JPH08503832

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 JPH08503832
[0001]
Mobile radio device with hands-free speech device The invention relates to a hands-free device
comprising combining means for combining at least two acoustic input signals and an adaptive
filter for filtering the combined output signal of the combining means. The present invention
relates to a mobile radio apparatus provided with a speech apparatus. Furthermore, the invention
relates to a hands-free speech device comprising combining means for combining at least two
acoustic input signals and an adaptive filter for filtering the combined output signal of the
combining means. A hand-free communication device having four microphones is known, for
example, from the known document "AMAICROPHONE ARRAY WITH ADAPTIVE POSTFILTERING
FOR NOISE REDUCTION IN REBERBERANT ROOM; Rainer Ze1 inski, ICASSP 88; pages 2578 to
2581". In the first processing step, the directional gain of the two-dimensionally arranged
microphones is used for noise reduction. In a second processing step, the microphone signal is
post-processed by means of an adaptive Wien filter (which evaluates the speech signal). For
calculating the parameters of the Wien filter, the autocorrelation function and the cross
correlation function of the input signal are obtained. The object of the present invention is to
improve the speech quality in a mobile radio device of the type mentioned at the outset as well as
in a hands-free device. The object is achieved according to the invention by means of a mobile
radio device of the type mentioned at the outset as well as an arrangement in which a high-pass
filter is provided for filtering the acoustic input signal in a hands-free device. The present
invention is based on the following recognition. It is based on the fact that the microphone signal,
that is to say the high-frequency disturbance component in the acoustic input signal in the motor
vehicle, causes various undesired effects in the calculation of the parameters for the adaptive
filter. This action degrades the speech quality of the output signal. For the processing of analog
acoustic input signals coming from microphones, for example configured as a microphone array,
these signals are firstly converted, for example, into digital signals in an A / D converter. The
08-05-2019
1
subsequent combining means are used to combine the digital input signals obtained as described
above into one possibly noiseless and clean output signal. An adaptive filter disposed after the
combining means is used to reduce disturbing noise.
This disturbing noise is caused, for example, by a wide variety of engine noises in the case of a
hands-free device provided in a motor vehicle. The use of particularly expensive microphones
leads to a deterioration of the output signal of the hands-free device. A reduction of this kind of
negative effect is achieved in a surprisingly simple way by means of high-pass noise of the
acoustic input signal. For example, a fourth-order digital IIR-filter can be used as the high-pass
filter. Such high pass filtering of the input signal eliminates the relatively poor signal processing
of the hands-free device which is attributable to the low frequency components of the input
signal. This results in an overall improvement in the speech quality of the output signal of the
hands-free speech system in the mobile radio system. This results in an improvement in the
speech quality of hands-free speech devices and mobile radios by filtering out relatively high
signal energy (which consists mainly of impairments) in the low frequency range. The
improvement of the speech quality of the hands-free speech apparatus in the mobile radio
apparatus can be obtained when the cutoff frequency of the high-pass filter is in the range of
200 to 400 Hz. In particular, when the cut-off frequency of the high-pass filtering filter is set to
300 Hz, significant improvement in the quality of the output signal is observed. This is because
frequencies below 300 Hz do not matter in any way for calls suitable for mobile radios. Further
improvement of the call quality of the hands-free device is ensured by the following. That is, it is
ensured by providing a further high-pass filter for filtering the output signal of the adaptive filter.
Such additional high-pass filtering further reduces low-frequency interference signal components
that have not been sufficiently suppressed by the adaptive filter. This is based on the following
recognition. That is, even if the input signal does not contain low frequency signal components,
additional various artifacts due to the time dispersion of the adaptive filter (this energy is
concentrated in the lower frequency region) are generated. Is based on Further low pass noise
attenuates this low frequency signal component and at the same time emphasizes relatively high
signal frequencies. This avoids a sense of hearing that may occur as a false response due to the
attenuation of high frequency signal components. A better speech quality of the hands-free
device in a mobile radio can be obtained if the cut-off frequency of the further high-pass filter is
in the region of 200 to 400 Hz.
In particular, when the cut-off frequency of another high-pass filter was 300 Hz, a significant
improvement in the quality of the output signal was observed. For example, considering transit
time differences of acoustic input signals picked up by a microphone array can be performed in a
simple manner by the following. That is, the coupling means is provided with a travel time
compensation adjustment device for compensating the travel time between the input signals. On
the other hand, the transit time compensation adjustment device is disposed in front of the
08-05-2019
2
summing device of the input signal and is used to compensate for transit time differences based
on the time difference between the speech signal components of the acoustic input signal, ie
microphone signal. Further improvement of the speech quality can be obtained by the following.
That is to say that the transit time compensation adjustment device comprises plausibility
checking means, wherein the detected distance interval and / or direction of the origin (starting
point) of the input signal to the caller exceeds a predetermined limit value. It is obtained by the
configuration provided to check whether the In this type of plausibility check, it is checked
whether the distance to the talker detected during focusing on the talker exceeds a
predetermined limit value. This limit value in this case is selectable depending on the internal
dimensions of the vehicle. In addition, the direction in which the input signal is picked up can
additionally be involved in this plausibility check. According to this, it is possible to exclude the
position of the wrong caller, that is, the position which is unlikely to be real (for example, the call
position behind the microphone, etc.). For example, poor speech quality caused by mis-estimated
correlation functions and / or quantitative problems can be improved by the following
configuration. That is, the hand-free communication device has a device for calculating the
parameter of the adaptive filter from the autocorrelation function and the cross correlation
function of the input signal, for example, the temporal averaging of the autocorrelation function
and / or the cross correlation function is performed Improved by such a configuration. According
to an advantageous means for the calculation of the cross correlation function, scaling of the
autocorrelation function and association of the autocorrelation function to the cross correlation
function are performed. The disturbing effects in hands-free calling can be reduced in a simple
manner by the following configuration. That is, the hands-free device can be reduced by a
configuration having switching means for not using the adaptive filter when coefficient overflow
occurs in the calculation of the autocorrelation function and / or the cross correlation function.
A Wien filter is preferably used as a filter suitable for the adaptive filter. In the following
specification, the invention will be described in detail on the basis of the illustrated embodiments.
BRIEF DESCRIPTION OF THE DRAWINGS FIG. 1 is a perspective view of a motor vehicle with a
hand-free device shown schematically. FIG. 2 is a view showing an embodiment of the handsfree
communication device. DESCRIPTION OF THE PREFERRED EMBODIMENT FIG. 1 shows a
perspective view of a car 7 with a driver 3 on it. Furthermore, it is also shown in FIG. 1 that a
mobile telephone 4 provided with a hands-free communication device 6 is disposed in the
automobile 7. In this case, the hands-free device 6 emits the transmitted signal from the remote
subscriber via the speaker 2 into the acoustic environment, ie into the car 7. This makes this
signal audible to the local subscriber 3 ("listen at loud volume"). A series of microphones 1 are
disposed inside the car 7 above the talker 3. This microphone 1 can be transmitted as an acoustic
input signal 9 to a remote subscriber who is connected to the mobile radio 4 via the radio
channel 5 via the hands-free device 6 and the mobile radio 4. Collect driver's 3 conversations and
other acoustic signals from the in-vehicle environment (hand-free calling). In the following
specification, both hand-free operation and listening at loud volume will be referred to
08-05-2019
3
simply as hand-free communication . The handsfree device 6 disposed in the vehicle may be a
component of the mobile radio 4 itself or may be a separate device connected with the mobile
radio 4. The other components that are typically used to operate the mobile wireless device are
not shown in the embodiment of FIG. 1 for clarity. There are fundamental problems with handsfree calling in a car. One of them is that the microphone 1 or the microphone group 1 picks up
another signal other than the talker's desired call signal from the acoustic environment. This
further signal is divided into two classes in terms of applicable technical solutions. One is an
acoustic echo of the speaker signal, and the other is local noise, such as noise due to engine, air,
tires, wind, rain, etc. These signals act as noise signals on the actual usage signal, ie on the
speech signal of the local subscriber. At the same time, the call quality of the telephone
connection formed via the mobile radio 4 and the radio channel 5 is decisively influenced. This
noise signal degrades the intelligibility of the used signal and impairs its quality even in the case
of the application of the call code method.
Therefore, the impairment due to the reduction of the transmission bit rate (e.g. GSM code) also
increases, which also reduces the identification rate of the call identification (which may for
example be provided for call control operations of the mobile telephone 4). For the original
radiotelephone communication conditions realized via the mobile radio 4, the first mentioned
effect of the two mentioned above is noticeable. In this case this action is only directly recognized
by the remote subscriber. It allows the remote subscriber to hear the echo of his own
conversation (becomes more pronounced above the signal transit time of about 30 ms, ie in the
case of GSM with a transit time of about 180 ms). The remote subscriber recognizes the effect of
local noise as a reduction in speech quality. The local subscriber 3 experiences a reaction only if
the feedback of the loudspeaker signal to the microphone 1 causes a stability problem in
radiotelephone communication. For example, under the call identification function such as the
call control of the mobile telephone 4, the effect of the failure becomes directly noticeable to the
local subscriber due to the deterioration of the identification rate. In addition to the desired
loudspeaker signal 2 and the corresponding echo, the local subscriber 3 hears further local noise.
Therefore, if the secondary noise level is high, the sense of hearing and intelligibility will be
impaired accordingly. Various concepts for noise reduction can be used to remove local noise
from the microphone signal. In this case, basically the following is always tried. That is, it is
attempted to analyze the use signal and the noise signal separately and to suppress the noise
signal without losing the use signal based on the basis of the required characteristics. FIG. 2
shows an embodiment of a hands-free device 6 which can be used, for example, in connection
with the mobile radio 4 shown in FIG. 1 (see FIG. 1). In the hands-free communication device 6
shown in FIG. 2, for example, in the hands-free communication device 6 consisting of
microphones 1 configured as a microphone array, the acoustic input signals 9 taken from the
individual microphones are analog / digital converters A high pass spike filter 11 for filtering the
input signal 9 is supplied via 22. The high-pass filter 11 belongs to the combining means 30 for
combining the input signal 9. In this case, the combining means 30 further comprise a device 12
08-05-2019
4
for optimum weighting of the input signal, a travel time compensation adjustment device 13, a
travel time evaluation device 17 and a summing device 15. The combined input signal 23 output
from the output of the combining means 30 is supplied to the adaptive filter 14.
This filter cooperates with the filter evaluation unit 19 to form the filter means 40 of the handfree unit 6. Another high pass filter 21 is provided for filtering the output signal 24 of the
adaptive filter 14. This filter outputs the output signal 22 of the handsfree device 6. In the
embodiment of the hand-free device 6 shown in FIG. 2, only the main parts related to the present
invention in the hand-free device 6 are shown, the other parts in a conventional hand-free device
are for clarity It is saved. The acoustic input signal 9 taken in using the microphone group 1 is
first converted into a digital signal using the conversion means 22 (A / D converter). The
following coupling means 40 are used for the following. That is, it is used to convert the digital
input signal obtained as described above into an output signal 23 which is as faultless as
possible. The quality of this signal is further improved by means of the following filter means 40.
For this, in the hands-free communication device 6 shown in FIG. 2, high-frequency filtering of
the input signal 9 subjected to digital conversion is first performed using the high-frequency
filtering filter 11. This can reduce the noise of the acoustic input signal 9 in a very simple
manner. This kind of high-pass filtering filter can be configured, for example, as a second-order
to fourth-order digital IIR (infinity impulse-response) filter. For example, 300 Hz is used as the
cut-off frequency of the high-pass filter 11. Frequencies below 300 Hz are not important for the
call signal to be processed for the mobile radio and the telephony based thereon. The high-pass
filtered input signal is optimally weighted using the device 12 in a known manner. In this case,
the high-pass filtered input signal output from the output side of the high-pass filtering filter 11
is also used for the evaluation of the transit time in the transit time evaluation unit 17 and for the
evaluation of the parameter 20 of the adaptive filter 14. Is also provided to the filter evaluation
unit 19 of FIG. In the embodiment of the hand-free device 6 shown in FIG. 2 prior to the coupling
of the input signal with the summing device 15, the transit time between the input signal 9 and
the microphone group 1 is furthermore achieved using the transit time compensator 13.
Compensation adjustments are made. For this, the travel time compensation adjustment device
13 has a plausibility check. This checking means is used to check whether the detected interval
of the input signal 9 origin to the caller exceeds a predetermined limit value.
Therefore, in the case of focusing on a caller, this type of validation device is checked whether
the detected distance to the caller exceeds a predetermined limit value. This limit value can be
selected according to the interior dimensions of the vehicle. For example, an interval of more
than one meter means that the caller should be outside the vehicle. Furthermore, the direction of
the input signal is also associated during this type of validation. This allows the location of the
caller on the data to be identified as being unlikely behind the microphone, ie in the direction of
the windshield. Similarly, the position of the caller on the data is also identified as being
08-05-2019
5
impossible on the left side of the position of the driver (in the case of the left handle). If such an
improbable talker position is identified, for example, the last valid talker position or a
corresponding prior evaluation value is used. This is ensured by means of the switching means
18 contained in the coupling means 30 in FIG. The microphone array 1 with subsequent travel
time compensation adjustment enables the separation of the active and the disturbed
components in the input signal 9 by its spatial selectivity. At the same time, if the interference
components in the microphone signals are not correlated with one another but are partially
eliminated during the addition, an additional array during focusing on the useful signal source
and averaging of the individual microphone signals Gain is obtained. The post-connected
adaptive filter 14 in the filter means 40 has the following purpose. That is, it has the purpose of
extracting the effective component of the combined signal 23 and further suppressing the
remaining fault signal. On the other hand, the autocorrelation function and the cross-correlation
function of the input signal 9 are calculated via the filter estimator 19 as parameters for the
adaptive filter 14 configured, for example, as a Wien filter. Furthermore, a check is made as to
whether coefficient overflow has occurred during the calculation of the autocorrelation function
and / or the cross correlation function. The adaptive filter 14 is not used when overflow occurs.
This can be produced by the switching means 18. In the signal 24 output from the output side of
the adaptive filter 14, the low frequency interference signal component included in the signal is
filtered by another high-pass filter 21. This is one that the adaptive filter 14 could not sufficiently
suppress. As a result, the output signal 22 of the hands-free device 6 has a further enhanced
signal quality.
08-05-2019
6