Device and method for suppressing a feedback

Translation of description of equivalent WO 2004047484 A1

Apparatus and method for suppressing a feedback

description

The present invention relates to audio systems and in particular to audio reproduction systems in live Ümgebungen.

In typical rock concerts highly dynamic exists to the effect that, for example, the singer moves on stage very much. The same applies to often for the guitarist. On the other hand, however, the speakers of such a demonstration environment are arranged statically. Therefore, it is inevitable that the singer together with his microphone, as well as the guitarist along with his attached to the guitar microphone sometimes gets a little closer to speakers and sometimes a little further arranges speakers off is connected. While the case is not a problem, in which a microphone is distant from a speaker, is the case where a microphone is placed very close to a loudspeaker, very problematic. After a high wear strengthening is present in the signal path from the microphone to the speaker performs a coupling of the loudspeaker signal to the microphone to the fact that the microphone / speaker system begins to oscillate. Such vibration is expressed as feedback at a certain frequency. It occurs whenever the amplitude and the phase nevertheless condition is satisfied. The specific phase condition that is currently best met determines the frequency which is typically relatively high, so that makes a feedback as a loud whistling noticeable. This pipe is not only unpleasant for the audience, but also for the artists. expressed signal theory, there is a strong time-varying channel from one or more speakers to one or more microphones.

Known feedback suppression techniques mix audible feedback sounds into the microphone and use filters to suppress a budding feedback.

Alternative feedback suppression techniques use a so-called pitch-shifting technique to move the feedback in inaudible parts of the spectrum, so that stable feedback tones are avoided.

While the first solution requires a short feedback to trigger a suppression, causing the other solution in some cases, a strange sound that makes as singing and voicing for Artists difficult.

In particular, in multi-channel systems, the two mentioned feedback suppression solutions are very difficult if not impossible.

The object of the present invention is to provide an improved concept for suppressing feedback.

This object is achieved by an apparatus according to claim 1, a method according to claim 10 or a computer program according to claim eleventh

The present invention is based on the finding that an effective feedback suppression can be achieved by providing a microphone signal which is a superposition of a desired signal and one of a speaker or multiple speakers originating feedback signal is then processed as possible before mixing or gain that the amount of feedback from the microphone signal is subtracted, so that after the subtraction only remains the useful signal.

Regardless of whether the feedback signal component, so if the microphone is located very close to the speaker in the event of an unfavorable channel is large or in the case of a favorable passage, so if the microphone is relatively far away from the speaker, is small, the feedback signal component is preferably continuously removed from the microphone signal. For this purpose, it is necessary to determine the feedback signal at the microphone component synthetically.

According to the invention this purpose, a marking operation is then performed, in that the signal emitted by the loudspeaker, can be detected. This is achieved in that either the microphone signal after the subtraction or to the microphone signal prior to the subtraction or the signal after mixing and amplification, ie, in the z. B. digitally present reproduction signal for a speaker, a test signal is embedded.

Furthermore, according to the present invention, a device for determining a characteristic of a transmission channel from the speaker to the microphone or directly for a feedback circulation from a microphone again back to itself using the received microphone signal, the superposition of the feedback signal and the useful signal , and using the known test signal, which has been embedded, are used.

A preferred procedure for determining the characteristic of the transmission channel in the area between the speaker and the microphone is to provide a cross correlation conduct between the microphone signal and the test signal. The cross-correlation, for example, directly delivers the impulse response of the channel between the betrachte- in speaker and microphone consideration. Alternative channel determination methods can also be used.

Using the determined characteristic of the transmission channel, a filter is adjusted, which filters the loudspeaker signal to obtain a filtered loudspeaker signal. the zeitva- variant channel is In other words, in a sense "simulate" from the speaker to the microphone to signal distribution synthetically calculate the fed into the microphone feedback, so that it is ready for the subtraction.

The present invention performs optimum feedback suppression when the channel changes only slowly. This is at concerts in view of caused by human performers movements very often the case. Even when an artist performs a very fast movement, so this rapid movement does not take long, such that movement on a single, fast moving again a slower moving or even a pause. The inventive system is capable of not only new to suppress feedback at the beginning of "settling", but also during the settling, to the effect that a feedback may already begun development yet again suppressed in the making, can that is subtracted out, be ,

On the other hand, a rapid movement often also means that the channel again changes to "good", so that the microphone back from the channel at a distance, which in turn, that perhaps the emergence located feedback subsides again without feedback suppression. the requirement of a time-invariant channel therefore is very low at the suppression concept of the present invention.

In the preferred embodiment of the present invention, the test signal is a pseudo-noise sequence can be produced with little effort, for example, easily using feedback shift registers, quickly and inexpensively, and if such a shift register is provided at several points available, readily reproducible. In particular, a plurality of shift register means that are falling follow generate such a pseudo-random, are initialized with the same seed value, or "germ". It is known that pseudo-noise sequences look like noise, but have a relatively large üblicherwei- se period. The intoxicating appearance of a pseudo-noise sequence is expressed in the frequency range considered in that the pseudo-noise signal has a white spectrum, such that all frequencies occur equally strong. If the dynamics of Mikrophonsig- is Nalles reasonably known, so this white pseudo noise signal can be directly incorporated, by ensuring that the level of blended pseudo noise signal is relatively small and does not cause audible interference, or resulting in only minor audible noise.

The effectiveness of the feedback suppression, ie to improve the channel simulation, it is preferred that the test signal, independent of whether it is a pseudo-noise signal or not, using a preferably of the already freed around its feedback proportion microphone signal, or using a from the amplified microphone signal, so to evaluate the loudspeaker signal, derived psychoacoustic masking threshold.

An addition of the thus evaluated test signal to the microphone signal or the loudspeaker signal causes the embedded test signal for the listener will not be audible, so that the listener of the constantly ongoing feedback suppression procedure does not notice anything.

In other words, in this case, the feedback suppression does not have any negative consequences as regards the vo viewers perceived reproduction quality. On the other hand, a test signal with the highest possible energy for an effective suppression, so for the most accurate determination of the impulse response of the channel between the loudspeaker and the microphone, so for precise simulation of the feedback portion, desirable in the loudspeaker signal. The maximum power is no compromise on audio quality achieved when the test signal is a pseudo-noise signal, that extends over the entire frequency range of interest, and is such a psycho weighted acoustic, that it is below the mark threshold of the loudspeaker signal. In signal components of the loudspeaker signal with high masking effect, the test signal is thus represented at a high energy, while signal components of the loudspeaker signal with low masking effect, for example, in tonal audio units, the test signal with relatively little or no energy is represented, to the extent that no Audioqualitätseinbu- KISSING for the listener arise.

It should be noted that, rather loud speaker signal passages are problematic in the case where the microphone is not directly in front of the speaker. Due to the fact that in such a loud Lautsprecherpas- usually say the masking threshold is relatively high, a significant test signal energy contained in such problematic loudspeaker signal components, which leads directly to the channel determination and thus the feedback cancellation takes place more accurately and thus more effective. The preferred for the present invention concept of using pseudo-noise test signal in conjunction with a psychoacoustic weighting or staining of the pseudo-Neuseh test signal therefore results that precisely in the case where a well-functioning feedback suppression used is, as in the case of noisy signals, also a good channel destination with high signal / noise ratio can be performed. The much needed in such a case good feedback ment oppression is inventively also provided.

The present invention is particularly suitable for multi- channel environments, where multiple microphones and multiple speakers. The use of embedded into the individual microphone signals different test signals, which are preferably orthogonal to each other, and the use of a cross correlation means for determining each relevant channel cause can be calculated for each microphone, the optimum amount of feedback. Thus there will be a flexible and at the individual microphone signals exactly matched feedback suppression, since each channel is simulated individually.

It can be seen that for the case in which a plurality of microphones and a plurality of loudspeakers are provided at different locations, the computing power can be preferably lent considerable for channel determination using a cross-correlation. However, this is not problematic, since a typical sound system, such as comprising a PA system, a mixer with considerable dimensions and considerable costs, and in such a setting some digital signal processors properties for calculating the channel and to suppress the feedback of play with respect will fall not much significant to the total cost of the plant.

On the other hand, the present invention causes an efficient feedback suppression without negative consequences for the audience on the one hand and in particular to the artists on the other hand, typically almost negligible cost based on the total system. In particular, emphasis is placed on that the artists are not disturbed in their artistic expression, such that they "eingetunte" z. B. audible Rückkopplungsunterdrü- hear ckungstöne or that in case of pitch-shifting perceived by the artists signals a different pitch than they have been, for example, sung by the artists. Although this known feedback suppression already nuances are sufficient in terms of pitch shift, these are still nuisances for artists who are likely to restrict him in his artistic expression. On the other hand, it is just the artist who ultimately determines which facility must be provided for him. A market acceptance of the concept of the invention is therefore to be expected, since the inventive feedback suppression concept the artists not harassed and even allowed him maximum freedom of movement, so that it without fear of unwanted feedback tones, the entire stage area can use for artistic expression, independent since - of whether he comes close to a feedback-prone speaker component or not.

Depending on the embodiment, the test signal can be directly embedded in the loudspeaker signals, before the analog / digital conversion and acoustic reproduction. In this case, the adjustment to the psychoacoustic characteristics of the loudspeaker signal will be best because the psychoacoustic model of the loudspeaker signal will be directly for meaningful, what an audience hears or not.

Embedding the loudspeaker signal also has the advantage that actually transfer functions can be simulated individually from each speaker to each microphone and used for feedback suppression. This alternative of the invention leads to a better sound quality for the listener, but requires greater computing power to the effect that if, for example, three microphones and three speakers are present, already determined nine different transmission channels with respect to the properties to be replicated with typically FIR filters and used for subtraction wherein prior to subtraction of the actual feedback insgesamten ment signal an addition of the delivered in the described case of three speakers three individual simulated feedback signals must be performed.

Another alternative of the present invention is to embed the test signal into the modified microphone signal, ie after the subtraction, that is, before the microphone signals are mixed and amplified to obtain an embedding signal. Embedding the signal is simultaneously used to be filtered and to supply the filtered signal to the subtracter. The psychoacoustic model is here preferably calculated on the basis of the modified microphone signal to th the masking threshold for optimal embedding to obtain.

The information on the psychoacoustic masking threshold may, however, also derived from the individual loudspeaker signals and the corresponding Einbettungsein- direction that lies before mixing / amplification, are supplied, so that there is better control of the test signal.

As has been explained, is the test signal on the one hand not be audible on the other hand be provided with the highest possible energy. If a psychoacoustic model derived from a signal that the speaker signal is not directly equivalent, but only approximately corresponds to the energy of the embedded test sig- signal is maintained by a certain safety margin below the psychoacoustic masking threshold, which, although the deterioration of audio quality in derogation, however, a poorer signal / noise ratio in the Ü bertragungskanalbesti flow and thus could lead to a poorer feedback suppression.

On the other hand, in this case, not as many channels to be calculated, so that this alternative calculation time poorer issued can be formed and thus can be used inexpensively in particular for smaller reproduction equipment or minimum-reproducing systems.

Again alternatively, the test signal can be inserted in the microphone signal prior to the feedback portion subtraction. If the feedback percentage is calculated exactly as the embedded test signal is the Rückkopplungsanteil- subtraction relative "without prejudice" to survive, so that this case can be similarly considered to the case in which the test signal is already embedded in the modified microphone signal.

Preferred embodiments of the present invention are explained in detail below with reference to the accompanying drawings. Show it:

Fig.la a preferred embodiment of the present invention in a multi-channel environment with embedding on the microphone side;

Fig.lb an alternative embodiment of the inventive concept with Rückkopplungsunterdrückungskon- embedding on the microphone side;

Figure 2 is an alternative embodiment of the present invention with embedding on the speaker side;

3 shows a schematic diagram of a transmission channel; and

4 is a schematic summary of the method for calculating an impulse response of the transmission channel shown in Figure 3 using a cross-correlation. Fig.l shows a preferred embodiment of the present invention in a multi-channel setting, wherein a plurality of microphones 10, 11, 12 and a plurality of loudspeakers 13, 14, 15 are arranged. phonseite between the microphones on the micro and the speakers on the speaker side is a signal processing device 16 which is any sound system that perform among other things, a mixture or amplification of the audio signal, which is fed by the microphones, can.

Signals from the three speakers 13, 14, 15 are superimposed on each microphone and form a feedback signal fi (t) for each microphone. The loudspeaker signals of the speakers 13, 14, 15 are transmitted through a free-space transmission channel 17, which can be defined such that the first microphone, a first transmission channel hi is defined by the three speakers that the second microphone 11, a second of the three speakers transmission channel h₂ it is defined that the three speakers for the third microphone 12, a third Ü bertragungskanal h₃ is defined.

In the embodiment shown in Fig.la to a test signal by using an embedding device 20, 21, 22 is embedded in a modified microphone signal to obtain for each microphone channel at the output of means 21, 21 and 22 a respective embedding signal. In particular, in the modified microphone signal of the first microphone 10, a first test signal pi is turned on a bed in order to obtain a first integration signal. In the modified microphone of the second microphone 11 of the signal is a second test signal p₂ embedded in order to obtain a second signal embedding. Finally, in the modified microphone signal of the third microphone 12, a third test signal p is embedded in order to obtain a third signal embedding. In order to come from a microphone signal at the output of each microphone 10, 11, 12 to a respective modes izierten microphone signal, each microphone is also a subtractor 30, 31, 32 associated. The subtracting means is configured to, subtract a simulated feedback portion that is in the ideal case, equal to the received feedback from a microphone fraction fi (t) from the microphone signal. So that in the ideal case at the output of each subtracter 30, 31, 32 a modified microphone signal is present, the sι the original useful signal (t), s₂(T) and s₃(T).

To simulate the feedback of play each microphone is a separate channel simulation filters 40, 41, 42 associated with the first simulation filter 40 is formed to have the same channel impulse response hι to have (t), as shown in block 17, wherein lb in Fig. Of representation associated with not only the free space channel in block 17, but also the transfer function by the block mixture / reinforcement 16. at this point it also strikes out that the simulated channel impulse response also already includes the necessary delay.

the second channel simulation filter 41 is analogous embodied det to the same channel impulse response h₂(T) to have, as outlined in block 17 (including mixing / gain). Finally, the third simulation filter 42 is formed to the same channel impulse response h₃is indicated reasonable (t) to have, as in block 17 (including mixing / gain).

The channel impulse responses for adjusting the simulation filter 40, 41, 42 are determined in the respective devices 50, 51, 52 for determining a characteristic of a transmission channel. To this end, receives the first device 50 for detecting the test signal, which has been fed into the modified microphone signal of the microphone 10th Analogously, replaced by the second means 51 for determining the Test signal p₂Which has been used in the device 21 for embedding. Finally, to determine the third microphone, the device 52 obtains the same test signal p₃Which has been fed into the modified microphone signal of the third microphone.

In a preferred embodiment of the present invention are the three test signals p_x, p₂, p₃ each pseudo-noise sequences that are orthogonal to each other, so that it by the means 50, '51, 52 for determining conducted cross-correlation with the respective test signal pi, p₂, p₃ can be distinguished from the other test signals provided with the modified microphone signals and loudspeaker signals thus radiated.

A cross-correlation as the first microphone 10 with the pseudo-noise sequence pi of the microphone signal will result in the provided with the pseudo-noise sequences modified microphone signals from the second and the third micro- phone are out correlated, so that only the fact of first microphone signal is to be subtracted feedback portion, which is problematic in terms of generating a feedback subtracted. It should be noted that typically, when the device 16 is not significant microphone / speaker assignment changes are carried out at short time intervals, the feedback signals from the two other microphones 11, 12 here are not critical, since such feedback signals in the signal processing path from said first microphone leads to the three speakers 13, 14, 15, in terms of a feedback generation are not critical.

In the embodiment of the present invention shown in Fig. La further embedding the signal is to Filterparameterberech- voltage for each microphone channel uses filtered and this microphone channel. In particular, the filter 40 for generating the filtered signal to be supplied to the device 30, the embedding signal at Output of the device 20. Accordingly, the filter 41 is fed with the signal from the embedding device 21st In addition, the filter 42 is fed with the signal from the embedding device 22nd

At this point it should be noted that the embodiment shown in Fig. La only the signal is subtracted, which is problematic for a feedback. Problematic for a feedback via the first microphone is only the (former) signal from the first microphone, the (later) is injected again. So it is in this case, no matter from which (s) Speaker (s) the first microphone signal is reproduced. The calculated by correlating the first microphone signal with the first test signal channel corresponds to a "feedback circulation", ie a circulation from the microphone, through mixing / amplification, one or more speakers and the free space channel back to the microphone (including the transfer characteristic of the microphone actually used) . it should also be pointed out back that the determined impulse response hi also includes "automatic" that has occurred in the feedback round trip delay, so that this no further precautions must be taken. Further, in this case the situation is transparent as to that for the spectral coloring the psychoacoustic masking threshold of the input in the signal embedding device can be used.

Alternatively, a loudspeaker signal could be recycled and fed into the filter. Depending on the principal illustration of a microphone to a speaker the association is such that the loudspeaker signal 13 is filtered and returned to the first microphone 10, in principle arbitrary. When the first microphone, the dominant association rather is the speaker 2, the speaker signal of the speaker would be 14 returned through the simulation filter 40 for the first microphone. The assignment of the loudspeaker signals for the microphones is thus only be seen as an example in Fig. la and may also vary from time to time depending on the mix in the signal processing device 16.

That shown in Fig. Lb to Fig. La alternative embodiment of the present invention differs from the embodiment shown in Fig. La to the effect that the loudspeaker signals are returned and not embedded signals, and in that the signals of the different speakers 13, 14, 15 in a summation means 23 are added up, and in that then the speaker sum signal is filtered with the corresponding different simulation filters 40, 41, 42, to generate the three synthesized feedback units that are the corresponding subtractors 30, 31, 32 which as shown in . lb is shown Fig. In this embodiment, it is assumed that overlie the loudspeaker signals of all speakers in the transmission channel 17 and, for example, a resultant feedback signal fι (t) lead, which is modified from signal components of the first, second and third speaker to a correspondingly definable transfer function. For the transfer of the sum signal of the three speakers that overlap in space transmission channel, for the first microphone a first transfer function hi is defined. For the transmission of the sum signal to the second microphone 11 is a transfer function h₂ defined and finally for the transmission of the sum signal to the third microphone 12, a resulting transfer function h Ü₃ defined.

These transfer functions hi, h₂, H₃ be in the devices 50, 51, 52 again preferably associated by cross-correlation with the appropriate, a specific microphone pseudo Neuseh sequence pi, p₂ or p₃ determined. The execution of the subtracters 30, 31, 32, the embedding means 20, 21, 22 as well as the simulation onsfilter 40, 41, 42 is designed as in the embodiment described with reference to FIG. la.

Description will be made on the schematic shown in Fig. 2 further embodiment. In contrast to the embodiments shown in FIGS. La and lb, the embedding of the test signal does not occur on the microphone side, but on the speaker side. This not only three different channels, but nxm different channels can be defined, where n is a number of speakers is greater than or equal to 1, and where m is a number of microphones is greater than or equal. 1 By correlation of the first microphone 10 of the output signal lose with the first test signal pi of the channel can be calculated from the loudspeaker 1 to the first microphone Ml, which is denoted by hu. By correlating the first microphone 10 of the output signal using the second pseudo noise sequence p₂ , the channel from the second speaker 14 for the first microphone 10 connected hι₂ is referred to, are calculated. Analogously, by correlating the first microphone 10, the microphone signals with the third pseudo-noise sequence p₃ the channel are simulated by the LS3 speaker for the first microphone Ml, which with hι₃ is designated.

Analogous to this procedure may be used for the outputs of the microphones 11 and 12, as is indicated to determine the basis of the devices 50, 51, 52nd The devices 50, 51, 52 are thus able to own channel to be calculated for the channel from each speaker to each microphone transfer function, with each speaker signal can be folded, which takes place in the simulating filters 40, 41, 42, and then, for example, within the subtraction means 30, 31 and 30, or in an upstream block of the three-channel output signals for each microphone to calculate the resulting amount of feedback by addition, to arrive at a resultant feedback portion. This is then of the fed into a respective microphone feedback signal fi (t) is subtracted to arrive at a modified microphone signal for each microphone, wherein each channel is selectively taken into account.

Depending on the embodiment, means may be executed for determining completely parallel 50 to a certain extent simultaneously hu the channel impulse responses, hχ₂ and hχ₃ to calculate. However, the appropriate device could be performed serially, in which case hu with regard to an optimal temporal synchronicity of the three channels, h_i2, H_i3 among themselves a buffer is preferred. At the expense of a certain error, however, could be dispensed with such an intermediate storage, so that the three mutually related impulse responses of each of the speakers 13, 14, 15 10, although not related to the first microphone to the same period of time, but on consecutive periods, which but then is not harmful when the signals change in an environment not too quickly, either with respect to the time required for correlation time.

Also, the filter means 40, 41, 42 may be performed serially or in parallel, with parallel execution provides the best results, such that a separate single simulation filter is provided for each possible channel the potential in Fig. 2 channels, such that the filter means 40, for example, actually comprises three individual simulation filter, hu whose filter coefficients using the corresponding channel impulse response, hι₂, H_i3 be adjusted. The adding up of three simulated feedback proportions from each speaker in a resulting feedback share could therefore be used in the filter device 40 immediately after the calculation of the corresponding impulse responses and the convolution of the loudspeaker signals with these impulse responses proceed. In the embodiment shown in Fig. 2 should, as with the in Figs. La and lb embodiments shown, the three test signals p_lf p, p₃ be each other as well as possible orthogonal. This condition is easy and safe to reach by pseudo-noise sequences, which property is lost even by a psychoacoustic filtering the test signals prior to embedding th.

In the embodiment shown in Fig. 2 it is noted that a loudspeaker signal is the signal that actually hears a listener. With respect to a non-audible embedding the test signals in the loudspeaker signals the embedding is therefore can be performed best when the loudspeaker signals are used for calculating the psychoacoustic masking thresholds.

So could wherein in Fig. Lb illustrated embodiment, a psychoacoustic model based on the respective loudspeaker signals 13, 14, are charged 15 and to be embedded in the respective microphone signals in the devices 20, 21 and 22 used. Thus, in the psychoacoustic model readily gains are taken into account, that take place between a microphone and a speaker in the establishment 16th However, if the device 16 as in the case of a mixing process a significant addition / subtraction or other processing of the microphone signals is performed so that a loudspeaker signal not only substantially reproduces the output of a single microphone, but reproduces the output signals of several microphones, so a embedding a test signal using the psychoacoustic masking threshold see inaccurate. This is because, on the one hand directly a single loudspeaker signal for calculating the psychoacoustic masking threshold can not be hergenommen, and on the other hand immediately see a microphone for calculating the psychoacoustic masking threshold can not be hergenommen. After the mixture in the mixer 16 but deterministically occurs, it is in such a case, preferably, a psychoacoustic masking threshold of the corresponding to calculate mixing simulated signal to obtain a loudspeaker signal, in which, if the loudspeaker signal is the combination of multiple microphone signals, the test signals from several microphones are differently or equally embedded, the test signals, but overall, the psychoacoustic masking threshold of a loudspeaker signal substantially follow so that an embedding with maximum power is achieved while no or only negligibly small audio quality losses are caused.

Below is summarized as the impulse response h (t) of a channel is determined by cross-correlation. For this purpose, the channel is acted upon with a time-discrete test signal p (t). The channel is on the output side, a reception signal y (t) from which, as is known, corresponds to the convolution of the input signal and with the channel impulse response. For the subsequent explanation of a procedure for determining the cross-correlation based on Fig. 4 is transferred to a matrix notation. Exemplary is a channel impulse response with only two values ​​h₀ and hi assumed without loss of generality. The channel impulse response h₀Hi can as a channel impulse response matrix H (t) can be written that has the band structure shown in Fig. 4, the remaining elements of the matrix are filled with zeros. In addition, the excitation signal p (t) is written as a vector, where it is assumed here that the excitation signal without loss of generality, only three samples p₀, Pi, p₂ Has.

It can be shown that the convolution shown in Fig. 3 to that shown in FIG. 4 matrix-vector multiplication corresponds, so that a vector y is the output signal results. The cross-correlation can be used as expected value E {...} of the multiplication of the output signal y (t) with the conjugate-complex transposed excitation signal p^{* τ} to be written. The expected value is calculated as the limit for N to infinity on in Fig. 5 illustrated summation of individual products for different excitation signals p ^. The multiplication and subsequent summation yields the cross-correlation matrix which is shown in Fig. 4 the top left, wherein this same weighted by the RMS of the excitation signal p, the at σ_p² is shown. For immediate obtaining the channel impulse response h (t) is, for example, taken the first line of the channel impulse response matrix, whereupon the individual components by σ_p² be divided to directly obtain the individual components of the channel impulse response ho, hi.

If, instead of a white excitation signal p (t) using a spectrally colored excitation signal, so increase the spectral coloring can be represented by a digital filtering, the filter is described by a filter coefficient matrix Q. In the equation shown in Fig. 4 in the last line also results in the output side, the correlation matrix H, but now weighted with the expected value on Q x Q^H, By dividing the individual impulse response coefficients h₀, Hi by the expectation value on Q x Q^H, That is by taking into account the coloring filter, for example, in the device 50 the channel impulse response can be determined in terms of their individual components for determining a property of the transmission channel of Fig. La, lb or 2 can immediately.

It should be noted that the cross-correlation concept for calculating the impulse response is an iterative approach, as can be seen for the expected value of the summation approach illustrated in Fig. 4. The first multiplying the response signal with the conjugate-complex-transposed excitation signal already provides a first very rough estimate of the channel impulse response, which is getting better with each subsequent multiplication and summation. If the matrix H (t) calculated by the iterative summation approach, it turns out that the top left set in Fig. 4 to zero elements of Ribbon matrix H (t) gradually go to zero, while remaining in the center, so the volume of the matrix, the coefficients of the channel impulse response h (t) and take certain values. Once again it should be noted that it is not necessary to calculate the whole matrix. It is sufficient to calculate only z. B. a row of the matrix H (t) to obtain the overall channel impulse response.

It should be noted that the inventive concept is not limited to the method for calculating the cross-correlation described with reference to 4 of FIG.. All of the other method for calculating the cross-correlation between a measurement signal and a response signal are also employable. Other methods for determining an impulse response in place of the cross-correlation may also be used.

It should be noted that the pseudo-noise sequences used depends in terms of their length should be dimensioned gig of the expected impulse response of the channel concerned. Thus, for larger a- kustische environments quite impulse responses with the length of a few seconds conceivable. This fact must be recognized in selecting an appropriate length of the pseudo noise sequences for correlation calculation.

Depending on the circumstances, the inventive method in hardware or in software may be implemented. The implementation may be on a digital storage medium, particularly a floppy disk or a CD with electronically readable control signals, which can cooperate with a programmable computer system such that the method is performed. Generally, the invention thus also in a computer program product stored on a machine readable carrier, the program code for performing the inventive method when the computer program product runs on a computer. In other words, the invention may thus as a A computer program can be realized with a program code for performing the method when the computer program runs on a computer.

At this point, it should again be noted that the inventive concept for arbitrary numbers of microphones and arbitrary numbers of speakers can be used. This means of course that the inventive concept can be used advantageously for a loud- speaker and a microphone already. This follows directly from the FIG. La, lb and 2, when the second and the third microphone 11, 12 and the second and third speakers 14, 15 are ignored and also the issues raised by these signals blocks are thought away.

It should be further noted that the embedding of the test signal does not necessarily follow into the modified microphone signal or the loudspeaker signal to achieved, but that also a embedding the test signal can be done in the microphone signal before the corresponding subtraction means, although the embedding of the test signal is preferred according to the subtractor. This is because in case of a not so favorable channel impulse response calculation and thus in the case of a not particularly precise synthesized feedback share the embedded test signal could be may be damaged by the subtraction of a not exactly matching feedback component u., Leading to a further aggravation of the channel simulation should lead by the devices 50, 51, 52nd

In preferred embodiments of the present invention in a multi-channel setting a non-audible wideband signal is thus embedded in each microtiter rophonsignal. This signal is adapted in terms of its spectral envelope adaptive to the recorded sound, wherein a psychoacoustic model can be used, which may in principle be as desired and can be also calculated based on the time domain data or based on frequency domain data. As a wideband signal, a pseudo-noise sequence is preferred because with such a sequence, orthogonality between multiple sequences can be readily achieved.

For each microphone, the recorded signal is compared before embedding with the pseudo-noise signal and uses to calculate the acoustic characteristics of all speakers to the corresponding microphone. As a comparison operation, a cross-correlation is preferred that when the depicted in FIG. 4 iterative algorithm is used, can be calculated calculation time-inexpensive with an arbitrarily scalable accuracy. In particular, the scalability provides the possibility of providing a fast but relatively coarse calculation for spe- zielJLe___j3ituationen, for example in a rock band, when a lot of movement on the stage there, while for other application scenarios, such as a rock band, in which the artistes e / her are static, such as scaling to a larger number could be performed by iteration values, since the individual channels less time-variant

Using a corresponding channel an in- verse filter is applied in order to suppress undesired components. The inverse filter is implemented according to the present invention by simulating filter and the corresponding associated subtracters. The use of microphone signals enables storage of spectrally shaped PNS-signals so that interference is avoided with the original sound signals, and that a psychoacoustic model for calculating the spectral shaping only once has to be calculated, and not in the appropriate means for detecting again must be calculated. Alternatively, as has been illustrated with reference to FIG. 2, a unique embedded PNS signal in the signal from each speaker. This approach of embedding on speaker side allows the measurement of a path from each loudspeaker to each microphone. A rejection filter is used separately for each speaker, creating a better sound quality is achieved, but at a cost higher computational complexity, however in view of the total cost of medium to larger sound systems not likely weighing particularly heavily.