JP2011205692

Patent Translate
Powered by EPO and Google
Notice
This translation is machine-generated. It cannot be guaranteed that it is intelligible, accurate,
complete, reliable or fit for specific purposes. Critical decisions, such as commercially relevant or
financial decisions, should not be based on machine-translation output.
DESCRIPTION JP2011205692
A method is provided for automatically determining equalizing filter characteristics for a vehicle
cabin indoor communication system. An indoor communication system includes at least one
speaker and at least one microphone, and emitting a test signal set by a speaker of the indoor
communication system, in particular by one speaker, the indoor to obtain a received test signal
Receiving the test signal by a microphone of the communication system, in particular by one
microphone, automatically determining a transfer function in the frequency domain based on the
test signal and the received test signal, equalizing filter characteristics based on the transfer
function Including the step of automatically determining. [Selected figure] Figure 1
Indoor communication system for vehicle cabins
[0001]
The invention relates to an indoor communication system for a vehicle cabin and a method for
automatically determining the characteristics of an equalizing filter for an indoor communication
system of a vehicle cabin.
[0002]
In many different environments, noise environments can make communication between different
people very difficult or nearly impossible, especially when the noise is at a similar noise level to
speech.
10-05-2019
1
As a specific example, strong noise behind by engine and wind noise is present in vehicle cabins
(eg car cabins etc). Furthermore, possible noise sources that can reduce people's mutual
understanding are car speakers or hands-free telephone systems. Communication between the
front and back passengers and between the driver and front passengers is, for example, very
difficult due to several different noise sources, especially when the vehicle is moving at high
speeds. It is.
[0003]
In the prior art, indoor communication systems for cars have been proposed to improve the
communication between humans in the cabin. According to this prior art, one microphone takes
charge of each passenger seat including the driver seat. This means that the microphones are
arranged near the respective seats or near the seating heads. Each microphone records the story
of each passenger, and a corresponding signal is output into the car through the speakers.
Usually, the speakers already present in the car are responsible for the seats of the various
passengers. For example, if speakers are installed at each door, each speaker may be responsible
for the passenger sitting next to it. This allows the speaker corresponding to the passenger other
than the current speaker to mainly output the speech signal. For example, if the driver is
speaking, the corresponding speech signal may be output on a different speaker other than the
driver's speaker.
[0004]
These prior art indoor communication systems, however, have the drawback that unwanted
feedback effects can occur, especially at high amplification levels. In principle, this obstacle can
be overcome by setting the stability limit (i.e. the maximum permissible amplification) in a very
conventional way by the manufacturer. In other words, this maximum amplification limit can be
set to such a value that under almost all circumstances no unwanted feedback effects can occur.
However, this has the result that in most cases the optimal (higher) amplification that may
sometimes be required can not be achieved.
[0005]
In this respect, the object of the invention is to provide a method for operating an indoor
communication system, in particular for automatically assuming the feedback frequency of an
indoor communication system, ie an indoor communication system for a vehicle cabin. Provide a
way to make decisions.
10-05-2019
2
[0006]
This object is solved by a method for automatically determining an equalizing filter of an indoor
communication system of a vehicle cabin according to claims 1 and 2 and by an indoor
communication system for a vehicle cabin according to claims 14 and 15. .
[0007]
Thus, the present invention provides a method for automatically determining equalizing filter
characteristics for an indoor communication system of a vehicle cabin, ie an indoor
communication system comprising at least one microphone and at least one speaker, (A) emitting
a set (predetermined) test signal by a speaker of the indoor communication system, in particular
by one speaker, (b) receiving the test signal by a microphone of the indoor communication
system to obtain the received test signal (C) automatically determining a frequency domain
transfer function based on the test signal and the received test signal; (d) automatically
determining equalizing filter characteristics based on the transfer function. It is a method.
[0008]
The invention further provides a method for automatically determining an equalizing filter of an
indoor communication system of a vehicle cabin, ie an indoor communication system comprising
at least one speaker and at least one microphone array connected to a beamformer. Do.
The method comprises: (a) emitting a setting test signal from a speaker of the indoor
communication system, in particular one speaker, (b) receiving and beamforming a test signal by
the microphone array of the indoor communication system, and the received test signal
Beamforming a signal emanating from the microphone using a beamformer connected to the
microphone to obtain a (c) automatically determining a frequency domain transfer function
based on the test signal and the received test signal (D) automatically determining equalizing
filter characteristics based on the transmission function.
[0009]
Here and in the following, the term "microphone" (if not explicitly used in the context of a
microphone array) is used to denote a single microphone or in other words to denote a
10-05-2019
3
microphone that is not part of a microphone array Used for
A speaker may include many components, such as high frequency and low frequency devices, for
example.
[0010]
The transmission function is a function that characterizes the transmission characteristics of the
transmission path.
These transmission characteristics may result from the geometry of the room, where test signals
are transmitted from one or more speakers to one or more microphones and / or microphone
arrays and objects that may be assumed to be located there. Furthermore, the transmission
characteristics can be influenced by the electrical elements along the signal path. For example,
between signal generators and speakers. The received signal is a function of the original test
signal, ie the transmission function. The test signal is a signal generated by a signal generator.
[0011]
The beamformer processes the signals originating from the microphone array to obtain a
combined signal. In its simplest form (delay and sum beamformer), the beamformer involves
delay correction and summing of the signal. Beamforming allows for the provision of a unique
directional type to the microphone array.
[0012]
Thus, according to the method of the present invention, a test signal is used to determine the
transmission characteristics of the signal path as a function of the frequency domain. Equalizing
filter characteristics are determined based on these transmission characteristics. Throughout the
specification, in contrast to conventional indoor communication systems, it is the detection of the
feedback frequency assumed by the transmission function that this method takes into account, it
being output that this transmission function can be used This is for proper correction of the
signal, which is to reduce the occurrence of feedback effects.
10-05-2019
4
[0013]
The indoor communication system in which this method is used may include one or more (single)
microphones and / or one or more microphone arrays. Such microphones or microphone arrays
may be associated with potential speakers (passengers) and may be equipped at appropriate
locations. Additionally, the indoor communication system may also include a microphone or
microphone array and one or more speakers associated with the corresponding talker position.
[0014]
In particular, in step (a), the setting test signal can be emitted (exactly) by one loudspeaker. In
this way, in particular, for each combination of loudspeakers / microphones or each combination
of loudspeakers / microphone arrays, a transfer function is obtained, and for each combination,
an appropriate equalizing filter is provided. Equalizing filter characteristics may be determined.
In step (a), test signals may be received simultaneously by the respective microphones or
microphone arrays. This allows to determine simultaneously the equalizing filters for each of
these microphones or microphone arrays for a given speaker.
[0015]
Step (a) may involve emitting binary white noise or colored noise, in particular pink noise. In
white noise, all frequencies (in the present context, only frequencies in the human hearing range
are considered) are present and uniformly distributed. This allows a comparison of the initial test
signal and the received test signal in a very simple manner. A more pleasing sound for people
present in vehicle cabins can be obtained using colored noise, in particular pink noise. Pink noise
can be obtained by filtering the white noise to reduce the amount in each octave.
[0016]
Step (c) involves processing the test signal and the received test signal in the time domain, in
particular using, in particular, an estimation method based on a matching filter or periodogram,
an autocorrelation function or a cross correlation function. Processing in the time domain is
10-05-2019
5
particularly useful when using wide frequency band test signals. For example, the NMLS (least
mean square) algorithm (Source: eg S. Haykin, B. Widrow, Least-Mean-Square Adaptive Filtering,
John Wiley and Sons, 2003) is used after processing in the time domain and impulses The
response may be transformed into the frequency domain to obtain a transfer function.
[0017]
As an alternative to using a wide frequency band test signal, step (a) involves emitting band pass
noise with various center frequencies. In this case, however, all frequency ranges should be valid.
[0018]
If band pass noise is emitted, step (c) may include determining the short power spectrum of the
received test signal and the square root of the short power spectrum of the test signal for each
band pass noise. The transfer function can also be obtained in this way.
[0019]
In the above method, step (d) may comprise determining the extrema of the magnitude of the
transfer function, in particular the maxima. The frequency at which the transfer function has a
(global or local) local maximum is taken as the possible feedback frequency.
[0020]
Step (d) may include determining the overall extrema of the magnitude of the transfer function.
In particular, the global maxima for the entire frequency range considered have a high
probability of configuring the feedback frequency.
[0021]
10-05-2019
6
Step (d) may further include determining the local extrema, the magnitude of the transfer
function at this value being different from the magnitude of the transfer function at the full
extremum by at most a set amount . In particular in the case of local maxima, other frequencies
(in addition to the global maxima) can also be considered as possible feedback frequencies.
[0022]
Step (d) may further include providing attenuation of a set frequency band of the signal centered
at the determined maximum value and / or setting frequency centered at the determined
minimum value. It is to provide amplification of the band.
[0023]
This gives particularly advantageous equalization.
With such an attenuation of the signal output by the loudspeaker, the stability limit (of the wide
frequency band) can be improved by a value corresponding to the applied attenuation. In
particular, when a small attenuation bandwidth is chosen, the auditory impression of the audio
signal in the vehicle cabin is hardly different.
[0024]
The attenuation step may be performed using a non-linear phase filter. The advantage this has is
that when reproducing a single signal, it is possible to improve the quality of the subjective
impression of the signal by inserting a small delay time difference.
[0025]
The invention further provides a method of automatically determining the maximum
amplification value of an indoor communication system of a vehicle cabin, which method
automatically determines equalizing filter characteristics using the method described above for a
set loudspeaker, Maximal as Gainmax = 1 / max {¦ H (e <jΩ>) G (e <jΩ>) ¦} using the transfer
function H (e <jΩ>) and the frequency response G (e <jΩ>) of the equalizing filter Automatically
determining amplification.
10-05-2019
7
[0026]
Determining this maximal amplification or gain allows, for example, to increase the maximal
amplification as set by the manufacturer.
[0027]
Furthermore, the present invention provides a computer program product, which comprises one
or more computer readable media having computer implemented instructions for performing the
steps of the above method.
[0028]
Furthermore, the invention provides an indoor communication system for supplying for a vehicle
cabin, a signal generator configured to generate a configuration test signal, connected to the
signal generator for generating the configuration test signal. At least one speaker, at least one or
more microphones for receiving a test signal generated to provide a received test signal, a signal
generator and signal processing means connected to the at least one or more microphones The
processing means is configured to automatically determine a transmission function in the
frequency domain based on the test signal and the received signal, and to automatically
determine equalizing filter characteristics based on the transmission function.
[0029]
Furthermore, the vehicle cabin indoor communication system provided is provided with: a signal
generator configured to generate a configuration test signal; at least one speaker connected to
the signal generator for emitting the configuration test signal; At least one or more microphone
arrays for receiving a test signal, a beamformer for each microphone array for beamforming a
signal diverging from the microphone array to provide a received test signal, a signal generator
and at least one Signal processing means connected to the microphone array, wherein the signal
processing means automatically determines the frequency domain transfer function based on the
test signal and the received test signal and automatically determines equalizing filter
characteristics based on the transfer function Means configured to It encompasses.
[0030]
These indoor communication systems allow, for example, the use of components already present
in the vehicle cabin, such as speakers and microphones or microphone arrays, for use in handsfree telephone systems.
10-05-2019
8
At the same time, in the case of microphone arrays, each microphone array is connected to the
signal processing means through a beamformer.
It should be understood that both the microphone and the microphone array can be used in a
vehicle cabin.
For example, a microphone array may be provided to front-seat passengers, but there is only one
microphone at the back-seat.
[0031]
An indoor communication system may include at least two speakers, and a signal generator may
be connected to the at least two speakers.
And it can be connected so that test signals can be emitted separately by the respective speakers.
This makes it possible to carry out the method independently for each speaker independently of
the other speakers.
[0032]
The signal generator may be configured to emit binary white noise or colored noise, in particular
pink noise.
[0033]
The signal processing means of the indoor communication system may be configured to process
the test signal and the received test signal in the time domain, in particular using an estimation
method based on a matching filter or periodogram, an autocorrelation function or a cross
correlation function. .
10-05-2019
9
In particular, the signal processing means may include an NMLS (Standardized Least Squares)
adaptive filter.
[0034]
As an alternative to a signal generator configured to emit wide frequency band noise, the signal
generator may be configured to generate band pass noise having various center frequencies.
[0035]
In this case, the signal processing means may be configured to determine the square root of the
quotient of the short time power spectrum of the received test signal and the short time power
spectrum of the test signal for each band pass noise.
[0036]
The signal processing means of the indoor communication system may be configured to
automatically determine the extrema, in particular the maxima of the magnitude of the
transmission function.
[0037]
The signal processing means may be configured to determine an overall extremum of the
magnitude of the transfer function.
[0038]
The signal processing means may then be configured to further determine the local extrema, for
which the magnitude of the transfer function is at most the set magnitude of the magnitude of
the transfer function and the global extremum. Only different.
[0039]
The signal processing means may be configured to automatically subtract the set frequency band
centered at the determined maximum value and / or automatically amplify the set frequency
band centered at the determined minimum value.
[0040]
In such cases, the signal processing means may include a non-linear phase filter to reduce the set
10-05-2019
10
frequency band.
[0041]
The signal processing means may be further configured to automatically determine the maximal
amplification value as the value shown below.
Gainmax = 1 / max {¦ H (e <jΩ>) G (e <jΩ>) ¦} where H (e <jΩ>) is the transfer function and G (e
<jΩ>) is the frequency of the equalizing filter It is a response.
[0042]
FIG. 1 is a schematic explanatory view of an embodiment of an indoor communication system of
a vehicle cabin.
FIG. 2 is a schematic illustration of an embodiment of an indoor communication system for
determining equalizing filter characteristics.
FIG. 3 is a flow diagram of an embodiment of a method for automatically determining equalizing
filter characteristics.
FIG. 4 is a diagram for explaining an embodiment of a method for reducing an assumed feedback
frequency.
FIG. 5 is a flow diagram of an embodiment of a method for correcting maximal amplification.
[0043]
Further features and advantages of the invention are described in the following with the
examples and figures.
10-05-2019
11
[0044]
The invention will now be described with respect to examples, which as well as the figures will be
described in the following detailed description.
The detailed description and figures are not intended to limit the invention to the specifically
disclosed embodiments, but rather the described exemplary embodiments merely illustrate the
various aspects of the invention and the scope of the invention It is defined by the appended
claims.
[0045]
An overview of an indoor communication system in a vehicle cabin 100 is shown in FIG.
Vehicle cabin 100 constitutes four passenger seats (not shown) and four passengers 101.
Associated with each passenger is a microphone array that includes two microphones 102 and
speakers 103. For front seat passengers, the microphone array is centered between the
passengers. They can be provided, for example, on the roof of a car. In the rear seat, microphone
arrays are provided to the left and right of the passenger, for example on the door of the car
respectively. Each speaker may have, for example, high frequency and low frequency devices.
[0046]
The microphone array records the signal. The signals emanating from the microphone array
enter digital signal processing means 104 and 104 '. In the example shown in FIG. 1, the first
signal processing means 104 is responsible for processing the signal emanating from the
microphone associated with the rear passenger, and the second digital processing means 104 'is
the microphone responsible for the front passenger Related to the signal emitted from However,
this is only a matter of choice, and the signals emanating from different microphone arrays and
the signals output by different speakers can also be processed in various combinations.
10-05-2019
12
[0047]
The signals emanating from the microphone array are first of all processed by an analog / digital
converter (not shown). This is followed by beamformers 105 and 105 '. Beamformers are useful
for obtaining temporal signal movement information and for obtaining spatial information of
both signal sources (eg, passengers) and noise signals (eg, car radio speakers). . Different types of
beamformers (e.g., compatible beamformers) may be used in the present invention.
[0048]
Apart from the method according to the invention, additional feedback effects can be avoided
during the processing of the indoor communication system. From this, the echo and feedback
elements are subtracted from the signal through matching echo cancelers 106 and 106 'and
feedback cancelers 107 and 107'. The matching notch filters 108 and 108 'may detect potential
feedback and attenuation frequencies.
[0049]
Based on temporal and spatial information from the beamformer, attenuators, the attenuators are
controlled, for example, such that only the loudest speakers are connected. However, such
corresponding components are not explicitly shown in FIG. Post processing 109 and 109 'may
apply boundary properties. The output signals of the digital signal processing means 104 and
104 'are first sent to the corresponding speaker 103. In the example shown in FIG. 1, the signals
emanating from the rear microphone array can only be output by the front speakers and vice
versa. However, this is not necessarily a necessary limitation, and the system can also be
configured such that the processed signal can be output by all speakers except those
corresponding to the microphone array of the input signal, for example.
[0050]
Furthermore, the processing signal may also be sent to feedback cancelers 107 and 107 'of
similar digital signal processing means and echo cancelers 106 and 106' of other digital signal
processing means.
[0051]
10-05-2019
13
As an alternative to the example shown in FIG. 1, it is possible to replace some or all of the
microphone array with a single microphone.
For example, an instruction microphone may be used instead of a microphone array.
[0052]
An example of an indoor communication system and its method of operation are shown in FIG.
Again, the vehicle cabin 100 and the passenger 101 (and the other passengers envisaged) are
shown. In this illustrated example, the method of determining equalizing filter characteristics is
performed for only one active speaker 103; the other speakers that are not active are shown in
dotted lines.
[0053]
A signal generator 201 is present to provide the test signal x (n). Such test signals are passed
through D / A converter and amplifier 202. Each speaker may be addressed independently of the
other speakers, as indicated by switch 203.
[0054]
The test signal 103 emitted by the speaker 103 is recorded by the microphone array 102. In
particular, the test signal can be recorded simultaneously by all microphone arrays; however, it is
also possible to record the signals consecutively by different microphone arrays. Signals
emanating from the microphone array are passed through an A / D converter and amplifier 204.
Thereafter, beamforming is performed at the beamformer 205 which produces a beamformed
signal y (n). This signal is passed to the signal processing means 206, which is connected not
only to the microphone array but also to the signal generator 201 in order to receive the test
signal x (n).
[0055]
10-05-2019
14
The signal processing means 206 responsible for determining the equalizing filter characteristics
can be part of the signal processing means 104 or 104 'of FIG. However, it is also possible to
provide this signal processing means 206 as a separating device. This is particularly useful if
later equipped with a vehicle that already has an equipped indoor communication system.
[0056]
The operation of the embodiment of the method for automatically determining equalizing filter
characteristics is shown in FIG. In a first step 301, a test signal is generated. In principle,
different test signals are possible. In particular, binary white noise can be used as a test signal.
This is an advantageous choice because all frequencies are present and uniformly distributed.
However, in most cases, colored noise may also be used because the passenger is sitting in the
vehicle cabin. For example, pink noise may occur. Other colored noises can also be used as long
as all frequencies are sufficiently effective.
[0057]
Instead of a wide frequency band signal, it is also possible to use narrow frequency band test
signals such as band pass noise with various center frequencies. In this case, all frequency ranges
remain valid.
[0058]
In the next step 302, a test signal is generated by one of the speakers. This corresponds to the
configuration of FIG. However, two or more speakers can alternatively emit a test signal at the
same time.
[0059]
Next, in step 303, a test signal is received by the microphone array, in particular by all the
microphone arrays. The signals originating from the microphone array and passing through the
10-05-2019
15
beamformer are sent to the signal processing means where the test signal and the received test
signal are used in step 304 to automatically determine the transfer function in the frequency
domain. In the case of wide frequency band test signals, the transmission behavior can be
determined in the time domain, for example using a suitable filter. For this reason, a normalized
least squares (NLMS) algorithm can be applied that uses small increments. After this, the impulse
response may be transformed to the frequency domain to obtain the transmission function H (e
<jΩ>). Instead of a fitting filter, one can also use estimation methods based on periodograms or
autocorrelation functions or cross correlation functions.
[0060]
If a narrow frequency band test signal is applied, the transfer function may be determined as the
square root of the quotient of the short time power spectrum.
[0061]
Both short time power spectra are determined for each band pass noise in order to obtain sample
points at the center frequency of the band pass signal.
[0062]
In the next step 305, equalizing filter characteristics are determined such that equalizing filters
for the microphone array / speaker combination are obtained and the reduced feedback risk for
the components of the combination.
[0063]
By performing these steps simultaneously for a particular speaker and all microphone arrays,
equalizing filters are obtained for each of these combinations.
The process may then be repeated with other speakers until the method is performed for all the
speakers present in the vehicle cabin.
[0064]
FIG. 4 illustrates an example of how to determine (at least a portion of) the equalizing filter
10-05-2019
16
characteristics in step 305 of FIG.
This is particularly useful when determining the transfer function using the matched filter
described above.
According to this example, one or more local maxima of the transfer function H (e <jΩ>) are
determined in step 401, in other words, the local maximum frequency Ω max
[0065]
Is determined. The number of frequencies thus determined is in particular based on the design
principle of the equalizing filter. For example, if these frequencies are sufficiently separated, then
values of frequencies of 3 to 5 may be determined. In other cases, a frequency value of 5 or more
may be determined.
[0066]
When narrow-band test noise is used, it is based on the above-mentioned short-time power
spectrum that the band pass range having the frequency Ωmax or the largest quotient is
determined.
[0067]
An assumed feedback frequency is then determined for the particular speaker / microphone
array combination.
[0068]
After determining the assumed feedback frequency at step 401, at step 402, attenuation is
provided for one or more frequency bands (depending on the number of assumed feedback
frequencies).
In this process, attenuation is applied to a minute band around the determined frequency Ωmax.
10-05-2019
17
In particular, some dB attenuation can be inserted where the remaining frequency range is not
attenuated.
[0069]
This step may be performed to amplify the minima of the transfer function. In such cases, the
frequency of the minima can be determined and the corresponding amplification can be applied.
In this way, the output signal is considered annoying by the listener.
[0070]
Since the transmission functions of different speaker / microphone array combinations are
usually different, an additional positive effect may be achieved when using a non-linear phase
equalizing filter. When producing a single signal through multiple speakers, the insertion of a
small delay difference increases the subjective quality impression (see, eg, M. Schroeder,
Computer Speech-Recognition, Compression, Synthesis, Springer 1999). This effect can be taken
into account when determining the equalizing filter.
[0071]
FIG. 5 illustrates the correction of the maximal amplification of the indoor communication system
after determining the transmission function at step 501, as described above. Maximum gain
(amplification) is determined in step 502. Gainmax = 1 / max {¦ H (e <jΩ>) G (e <jΩ>) ¦} where H
(e <jΩ>) corresponds to the transfer function and G (e <jΩ>) corresponds to the equalizing filter
Frequency response. This reciprocal value may be determined for each speaker / microphone
array combination.
[0072]
Knowing this stability limit, the maximal amplification can be corrected in step 503. This enables
the operation of the indoor communication system near the stability limit. And this is especially
useful in the case of wide external noise. By determination of the transfer function, this stability
10-05-2019
18
limit can be determined individually for each vehicle (and for each attached speaker and
microphone or microphone array), + compensating for the maximal amplification initially set by
the manufacturer May be used to (increase). In this way, the quality of the indoor communication
system can be further improved.
[0073]
In order for the equalizing filter to be used to reproduce the signal of the indoor communication
system, the equalizing filter should be placed before the mixing matrix of the audio amplifier. In
this case, the equalizing filter does not modify the sound of other audible sources of the vehicle.
[0074]
Furthermore, the method may also be used to improve the audible output of other audible
components such as hands free systems and speech recognition. If it is detected by the above
method that the so-called center speaker located near the front microphone produces a
microphone signal larger than the other speakers, its output power can be reduced accordingly.
Overdrive of the microphone with audible output can be avoided in this way, increasing the
performance of the hands free system or speech recognition system.
[0075]
Furthermore, modifications and variations of the present invention will be apparent to those
skilled in the art in light of this specification. Accordingly, this specification is to be construed as
illustrative only and is for the purpose of teaching those skilled in the art the general manner of
carrying out the present invention. It is to be understood that the forms of the invention shown
and described herein are to be taken as the presently preferred embodiments.
[0076]
100 Vehicle cabin 101 Passenger 102 Microphone 103 Speaker
10-05-2019
19