio source signal in such a manner, that a respective limited channel-specific audio signal of the multiple limited audio signals is assigned to a respective audio signal of the multichannel audio source signal and is limited to a frequency range below the prescribed lower frequency and/or above the prescribed upper frequency, and

   - outputting the multiple limited channel-specific audio signals via the multiple loudspeakers (17-20) in the vehicle (10).
3. A method according to Claim 2, characterized in that the multichannel audio source signal is divided into a middle signal component, which is the same on all channels, and a respective side signal component per audio channel of the multichannel audio source signal and the middle signal component is used for generating the limited mono audio signal and the respective side signal components are used for generating the multiple limited channel-specific audio signals.
4. A method according to Claim 3, characterized in that the middle signal is formed by an averaging of respective sample values of the audio channels of the multichannel audio source signal and the respective side signal components are formed by subtraction of the middle signal from the respective audio signals of the multichannel audio source signal.
5. A method according to any one of the preceding claims, characterized in that the voice audio signal received via the microphone (13) is limited to a frequency range between the prescribed lower frequency and the prescribed upper frequency and the influence of the limited mono audio signal outputted via the multiple loudspeakers (17-20) on the limited voice audio signal is compensated.
6. A method according to any one of the preceding claims, further characterized by:

   - The generation of multiple weighting factors assigned to the respective loudspeakers (17-20) in dependence on the multichannel audio source signal, and

   - the outputting of a limited mono audio signal via the respective loudspeaker (17-20), which signal is weighted with the weighting factor assigned to the respective loudspeaker.
7. An audio signal processing device for a vehicle (10), which is designed

   - to generate a mono audio signal on the basis of a multichannel audio source signal,

   - to limit the mono audio signal to a frequency range between a prescribed lower frequency and a prescribed upper frequency,

   - to output the limited mono audio signal via multiple loudspeakers (17-20) in the vehicle (10), and

   - to output the limited mono audio signal to a compensation device (25), in order to compensate an influence of the limited mono audio signal outputted via the multiple loudspeakers (17-20) on a voice audio signal received in the vehicle (10) via a microphone 13) by means of the limited mono audio signal,

   characterized in that the prescribed lower frequency has a value in the range of 100 Hz to 300 Hz and the prescribed upper frequency has a value in the range of 4kHz to 8kHz.
8. An audio signal processing device according to Claim 7, characterized in that the audio signal processing device (15) is designed for carrying out the method according to any one of Claims 1-6.

Images