EP1445761A1 - Apparatus and method for operating voice controlled systems in vehicles - Google Patents

Abstract

A process for a speech support system such as a communication or intercom unit in a motor vehicle comprises a microphone (30) and loudspeaker (31) with a band pass filter (32) between them which is adjusted according to a comparison of signal (S) power between a search frequency and an integral multiple frequency or power comparison. Independent claims are also included for the following: (a) a device as above comprising decision logic; (b) an operating process as above using feedback;and (c) an operating device using feedback.

Description

The invention relates to a method and a device for the operation of voice-assisted systems, such as communication and / or voice / intercom devices in motor vehicles, in which a microphone arrangement Voice signals recorded and passed to at least one speaker become.

Methods of this kind are used in motor vehicles for voice-assisted Intercom operation or in support of voice input electronic or electrical assemblies used. The fundamental The problem here is that in the motor vehicle depending on the operating condition appropriate background noise is present. This covers the voice commands. Speech and intercom systems in motor vehicles are predominantly large Vehicles, minibuses and the like advantageous. You can, however, also at normal passenger cars are used. When using voice-controlled input units for electrical components in the vehicle is the Suppressing the background noise or filtering out the voice command yet really important.

Thus, from EP 0078014 B1 a speech recognition device for a motor vehicle in which in the amplifier system of the speech recognition device via Sensors is reported or fed, whether the engine is in operation and / or the Vehicle moves. Thereafter, then a level influencing directed with the trying to filter out the voice command from the background noise.

From WO 97/34290 a filtering is known in the periodic interference signals be filtered out by determining their period and using generator is outinterfered, so that the speech signal is left over.

From DE 197 05 471 A1 is known, a speech recognition using a To support transversal filtering.

From DE 41 06 405 C2 a method is known in which a noise subtraction from the voice signal using a plurality of microphones. A Intercom with multiple microphones also discloses the DE 199 58th 836 A1.

DE 39 25 589 A1 discloses the use of a multiple microphone arrangement known, wherein when used in the vehicle one of the microphones in the engine compartment and another is arranged in the passenger compartment. Then there is a subtraction both signals. The disadvantage here is that only the engine noise or the actual operating noise of the vehicle itself from the total signal in Passenger compartment is deducted. Specific background noises are hereby left unconsidered. Likewise, a feedback suppression is missing. All over where microphones and loudspeakers are arranged in acoustically coupled proximity, it happens that the acoustic signal decoupled at the loudspeaker is again in the microphone is fed back. It comes to a so-called feedback and a following override. Solutions to avoid such Overloading are known from EP 1 077 013 B1, WO 02/069487 A1 and WO 02/21817 A2 known.

The invention is therefore based on the object, a method and a device of the generic type in such a way that the verbal Communication of the occupants of a vehicle is improved.

This object is achieved by a method according to claim 1 and a device solved according to claim 28. This is to operate a voice-assisted Systems, such as a communication and / or voice / intercom in a motor vehicle, with at least one microphone and at least one speaker for reproducing a signal generated by means of the microphone as well as between the bandpass filter arranged on the microphone and loudspeaker depending on a comparison of the power of the microphone signal generated at an examination frequency with the power of the means of Microphone generated signal at least one substantially integer Multiples, ie a substantially harmonic, the examination frequency or depending on a comparison of the power of the microphone generated by the Signal at an examination frequency with the power of the microphone generated signal at the examination frequency at least one earlier Time set. The examination frequency is one or more Frequencies of the signal generated by the microphone in question. In an advantageous manner Embodiment of the invention is the frequency as the examination frequency selected, in which the power of the signal generated by the microphone in the substantial maximum. Alternatively, several frequency components with large Services selected as examination frequencies.

In a further advantageous embodiment of the invention, the bandpass filter is both depending on a comparison of the power of the microphone generated by the Signal at the examination frequency with the power of the microphone generated signal at least a substantially integer multiple of the Examination frequency as well as depending on a comparison of the performance of the by means of the microphone generated signal at the examination frequency with the Power of the signal generated by the microphone at the examination frequency set at least at an earlier date.

In a further advantageous embodiment of the invention, the bandpass filter is so set it to the proportion of the signal generated by the microphone with a Locking frequency (only) locks when the power of the microphone generated signal at the examination frequency by more than an upper Limit is greater than the power of the signal generated by the microphone at the first harmonic of the examination frequency. Blocking frequency in the sense of Invention may also be a frequency range and not just a single frequency.

In a further advantageous embodiment of the invention is the upper limit between 20 and 40dB. Distributively, the upper limit is substantially 30dB.

In a further advantageous embodiment of the invention, the bandpass filter is so set it to the proportion of the signal generated by the microphone with the Locking frequency does not lock when the power of the microphone generated by the Signal at the examination frequency less than a lower limit is greater than the power of the signal generated by the microphone at the first Harmonics of the examination frequency.

In a further advantageous embodiment of the invention is the lower limit between 5 and 20dB. Advantageously, the lower limit is substantially 12dB.

In a further advantageous embodiment of the invention is by means of a comparison of Power of the signal generated by the microphone at the examination frequency with the power of the signal generated by the microphone at the Examination frequency at least earlier decided whether the Power of the signal generated by the microphone at the examination frequency increases exponentially.

In a further advantageous embodiment of the invention, the bandpass filter is so adjusted that the proportion of the signal generated by the microphone in the Locking frequency locks if it is decided that the power of the means of the Microphone at the examination frequency increases exponentially.

In a further advantageous embodiment of the invention, the bandpass filter is so set it to the proportion of the signal generated by the microphone with the Locking frequency (only) locks when the power of the microphone generated signal at the examination frequency longer than a first response time is greater than a threshold, with the first response time more advantageous greater than essentially 750ms.

In a further advantageous embodiment of the invention, the power at more than an examination frequency and set the bandpass filter such that it only the proportion of the signal generated by means of the microphone with the blocking frequency locks when the power of the signal generated by the microphone at a Examination frequency longer than a second response time is greater than the power the signal generated by the microphone at any other examination frequency, wherein the second response time is advantageously greater than substantially 750ms.

In a further advantageous embodiment of the invention, the setting of Bandpass filter with respect to the examination frequency earliest after expiration of a Minimum dead time repeated. The minimum dead time is advantageously 200ms to 300ms.

In a further advantageous embodiment of the invention, the bandpass filter is so adjusted that it the proportion of the signal generated by the microphone at a Frequency range around the blocking frequency blocks if after expiration of a Repetition time, which is greater than the minimum dead time, the performance of the means of Microphone generated signal at the examination frequency by more than the upper Limit is greater than the power of the signal generated by the microphone at the substantially first harmonic of the examination frequency and / or when It is decided that the power of the signal generated by the microphone at the examination frequency increases exponentially.

In a further advantageous embodiment of the invention, the bandpass filter is so adjusted that it the proportion of the signal generated by the microphone at a enlarged frequency range around the blocking frequency blocks when after expiration of a Repetition time, which is greater than the minimum dead time, the performance of the means of Microphone generated signal at the examination frequency by more than the upper Limit is greater than the power of the signal generated by the microphone at the substantially first harmonic of the examination frequency and / or when It is decided that the power of the signal generated by the microphone at the examination frequency increases exponentially.

The aforementioned object is also achieved by a method according to claim 30 and a device according to claim 53 solved. It is used to operate a voice-assisted system, such as a communication and / or voice / intercom device in a motor vehicle, with at least one microphone and at least one speaker for playing a generated by means of the microphone Signal and one between the microphone and the speaker arranged Bandpass filter the power of the signal generated by the microphone at least three examination frequencies determined by evaluating the performance of the detected by the microphone signal at the examination frequencies is whether there is feedback, and the bandpass filter is set such that it is a part of the microphone that is around a cutoff frequency generated signal locks when it is determined that feedback exists.

Notch frequency in the sense of the invention (and thus also in the sense of all claims) the examination frequency at which the power of the microphone generated signal is maximum. In an advantageous embodiment of the invention is the Rejection frequency but the examination frequency added with a correction frequency, where the power of the signal generated by the microphone is maximum, i. to the examination frequency at which the power of the microphone generated by the Signal is maximum, a correction frequency is added. This correction frequency is Advantageously, depending on the power of the generated by means of the microphone Signal at the examination frequency at which the power of the microphone generated signal is maximum, as well as the power of the microphone generated by the Signal at least one, especially immediately, next to this Examination frequency lying examination frequency formed.

- fkorr

die Korrekturfrequenz,

- fdist

der Abstand zwischen der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und einer die größte Leistung aufweisenden Untersuchungsfrequenz unmittelbar neben der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

- Pmax

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist (Pmax ist also die Leistung bei der Untersuchungsfrequenz, die größer ist als die Leistung jeder anderen Untersuchungsfrequenz),

- Pmaxneigh

die Leistung des mittels des Mikrofons erzeugten Signals bei der die größte Leistung aufweisende Untersuchungsfrequenz unmittelbar neben der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und

- sign

ein Vorzeichen

ist, wobei sign positiv ist, wenn die die größte Leistung aufweisende Untersuchungsfrequenz unmittelbar neben der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, größer ist als die Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und wobei sign sonst negativ ist.Thus, the correction frequency, for example, according to fkorr = sign * fdist * Pmaxneigh / (Pmax + Pmaxneigh) be formed, where

- fkorr

the correction frequency,

- fdist

the distance between the examination frequency, at which the power of the signal generated by the microphone is at its maximum, and one of the highest power, the examination frequency immediately adjacent to the examination frequency at which the power of the signal generated by the microphone is maximal,

- Pmax

the power of the signal generated by the microphone at the examination frequency at which the power of the signal generated by the microphone is maximal (Pmax is the power at the examination frequency which is greater than the power of any other examination frequency),

- Pmaxneigh

the power of the signal generated by means of the microphone at the highest power examination frequency immediately adjacent to the examination frequency at which the power of the signal generated by the microphone is maximum, and

- sign

a sign

where sign is positive when the highest power examination frequency immediately adjacent to the examination frequency at which the power of the microphone generated signal is maximum is greater than the examination frequency at which the power of the microphone generated signal is maximum , and where else sign is negative.

This is explained in more detail with reference to the following example:
192 examination frequencies f ₁ , f ₂ , .... f _{192 are} assumed. f ₁ is equal to 40Hz. fdist is 40Hz for all exam frequencies. In addition, for the powers of the signal generated by the microphone at the examination frequencies f ₁ , f ₂ , .... f ₁₉₂ :
P (f ₁ , f ₂ , ..... f ₉₄ ) = 1
P (f ₉₅ ) = 4
P (f ₉₆ ) = 16
P (f ₉₇ ) = 2
P (f ₉₈ , f ₉₉ .... f ₁₉₂ ) = 1
Then:
fkorr = (-) * 40Hz * 4 / (16 + 2) = -8Hz
The examination frequency at which the power of the signal generated by the microphone is maximum is thus 3840 Hz and the blocking frequency 3832 Hz.

- fkorr

die Korrekturfrequenz,

- Δf

der Abstand zwischen zwei Untersuchungsfrequenzen,

- Pmax

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

- Pneighright

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz unmittelbar oberhalb (also 'rechts' neben) der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und

- Pneighleft

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz unmittelbar unterhalb (also 'links' neben) der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

ist.It has proven to be advantageous at least in certain embodiments, the correction frequency according to fkorr = Δf * (Pneighright Pneighleft) / (Pmax + | Pneighright Pneighleft |) to form, being

- fkorr

the correction frequency,

- Δf

the distance between two examination frequencies,

- Pmax

the power of the signal generated by the microphone at the examination frequency at which the power of the signal generated by the microphone is at a maximum,

- Pneighright

the power of the signal generated by the microphone at the examination frequency immediately above (ie 'right' next to) the examination frequency at which the power of the signal generated by the microphone is at a maximum, and

- Pneighleft

the power of the signal generated by the microphone at the examination frequency immediately below (ie 'left' next to) the examination frequency at which the power of the signal generated by the microphone is maximal,

is.

On the basis of the above numerical example, in this case: fkorr = 40Hz * (2-4) / (16+ | 4-2 |) = -4.44Hz The examination frequency at which the power of the signal generated by the microphone is maximum is thus 3840 Hz and the blocking frequency 3835.56 Hz.

In a further advantageous embodiment of the invention, the distances between at least part of the examination frequencies or all examination frequencies equidistant.

In further advantageous. Embodiment of the invention is a presence of Feedback only detected when the power of the microphone signal generated at the examination frequency at which the power of the means of the Microphone generated signal maximum is greater by more than an upper limit is the power of the signal generated by the microphone at the first Harmonics of this examination frequency, the upper limit advantageously between 20 and 40 dB, in particular at substantially 30 dB, lies.

In a further advantageous embodiment of the invention is not present Feedback detected when the power of the microphone generated by the Signal at the examination frequency at which the power of the microphone generated signal is less than a lower limit greater than the power of the signal generated by the microphone at the first harmonic this examination frequency, the lower limit advantageously between 5 and 20dB, especially at substantially 12dB.

In a further advantageous embodiment of the invention is a presence of Feedback (only) then detected when the power of the microphone signal generated at the examination frequency at which the power of the means of the Microphone generated signal is maximum, at least approximately, exponentially increases.

In a further advantageous embodiment of the invention is a presence of Feedback (only) then detected when the power of the microphone generated signal at least one examination frequency longer than a first Response time is greater than a threshold. The first response time is advantageously greater than substantially 750ms. The threshold can depending on the power of the signal S or on the sum of the powers of all Examination frequencies are selected.

In a further advantageous embodiment of the invention is a presence of Feedback (only) then detected when the power of the microphone generated signal at least one examination frequency longer than a first Response time greater than the power of the signal generated by the microphone at every other examination frequency is. The second response time is advantageously greater than essentially 750ms.

In a further advantageous embodiment of the invention, the setting of Bandpass filter repeated at the earliest after expiration of a minimum dead time, the advantageously between 100ms to 300ms.

In a further advantageous embodiment of the invention, the performance of the means of Microphone generated signal at least 50, especially at 150 to 300, Examination frequencies determined.

In a further advantageous embodiment of the invention, the bandpass filter is a Notch filter or a filter bank with at least one notch filter. The filter bank can e.g. Include 10 notch filters.

Fig. 1

ein Kraftfahrzeug,

Fig. 2

ein Ausführungsbeispiel für eine erfindungsgemäße Einrichtung,

Fig. 3

ein Notchfilter,

Fig. 4

eine Filterbank,

Fig. 5

ein Ausführungsbeispiel für einen in einer Entscheidungslogik implementierten Ablaufpan,

Fig. 6

ein Leistung-Frequenz-Diagramm,

Fig. 7

ein Ausführungsbeispiel für Abfrage 41 in Fig. 5,

Fig. 8

ein Leistung-Frequenz-Diagramm,

Fig. 9

ein Leistung-Frequenz-Diagramm,

Fig. 10

ein weiteres Ausführungsbeispiel für Abfrage 41 in Fig. 5,

Fig. 11

ein weiteres Ausführungsbeispiel für einen in einer Entscheidungslogik implementierten Ablaufpan,

Fig. 12

ein Ausführungsbeispiel für die Abfragen 41 und 82,

Further advantages and details emerge from the following description of exemplary embodiments. Showing:

Fig. 1

a motor vehicle,

Fig. 2

an embodiment of an inventive device,

Fig. 3

a notch filter,

Fig. 4

a filter bank,

Fig. 5

an embodiment of a sequence implemented in a decision logic,

Fig. 6

a power-frequency diagram,

Fig. 7

an embodiment for query 41 in Fig. 5,

Fig. 8

a power-frequency diagram,

Fig. 9

a power-frequency diagram,

Fig. 10

another embodiment for query 41 in FIG. 5,

Fig. 11

a further embodiment of a sequence implemented in a decision logic,

Fig. 12

an embodiment for the queries 41 and 82,

Fig. 1 shows the interior view of a motor vehicle 1 from above. Designate Reference numerals 2 and 3, the front seats and reference numerals 4, 5 and 6, the rear seats of the Motor vehicle. Reference numerals 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 and 20 designate speakers. Reference numerals 21, 22, 23 and 24 denote microphones. The Speakers 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19 and 20 are included partly to a music system and partly to a communication or speech / intercom device. They can also be used by both systems.

In the present embodiment, the speakers 9, 17, 18, 19, 20 enter from the microphone 21 generated signal, the speakers 7, 17, 18, 19, 20 a of the Microphone 22 generated signal, the speakers 7, 9, 19, 20 a from the microphone 23rd generated signal and the speakers 7, 9, 17, 18 a generated by the microphone 24 Signal off. In this way, the possibility of verbal communication in one Motor vehicle supported. The communication is in principle the better ever stronger a signal between one of the microphones 21, 22, 23, 24 and one of the Loudspeaker 7, 9, 17, 18, 19, 20 is amplified. Limited is the possibility of one However, such gain due to possible feedback effects caused by by means of a loudspeaker 7, 9, 17, 18, 19, 20 radiated sound passing through a Microphone 21, 22, 23, 24 received and then amplified and through the Speaker 7, 9, 17, 18, 19, 20 is emitted.

To reduce such feedback is shown in FIG. 2 between a microphone 30, which may be one of the microphones 21, 22, 23, 24, and a speaker 31, one of the speakers 7, 9, 17, 18, 19, 20 be can, a bandpass filter 32 is provided. This filters a signal S generated by the microphone 30 and provides a filtered signal S 'in which certain frequency ranges are filtered out, for which a decision logic 33 has recognized the risk of feedback. For this purpose, the decision logic 33 determines filter parameters f _c and Q by means of which the bandpass filter 32 is set.

For amplifying the signal S and / or the signal S 'can not be shown Amplifier can be provided. However, the amplifier function can also by the Bandpass filters are taken.

3 shows the characteristic curve of a bandpass filter designed as a notch filter, wherein the gain V of the bandpass filter is plotted over the frequency f. Here f _c denotes the center frequency of the bandpass filter and Q its quality. For filtering a plurality of frequency ranges, the bandpass filter 32 is advantageously designed as a filter bank, as shown in FIG. The filter bank advantageously comprises up to 10 notch filters.

FIG. 5 shows an exemplary embodiment of a sequence chip implemented in a decision logic 33. In this case, an examination frequency is first determined in a step 40. For this purpose, the frequency f of the signal S analyzed and, as exemplified in Fig. 6, the power P of the signal S on, for example, 192 different test frequencies f _n, f _{n + 1,} f _{n + 2,} f _{n + 3,} f _{n + 4} , f _{n + 5} , f _{n + 6} , f _{n + 7} , f _{n + 8} determined, which are eg 40Hz apart. For the examination frequency f _{n + 5} , at which the power is maximum, the subsequent sequence is run through. However, it is also possible to go through the following procedure for more than one examination frequency.

It has proved to be advantageous to determine the power at the examination frequencies _fn , fn _{+ 1} , fn _{+ 2} , fn _{+ 3} , fn _{+ 4} , fn _{+ 5,} fn _{+ 6,} fn _{+ 7,} f _{n + 8} , ie to form an average value over time, and this time average of the power instead of the actual power of the signal S at the examination frequencies f _n , f _{n + 1,} f _{n + 2,} f _{n + 3 ,} f _{n + 4, n +} f _5, f _{n + 6, n +} f _7, f _{n + 8} to be examined. If the power of the signal S is mentioned in the description and the claims, this can thus also include the average value of the power formed over a certain period of time. Furthermore, the term of the power according to the invention may include the amplitude or its time average. Also included in the sense of the invention are other modifications of the power, the amplitude or their time averages, such as normalized quantities. For example, under the power of the signal S at an examination frequency f _n in the sense of the invention, the value of the power of the signal S at this examination frequency f _{n can be} divided by the sum of the power of the signal S at all examination frequencies f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , f _{n + 6} , f _{n + 7} , f _{n + 8} .

The step 40 is followed by a query 41, if there is a risk of feedback. Details of this query are made with reference to FIGS. 7 and 10. Unless the danger the feedback consists, the query 41 follows a query 42, whether that of the Microphone 30 already generates signal S by means of the bandpass filter the examination frequency has been reduced around.

If the signal S generated by the microphone 30 is not already reduced by signal components around the examination frequency by means of the bandpass filter, the query 42 is followed by a step 43 in which the filter parameters, ie the center frequency f _c and the quality Q of the bandpass filter Filters are generated. The center frequency f _c is an example of the blocking frequency in the sense of the claims. However, the blocking frequency in the sense of the claims can also be, in particular, the frequency range around the center frequency f _c , which the bandpass filter actually filters out of the signal S generated by the microphone 30.

- fkorr

die Korrekturfrequenz,

- fdist

der Abstand zwischen der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und einer die größte Leistung aufweisenden Untersuchungsfrequenz unmittelbar neben der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

- Pmax

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

- Pmaxneigh

die Leistung des mittels des Mikrofons erzeugten Signals bei der die größte Leistung aufweisenden Untersuchungsfrequenz unmittelbar neben der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und

- sign

ein Vorzeichen

ist, wobei sign positiv ist, wenn die die größte Leistung aufweisende Untersuchungsfrequenz unmittelbar neben der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, größer ist als die Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und wobei sign sonst negativ ist.The center frequency f _c can be set, for example, equal to the examination frequency. In an advantageous embodiment of the invention, however, the center frequency f _c is the examination frequency added at a correction frequency at which the power of the signal generated by the microphone is maximum, ie at the examination frequency at which the power of the signal generated by the microphone is maximum adds a correction frequency. This correction frequency is advantageously formed as a function of the power of the signal generated by the microphone at the examination frequency at which the power of the signal generated by the microphone is maximum, and the power of the signal generated by the microphone at at least one adjacent to this examination frequency examination frequency. Thus, the correction frequency, for example, according to fkorr = sign * fdist * Pmaxneigh / (Pmax + Pmaxneigh) be formed, where

- fkorr

the correction frequency,

- fdist

the distance between the examination frequency, at which the power of the signal generated by the microphone is at its maximum, and one of the highest power, the examination frequency immediately adjacent to the examination frequency at which the power of the signal generated by the microphone is maximal,

- Pmax

the power of the signal generated by the microphone at the examination frequency at which the power of the signal generated by the microphone is at a maximum,

- Pmaxneigh

the power of the signal generated by the microphone at the highest power examination frequency immediately adjacent to the examination frequency at which the power of the signal generated by the microphone is maximum, and

- sign

a sign

where sign is positive when the highest power examination frequency immediately adjacent to the examination frequency at which the power of the microphone generated signal is maximum is greater than the examination frequency at which the power of the microphone generated signal is maximum , and where else sign is negative.

- fkorr

die Korrekturfrequenz,

- Δf

der Abstand zwischen zwei Untersuchungsfrequenzen,

- Pmax

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

- Pneighright

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz unmittelbar oberhalb der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist, und

- Pneighleft

die Leistung des mittels des Mikrofons erzeugten Signals bei der Untersuchungsfrequenz unmittelbar unterhalb der Untersuchungsfrequenz, bei der die Leistung des mittels des Mikrofons erzeugten Signals maximal ist,

ist.In the present embodiment, the correction frequency according to fkorr = Δf * (Pneighright Pneighleft) / (Pmax + | Pneighright Pneighleft |) formed, where

- fkorr

the correction frequency,

- Δf

the distance between two examination frequencies,

- Pmax

the power of the signal generated by the microphone at the examination frequency at which the power of the signal generated by the microphone is at a maximum,

- Pneighright

the power of the signal generated by the microphone at the examination frequency immediately above the examination frequency at which the power of the signal generated by the microphone is at a maximum, and

- Pneighleft

the power of the signal generated by the microphone at the examination frequency immediately below the examination frequency at which the power of the signal generated by the microphone is at a maximum,

is.

The quality Q is set to a predetermined value of e.g. 1 / 40Hz set.

Step 43 is followed by inquiry 45 as to whether the program should be terminated. Should that If the program does not terminate, query 45 is followed by step 40. Otherwise the program is ended.

If the signal S generated by the microphone 30 is already reduced by signal components around the examination frequency by means of the bandpass filter, the query 43 is followed by a step 44, in which the quality Q is reduced. Thereby, the bandpass filter is adjusted so that it blocks the proportion of the signal generated by the microphone at an increased frequency range around the center frequency f _c around. Step 44 is followed by step 40.

If there is no risk of feedback, query 41 is followed by query 45 or optionally, a step 46 in which the filtering of the microphone 30 generated by the microphone Signal S is terminated around the examination frequency.

In a particularly advantageous embodiment, it is provided that the query 41 is repeated at the earliest after expiration of a minimum dead time, the minimum dead time in the present embodiment is 200ms to 300ms.

Fig. 7 shows an embodiment for the query 41. Here, a query 50 is first provided, whether the power of the signal generated by the microphone 30 S at the examination frequency by not less than a lower limit .DELTA.l greater than the power of the microphone 30 generated signal S at the first harmonic (ie twice) of the examination frequency. The lower limit Δ1 is for example between 5 and 20 dB. Advantageously, the lower limit Δ1 is substantially 12dB. This query is illustrated by way of example in FIG. 8, where f _{H0 denotes} the examination frequency, f _H1 , f _H2 , f _H3 and f _H4 the first, second, third and fourth harmonics of the examination frequency and f _{H1 / 2} the first subharmonic of the examination frequency. P denotes the power at a frequency f. With query 50 is thus queried whether P (f H0 ) -P (f H1 ) ≥ Δ1 Optionally, query 50 may be provided for one or more of the queries P (f H0 ) -P (f H1 / 2 ) ≥ Δ1 P (f H0 ) -P (f H2 ) ≥ Δ1 P (f H0 ) -P (f H3 ) ≥ Δ1 P (f H0 ) -P (f H4 ) ≥ Δ1 to supplement, where appropriate, other limits can be selected.

The examination frequencies _fn , fn _{+ 1} , fn _{+ 2} , fn _{+ 3} , fn _{+ 4} , fn _{+ 5} , fn _{+ 6} , fn _{+ 7} , fn _{+ 8} in Fig. 6 are of to distinguish the subharmonic / harmonics f _{H1 / 2} , f _H1 , f _H2 , f _H3 and f _H4 in Fig. 8 and Fig. 9, respectively. For example, assume 192 examination frequencies f ₁ , f ₂ , .... f ₁₉₂ , which are 40Hz apart, where f ₁ is 40Hz, and f ₄₄ = f _H0 , ie the examination frequency at which the power of the microphone 30th is maximum, so f _H1 = f ₈₈ and f _H2 = f ₁₂₂ .

If the power of the signal S generated by means of the microphone 30 at the examination frequency is not less than a lower limit Δ1 greater than the power of the signal S generated by the microphone 30 at the first harmonic of the examination frequency, the query 50 is followed by a query 51. The query 51 queries whether the power of the signal S generated by means of the microphone 30 at the examination frequency is not greater than an upper limit Δ2 greater than the power of the signal S generated by the microphone 30 at the first harmonic of the examination frequency. The upper limit Δ2 is for example between 20 and 40 dB. Advantageously, the upper limit Δ2 is substantially 30 dB. This query is illustrated by way of example in FIG. 9, where again f _{H0 denotes} the examination frequency, f _H1 , f _H2 , f _H3 and f _H4 the first, second, third and fourth harmonics of the examination frequency and f _{H1 / 2} the first subharmonic of the examination frequency. P again denotes the power at a frequency f. With query 51 is thus queried whether P (f H0 ) -P (f H1 ) ≥ Δ2 Optionally, query 51 may be provided for one or more of the queries P (f H0 ) -P (f H1 / 2 ) ≥ Δ2 P (f H0 ) -P (f H2 ) ≥ Δ2 P (f H0 ) -P (f H3 ) ≥ Δ2 P (f H0 ) -P (f H4 ) ≥ Δ2 to supplement, where appropriate, other limits can be selected.

Is the power of the signal generated by the microphone 30 S at the Examination frequency does not exceed by more than an upper limit Δ2 Power of the signal S generated by the microphone 30 at the first Harmonics of the examination frequency, the query 51 follows a query 52, by means of the signal generated by comparing the power of the microphone 30 S at the examination frequency with the power of the microphone 30th generated signal S at the examination frequency to at least one earlier Time is inquired whether the power of the signal generated by the microphone at the examination frequency increases exponentially.

Fig. 10 shows a further embodiment for the query 41. It is first a query 60 is provided as to whether the power generated by the microphone 30 Signal S at the examination frequency is greater than a predetermined limit value. In This case is followed by a query 61 corresponding to query 50. The queries 62 and 63 correspond to the queries 51 and 52.

Fig. 11 shows a preferred embodiment for one in the decision logic 33 implemented expiration. The process starts with a step 81, which is the step 40 in Fig. 5 corresponds. Step 81 is followed by a query 41 in FIG. 5 Query 82, if there is a risk of feedback. Embodiments for the Query 82 is shown in FIG. 7 and FIG. 10. In connection with the exemplary embodiment Figure 11 has an implementation of a feedback detection (query 82), as is explained in more detail in Fig. 12, found to be advantageous.

Unless the risk of feedback exists or is determined, follows the Query 82 is query 83 corresponding to query 45 to see if the program terminates shall be. If the program is not terminated, then query 93 follows Step 81. Otherwise, the program will exit.

If there is a risk of feedback, query 82 is followed by query 42 corresponding query 83, whether the generated by the microphone 30 signal S already by means of the bandpass filter around signal portions around the examination frequency around is reduced. If the signal S generated by the microphone 30 is already using the Bandpass filter reduced by signal components around the examination frequency, so the query 83 follows a query 85 otherwise a query 84.

Query 84 queries whether a notch filter is available. If a notch filter is available, query 84 is followed by a step 88 corresponding to step 43, in which the filter parameters, ie the center frequency f _c and the quality Q of the bandpass filter for the specific embodiment, are generated. If query 84 indicates that no notch filter is available, query 84 is followed by a step 86 in which the power of signal S is reduced by a reduction factor which is advantageously between 2 dB and 5 dB, in particular at substantially 3dB , Step 86 is followed by a step 87 in which the entire run is stopped for a stop time of substantially 3 seconds. However, this step should only be executed once per run.

By means of the query 85 is queried whether by a further expansion of the Frequency range in which the bandpass filter blocks, so by further reduction of its quality Q, a predetermined minimum quality would be undercut. Would by a further widening of the frequency range a predetermined Below minimum quality, the query 85 follows a step 89, otherwise a step 91. In step 91 corresponding to step 44, the Q is reduced.

Steps 87, 88 and 91 are followed by a step 92, in which the process has a minimum dead time long, wherein the minimum dead time in the present embodiment 100ms.

In step 89, the power of the signal S is reduced by a reduction factor advantageously between 2dB and 5dB, especially at substantially 3dB, reduced. The step 89 is followed by a step 90 in which the entire run for a Stopping time of substantially 3s is stopped.

FIG. 7 shows an exemplary embodiment for the query 82, according to which also query 41 can be implemented. Initially, a query 95 is provided, whether the Power of the signal S generated by the microphone 30 at the examination frequency greater than 750ms greater than the power of the microphone 30 generated signal S every other examination frequency. Is the performance of by the microphone 30 generated signal S at the examination frequency longer than 750 ms is greater than the power of the signal generated by the microphone 30 S every other examination frequency, query 95 is followed by a query 96. Otherwise, query 95 follows query 93.

By means of the query 96 is queried whether the power of the microphone 30th generated signal S at the examination frequency by not less than 12dB larger is the power of the signal S generated by the microphone 30 at the first one Harmonic (that is twice) of the examination frequency. Is the power of the signal S generated by the microphone 30 at the examination frequency not less than 12dB greater than the power of the microphone 30 generated signal S at the first harmonic of the examination frequency, it follows query 96 has a query 97. Otherwise, query 96 follows query 93.

By means of the query 97 is queried whether the power of the microphone 30th generated signal S at the examination frequency longer than 750ms greater than one Threshold is. Is the power of the generated by the microphone 30 signal S at the examination frequency longer than 750ms greater than a threshold query 97 follows query 83. Otherwise, query 95 follows query 93.

The feedback detection according to the invention is not based on the embodiments limited according to Fig. 7, Fig. 10 and Fig. 12. It can e.g. be provided that the queries 52 and 63 follow the no outputs of the queries 50 and 61, respectively. moreover can be provided, the embodiments of FIG. 7, Fig. 10 and Fig. 12 with their binary decision logic by a fuzzy decision logic, so fuzzy logic or replace neural networks.

LIST OF REFERENCE NUMBERS

1

motor vehicle

2, 3

front seats

4, 5, 6

rear seats

7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 31

speaker

21,22,23,24,30

Microphones

32

Bandpass filter

33

decision logic

40, 41, 43, 44, 46, 81, 84, 86, 87, 88, 89, 90 91.92

steps

41, 42, 45, 50, 51, 52, 60, 61, 62, 63, 82, 83, 84, 85, 93, 95, 96, 97

Interrogate

f

frequency

f _H0

test frequency

f _H1

first harmonic of the examination frequency

f _H2

second harmonic of the examination frequency

f _H3

third harmonic of the examination frequency

f _H4

fourth harmonic of the examination frequency

f _{H1 / 2}

first subharmonic of the examination frequency

^f n, f _{n + 1,} f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , f _{n + 6} , f _{n + 7} , f _{n + 8} , f ₁ ,. f ₂ ,. f _44,. f ₈₈ , - f _94,. f ₉₅ , f _97,. f _98,. f _122,. f ₁₉₂

frequency points

f _c

center frequency

FDIST

Distance between the examination frequency at which the Power of the signal generated by the microphone maximum, and one of the greatest performance Examination frequency immediately adjacent to the examination frequency, in which the performance of the means of Microphone generated signal is maximum

Fcorr

correction frequency

Q

quality

P

power

Pmax

Performance of the signal generated by the microphone at the examination frequency at which the performance of the of the microphone signal is maximum

Pmaxneigh

Performance of the signal generated by the microphone at the highest performance examination frequency immediately adjacent to the examination frequency, at which produced the power of the microphone Signal is maximum

Pneighleft

Performance of the signal generated by the microphone at the examination frequency immediately below the examination frequency, in which the performance of the means of Microphone generated signal is maximum

Pneighright

Performance of the signal generated by the microphone at the examination frequency immediately above the Examination frequency at which the performance of the of the microphone signal is maximum

S

signal

S '

filtered signal

sign

sign

V

reinforcement

Δ1

lower limit

Δ2

upper limit

.delta.f

Distance between two examination frequencies

Claims (54)

Method for operating a voice-assisted system, such as a communication and / or voice / intercom device in a motor vehicle (1), comprising at least one microphone (30) and at least one loudspeaker (31) for reproducing a signal generated by means of the microphone (30) (S) and a between the microphone (30) and the speaker (31) arranged bandpass filter (32), characterized in that the bandpass filter (32) in response to a comparison of the power of the microphone (30) generated Signal (S) at an examination frequency (f _{n + 5} ) with the power of the signal (S) generated by the microphone (30) at at least a substantially integer multiple of the examination frequency (f _{n + 5} ) or in dependence of a comparison of the power the signal (S) generated by means of the microphone (30) at an examination frequency (f _{n + 5} ) with the power of the signal (S) generated by means of the microphone (30) at the examination frequency ( f _{n + 5} ) is set at least one earlier. A method according to claim 1, characterized in that the bandpass filter (32) in response to a comparison of the power of the signal generated by the microphone (30) signal (S) at the examination frequency (f _{n + 5} ) with the power of the microphone (30) generated signal (S) at least a substantially integral multiple of the examination frequency (f _{n + 5} ) and in dependence of a comparison of the power of the microphone (30) generated signal (S) at the examination frequency (f _{n + 5} ) is adjusted with the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) at least at an earlier time. A method according to claim 1 or 2, characterized in that the band-pass filter (32) is set to block the proportion of the signal (S) produced by the microphone (30) at a blocking frequency when the power of the microphone (30) generated signal (S) at the examination frequency (f _{n + 5} ) by more than an upper limit (Δ2) is greater than the power of the signal (S) generated by the microphone (30) at the first harmonic of the examination frequency ( fn _{+ 5} ). Method according to Claim 3, characterized in that the upper limit (Δ2) is between 20 and 40 dB. A method according to claim 4, characterized in that the upper limit (Δ2) is substantially 30dB. Method according to one of the preceding claims, characterized in that the band-pass filter (32) is set such that it does not block the proportion of the signal (S) generated by the microphone (30) with the cut-off frequency when the power of the means of the Microphones (30) generated signal (S) at the examination frequency (f _{n + 5} ) by less than a lower limit (Δ1) is greater than the power of the signal (S) generated by the microphone (30) at the first harmonic of the examination frequency (fn _{+ 5} ). A method according to claim 6, characterized in that the lower limit (Δ1) is between 5 and 20dB. A method according to claim 7, characterized in that the lower limit (Δ1) is substantially 12dB. Method according to one of the preceding claims, characterized in that by means of a comparison of the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) with the power of the signal generated by the microphone (30) ( S) at the examination frequency (f _{n + 5} ) is decided at least earlier times, whether the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) increases exponentially. A method according to claim 9, characterized in that the band-pass filter (32) is adjusted so that it blocks the proportion of the signal (S) generated by the microphone (30) at the rejection frequency, when it is decided that the power of the of the microphone (30) signal (S) at the examination frequency (f _{n + 5} ) increases exponentially. Method according to one of the preceding claims, characterized in that the band-pass filter (32) is set such that it only blocks the proportion of the signal (S) generated by the microphone (30) with the blocking frequency if the power of the signal by means of Microphone (30) generated signal (S) at the examination frequency (f _{n + 5} ) is longer than a first response time greater than a threshold. A method according to claim 11, characterized in that the first response time is greater than substantially 750ms. Method according to one of the preceding claims, characterized in that the power at more than one examination frequency (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , f _{n + 6,} f _{n + 7} , f _{n + 8} ), and that the bandpass filter (32) is set to block the portion of the signal (S) generated by the microphone (30) at the rejection frequency only when the power of the signal (S) generated by means of the microphone (30) at an examination frequency (f _{n + 5} ) longer than a second response time is greater than the power of the signal (S) generated by the microphone (30) at any other examination frequency ( f _n, f _{n + 1,} f _{n + 2,} f _{n + 3, n +} f _4, f _{n + 6, n +} f _7, f _{n + 8).} A method according to claim 13, characterized in that the second response time is greater than substantially 750ms. Method according to one of the preceding claims, characterized in that the setting of the bandpass filter (32) with respect to the examination frequency (f _{n + 5} ) is repeated at the earliest after the expiration of a minimum dead time. A method according to claim 15, characterized in that the minimum dead time is 100ms to 300ms. A method according to claim 15 or 16, characterized in that the bandpass filter (32) is adjusted so that it blocks the proportion of the signal (S) generated by the microphone (30) at a frequency range around the cut-off frequency, if after expiration of a Repetition time greater than the minimum dead time, the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) is greater than the upper limit (Δ 2) greater than the power of the means the microphone (30) generated signal (S) at the first harmonic of the examination frequency (f _{n + 5} ) and / or if it is decided that the power of the signal generated by the microphone (30) (S) at the examination frequency (f _{n +5} ) increases exponentially. A method according to claim 15, 16 or 17, characterized in that the bandpass filter (32) is set to block the portion of the signal (S) produced by the microphone (30) at an increased frequency range around the notch frequency after a repetition time greater than the minimum dead time, the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) is greater than the upper limit (Δ 2) than the upper limit Power of the signal (S) generated by means of the microphone (30) at the first harmonic of the examination frequency (f _{n + 5} ) and / or if the power of the signal (S) generated by the microphone (30) is decided at the examination frequency (f _{n + 5} ) increases exponentially. A method according to claim 18, characterized in that the frequency range is increased by the blocking frequency only up to a minimum quality. A method according to claim 19, characterized in that the signal (S) generated by means of the microphone (30) is interrupted for an interruption period when the frequency range is increased by the blocking frequency up to the minimum quality. A method according to claim 20, characterized in that the interruption time is greater than substantially 1s to 5s. A method according to claim 21, characterized in that the interruption duration is greater than substantially 3s. Method according to one of Claims 3 to 22, characterized in that the blocking frequency is the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by means of the microphone (30) is maximal. Method according to one of Claims 3 to 22, characterized in that the blocking frequency is the examination frequency (f _{n + 5} ) added at a correction frequency at which the power of the signal (S) generated by means of the microphone (30) is maximal. Method according to Claim 24, characterized in that the correction frequency depends on the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the signal generated by means of the microphone (30) (S) is maximum, and the power of the signal (S) generated by means of the microphone (30) is formed at at least one examination frequency (f _{n + 4} ) lying next to this examination frequency (f _{n + 5} ). A method according to claim 25, characterized in that the correction frequency according to fkorr = sign * fdist * Pmaxneigh / (Pmax + Pmaxneigh) is formed, where

- fkorr

the correction frequency,

- fdist

the distance between the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum and one of the highest power examination frequencies (f _{n + 4} ) immediately adjacent to the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum,

- Pmax

the power of the signal (S) generated by means of the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by means of the microphone (30) is maximal,

- Pmaxneigh

the power of the signal (S) generated by means of the microphone (30) at the highest power examination frequency (f _{n + 4} ) immediately adjacent to the examination frequency (f _{n + 5} ) at which the power of the microphone (30) generated Signal (S) is maximum, and

- sign

a sign

where sign is positive when the highest power examination frequency (f _{n + 4} ) is immediately adjacent to the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximal, is greater than the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum, and where sign is otherwise negative. A method according to claim 25, characterized in that the correction frequency according to fkorr = Δf * (Pneighright Pneighleft) / (Pmax + | Pneighright Pneighleft |) is formed, where

- fkorr

the correction frequency,

- Δf

the distance between two test frequencies (f _n, f _{n + 1,} f _{n + 2,} f _{n + 3,} f _{n + 4,} f _{n + 5,} f _{n + 6,} f _{n + 7,} f _{n + 8),}

- Pmax

the power of the signal (S) generated by means of the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by means of the microphone (30) is maximal,

- Pneighright

the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 6} ) immediately above the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum, and

- Pneighleft

the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 4} ) immediately below the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum,

is. Device for the operation of speech-assisted systems according to a method according to one of the preceding claims, wherein the device comprises at least one microphone (30) and at least one loudspeaker (31) for reproducing a signal (S) generated by means of the microphone (30) as well as between the microphone (30) and the loudspeaker arranged bandpass filter (32), characterized in that the device comprises a decision logic for adjusting the band-pass filter (32) in dependence on a comparison of the power of the microphone (30) generated signal (S) at an examination frequency (f _{n + 5} ) with the power of the signal (S) generated by the microphone (30) at at least a substantially integer multiple of the examination frequency (f _{n + 5} ) or in dependence of a comparison of the power of the microphone (30) generated signal (S) at an examination frequency (f _{n + 5} ) with the power of the microphone (30) produce gten signal (S) at the examination frequency (f _{n + 5} ) at least one earlier point in time. Device according to claim 28, characterized in that the bandpass filter (32) is a filter bank with at least one notch filter. Method for operating a voice-assisted system, such as a communication and / or voice / intercom device in a motor vehicle (1), comprising at least one microphone (30) and at least one loudspeaker (31) for reproducing a signal generated by means of the microphone (30) (S) and a bandpass filter (32) arranged between the microphone (30) and the loudspeaker (31), the power of the signal (S) generated by means of the microphone (30) being at least three examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , f _{n +6} , fn _{+ 7} , fn _{+ 8} ), wherein by evaluating the power of the signal (S) generated by the microphone (30) at the examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , f _{n + 6} , f _{n + 7} , f _{n + 8} ) is determined whether there is feedback, and wherein the bandpass filter (32) is adjusted to disable a portion of the signal (S) generated by the microphone (30) around a notch frequency when it is determined that the feedback exists. A method according to claim 30, characterized in that the notch frequency is the examination frequency (f _{n + 5} ) at which the power of the microphone (30) generated signal (S) is maximum. A method according to claim 30, characterized in that the notch frequency is the added with a correction frequency examination frequency (f _{n + 5} ) at which the power of the microphone (30) generated signal (S) is a maximum. Method according to Claim 32, characterized in that the correction frequency depends on the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the signal generated by means of the microphone (30) (S) is maximum, and the power of the signal (S) generated by means of the microphone (30) is formed at at least one examination frequency (f _{n + 4} ) lying next to this examination frequency (f _{n + 5} ). A method according to claim 33, characterized in that the correction frequency according to fkorr = sign * fdist * Pmaxneigh / (Pmax + Pmaxneigh) is formed, where

- fkorr

the correction frequency,

- fdist

the distance between the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum and one of the highest power examination frequencies (f _{n + 4} ) immediately adjacent to the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum,

- Pmax

the power of the signal (S) generated by means of the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by means of the microphone (30) is maximal,

- Pmaxneigh

the power of the signal (S) generated by means of the microphone (30) at the highest power examination frequency (f _{n + 4} ) immediately adjacent to the examination frequency (f _{n + 5} ) at which the power of the microphone (30) generated Signal (S) is maximum, and

- sign

a sign

where sign is positive when the highest power examination frequency (f _{n + 4} ) is immediately adjacent to the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximal, is greater than the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum, and where sign is otherwise negative. A method according to claim 33, characterized in that the correction frequency according to fkorr = Δf * (Pneighright Pneighleft) / (Pmax + | Pneighright Pneighleft |) is formed, where

- fkorr

the correction frequency,

- Δf

the distance between two test frequencies (f _n, f _{n + 1,} f _{n + 2,} f _{n + 3,} f _{n + 4,} f _{n + 5,} f _{n + 6,} f _{n + 7,} f _{n + 8),}

- Pmax

the power of the signal (S) generated by means of the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by means of the microphone (30) is maximal,

- Pneighright

the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 6} ) immediately above the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum, and

- Pneighleft

the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 4} ) immediately below the examination frequency (f _{n + 5} ) at which the power of the signal (S) generated by the microphone (30) is maximum,

is. Method according to one of Claims 30 to 35, characterized in that the distances between at least a part of the examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , fn _{+ 6} , fn _{+ 7} , fn _{+ 8} ) are equidistant. Method according to one of claims 30 to 36, characterized in that the distances. between the test frequencies _(n f, f _{n + 1,} f _{n + 2,} f _{n + 3,} f _{n + 4,} f _{n + 5,} f _{n + 6,} f _{n + 7,} f _{n + 8)} are equidistant. Method according to one of Claims 30 to 37, characterized in that feedback is detected only when the power of the signal (S) generated by means of the microphone (30) at the examination frequency (f _{n + 5} ) at which the Power of the signal (S) generated by means of the microphone (30) is at most greater than the power of the signal (S) generated by the microphone (30) at the first harmonic of this examination frequency (f _{n + 5} ). A method according to claim 38, characterized in that the upper limit (Δ2) is between 20 and 40 dB. A method according to claim 39, characterized in that the upper limit (Δ2) is substantially 30dB. Method according to one of claims 30 to 40, characterized in that the absence of feedback is determined when the power of the signal (S) generated by the microphone (30) at the examination frequency (f _{n + 5} ) at which the power of the means of the microphone (30) is less than a lower limit (Δ1) than the power of the signal (S) generated by the microphone (30) at the first harmonic of this examination frequency (f _{n + 5)} ). A method according to claim 41, characterized in that the lower limit (Δ1) is between 5 and 20dB. A method according to claim 42, characterized in that the lower limit (Δ1) is substantially 12dB. Method according to one of claims 30 to 43, characterized in that a presence of feedback is only detected when the power of the signal (S) generated by means of the microphone (30) at the examination frequency (f _{n + 5} ) at which the Power of the signal (S) generated by the microphone (30) is maximally, at least approximately, increases exponentially. Method according to one of claims 30 to 44, characterized in that a presence of feedback is determined only if the power of the signal generated by the microphone (30) (S) at least one examination frequency (f _{n + 5} ) longer than one first response time is greater than a threshold. A method according to claim 45, characterized in that the first response time is greater than substantially 750ms. Method according to one of claims 30 to 46, characterized in that a presence of feedback is determined only if the power of the signal generated by the microphone (30) (S) at least one examination frequency (f _{n + 5} ) longer than one first response time greater than the power of the signal (S) produced by means of the microphone (30) at every other examination frequency (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 6)} , fn _{+ 7} , fn _{+ 8} ). A method according to claim 47, characterized in that the second response time is greater than substantially 750ms. Method according to one of claims 30 to 48, characterized in that the setting of the band-pass filter (32) is repeated at the earliest after expiration of a minimum dead time. A method according to claim 49, characterized in that the minimum dead time is 100ms to 300ms. Method according to one of claims 30 to 50, characterized in that the power of the signal (S) generated by means of the microphone (30) at at least 50 examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , fn _{+ 4} , fn _{+ 5} , fn _{+ 6} , fn _{+ 7} , fn _{+ 8} ). A method according to claim 51, characterized in that the power of the signal (S) generated by means of the microphone (30) at 150 to 300 examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , fn _{+ 5} , fn _{+ 6} , fn _{+ 7} , fn _{+ 8} ). Means for operating voice-assisted systems according to a method according to one of claims 30 to 50, wherein the device comprises at least one microphone (30) and at least one loudspeaker (31) for reproducing a signal (S) generated by means of the microphone (30) and an intermediate comprising the microphone (30) and the loudspeaker arranged bandpass filter (32), characterized in that the device is a decision logic for for determining the power of the signal (S) generated by means of the microphone (30) at at least three examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n + 5} , f _{n + 6} , fn _{+ 7} , fn _{+ 8} ), for determining a possible feedback by evaluating the power of the signal (S) generated by the microphone (30) at the examination frequencies (f _n , f _{n + 1} , f _{n + 2} , f _{n + 3} , f _{n + 4} , f _{n +5} , fn _{+ 6} , fn _{+ 7} , fn _{+ 8} ), and for setting the bandpass filter (32) so as to disable a portion of the signal (S) generated by the microphone (30) around a notch frequency when it is determined that the feedback exists; having. Device according to claim 53, characterized in that the bandpass filter (32) is a filter bank with at least one notch filter.

Images