EP1143416B1 - Time domain noise reduction - Google Patents

Abstract

A process for noise reduction during the transmission of acoustic useful signals includes the following steps of: (a) determining when a speech pause is present; (b) branching the incoming TC signal from the main signal path and utilizing a Fourier transformation to generate a frequency spectrum; (c) storing in a buffer memory (3) the last frequency spectrum recorded during the last speech pause; (d) using an inverse Fourier transformation on the respective last recorded frequency spectrum to generate a simulated noise signal; (e) subtracting the simulated noise signal in the time domain from the current incoming TC signal. As a result, the original signal is maintained uncorrupted up to the actual noise subtraction. With a simple arrangement and less computing effort than before, the process enables an overall acoustic impression to be produced, which is as agreeable as possible to the human ear and which can be matched to individual requirements. Simple optimization to the spectral processing requirements of noise signals can be realized independently of the voice signal processing requirements.

Description

The invention relates to a method for the reduction of noise signals in Telecommunication (= TK) systems for the transmission of acoustic Useful signals, in particular human speech.

One known method for noise reduction is the so-called "spectral subtraction", which is described, for example, in the publication "A new approach to noise reduction based on auditory masking effects" by S. Gustafsson and P. Jax, ITG-Fachforum, Dresden, 1998 , This is a spectral noise reduction technique which takes into account an acoustic masking threshold (for example according to the MPEG standard). Another known method for noise reduction is described in US 6 038 532.
During natural communication between people, one usually adjusts the amplitude of the spoken language automatically to the acoustic environment. However, in the case of voice communication between remote locations, the call partners are not in the same acoustic environment and therefore are not aware of the acoustic situation at the other party's location. Therefore, a problem arises more intensively when one of the partners is forced to speak very loudly due to its acoustic environment, while the other partner generates low amplitude voice signals in a quiet acoustic environment.

Particularly exacerbated noise problems arise in newer applications of communication systems, such as mobile phones the terminals are designed so small that an immediate spatial proximity between speaker and microphone unavoidable is. Due to the direct sound transmission, in particular by structure-borne noise between speaker and microphone, the acoustic interference signal come in the same order of magnitude as the useful signal of the speaker on respective terminal or this even exceed in amplitude. Such a thing Noise problem also occurs with several spatially adjacent Terminals, for example in an office or conference room with many Telephone connections in a significant extent, since a coupling of every loudspeaker signal is sent to each microphone.

In addition there is the problem that on a TK channel also an "electronic generated "noise arises and mitübertragen as background to the useful signal becomes. To increase the comfort of the phone, that's why you strives to minimize any kind of noise in relation to the useful signal to keep.

Finally, one also strives to noise such as unwanted background noise (Street noise, factory noise, office noise, canteen noise, aircraft noise, etc.) to reduce or completely suppress.

In the known compander method, as described for example in DE 42 29 912 A1 is described, the degree of noise reduction according to a fixed predetermined transfer function. The compander has first the property of voice signals with a specific (pre-set) "normal speech signal level" (possibly called normal volume) practical unchanged from its input to the output. But now once the input signal is too loud, e.g. because a speaker is too close Microphone comes, so limits a dynamic compressor to the output level almost the same value as normally, adding the current gain in the Kompander is linearly lowered with increasing input volume. By this property remains the language at the output of the compander system about the same - no matter how strong the input volume fluctuates. Now, on the other hand, a signal with a level that is smaller than the Normal level is given to the input of the compander, so will that Signal is additionally attenuated by the gain being back-regulated to If possible, transmit background noise only attenuated. Of the Kompander thus consists of two sub-functions, a compressor for Speech signal levels greater than or equal to a normal level and one Expander for signal levels lower than the normal level.

In the above-mentioned spectral subtraction is for this purpose First, the noise measured in the speech pauses and in the form of a Power density spectrum continuously stored in a memory. The power density spectrum is won over a Fourier transformation. At the Occurrence of speech, the stored sound spectrum "as the best current estimate "subtracted from the current disturbed speech spectrum, then transformed back into the time domain to create in this way a To obtain noise reduction for the disturbed signal.

A disadvantage of such methods is the complex determination of this acoustic Masking threshold and the execution of all with this procedure connected arithmetic operations. Another disadvantage of the spectral subtraction is that by the process of a basically inaccurate spectral noise estimation and subsequent subtraction also errors in the Output signal occur, which are noticeable as "musical tones".

In the extended spectral signal processing, which is also described in the citation mentioned above, first the power density spectra for the noise and for the speech itself are estimated by means of a spectral subtraction. From the knowledge of these partial spectra, a spectral acoustic masking threshold R _T (f) for the human ear is then calculated using the rules from the MPEG standard, for example. With the help of this masking threshold and the estimated noise and speech spectra, a filter pass curve H (f) is calculated according to a simple rule, designed to let the essential spectral parts of the speech pass as unaltered as possible and reduce the spectral parts of the noise as much as possible.

Then the original disturbed speech signal is given only by this filter, in this way, a noise reduction for the disturbed signal receive. The advantage of this method is that of the disturbed Signal "Nothing is added or subtracted" and therefore error in the estimates are less or hardly perceptible. The disadvantage is again the significant greater computational effort.

Of particular disadvantage with all these known methods is the fact that the incoming original signal already before the actual subtraction a simulated sound signal anyway a signal processing process subject to and thus fundamentally falsified.

Object of the present invention is in contrast, a method possible low complexity with the features described above, in a technically inexpensive way a noise reduction or noise suppression is achieved, and at the original signal remains untouched until the actual noise deduction. there The procedure should be simple, especially with less computational effort as far as possible, one for the human ear possible pleasant overall acoustic impression, depending on the taste can be adapted to individual needs. Finally, should the new method completely independent of the requirements for a voice signal processing can be performed and thus a simple optimization to the requirements of spectral processing of noise signals enable.

(a) Feststellen mittels Sprach-Pausen-Detektion, wann in der zu übertragenden Mischung aus Nutzsignalen und Störsignalen ein Sprachsignal enthalten ist oder wann eine Sprachpause vorliegt;

(b) Abzweigen des ankommenden TK-Signals von Hauptsignalpfad und Anwenden einer Fourier-Transformation auf das abgezweigte TK-Signal zur Erzeugung eines Frequenzspektrums des abgezweigten TK-Signals;

(c) Speichern des letzten während der letzten Sprachpause aufgenommenen Frequenzspektrums in einem Zwischenspeicher;

(d) Anwenden einer inversen Fourier-Transformation auf das jeweils letzte aufgenommenen Frequenzspektrum zur Erzeugung eines nachgebildeten Geräuschsignals;

(e) Abziehen des nachgebildeten Geräuschsignals im Zeitbereich vom aktuell ankommenden TK-Signal.

According to the invention this object is achieved in a simple yet effective manner by the following method steps:

(A) detecting by means of speech pause detection, when in the mixture of useful signals and interference signals to be transmitted a speech signal is included or when there is a speech break;

(b) branching the incoming TK signal from the main signal path and applying a Fourier transform to the branched TK signal to produce a frequency spectrum of the branched TK signal;

(c) storing the last frequency spectrum recorded during the last speech break in a buffer;

(d) applying an inverse Fourier transform to the latest recorded frequency spectrum to produce a replicated noise signal;

(e) subtracting the simulated noise signal in the time domain from the currently arriving TK signal.

The inventive method made possible by the separate replica the noise signal in the frequency domain regardless of processing the original voice signal direct deduction of the replicated Noise signal from the original, unadulterated input signal, which neither a Fourier transform nor an inverse Fourier transform is subjected. With a corresponding phase correction in the frequency domain is even a noise subtraction from the original signal with no time delay possible. The inventive method is less complex as the above-described known prior art methods, requires less computing power and leads to better frequency resolution.

By separating the noise simulation from the transmission of the original signal allows the process of the invention in a particularly preferred Variant that in step (d) only a selected part of the generated Frequency spectrum used to generate the simulated noise signal becomes. Thus, the for carrying out the method according to the invention required computing power further minimized or the process itself be done even faster.

A development of this variant of the method is characterized in that the selection of the for generating the simulated noise signal used part of the frequency spectrum according to criteria of psychoacoustics according to the mean values of the perceptual spectrum of the human Hearing takes place.

The value for the sound signal to be reproduced is not only from the instantaneous power value of an original signal in speech pauses alone, but also from a weighted spectral course of the corresponding signal determined and in total over the function thus gained a hearing-correct, i.e. Achieved a psychoacoustically pleasing-sounding noise reduction.

Because it is not an easily reproducible measure of an acoustically pleasant sounding Noise reduction is there, all quality assessments are extensive Hearing tests dependent, which then optimized by means of statistical Methods are evaluated to a rating scale, (similar to Speech codecs).

The basic procedures for this are, for example, the textbook by E. Zwicker, "Psychoacoustics", Springer-Verlag Berlin, 1982, in particular Pages 51-53.

Through the psychoacoustic evaluation can not only the perceptible quality the overall signal are optimized, but there are also still Further savings in the required computing power possible, for example Masking effects are exploited or only those frequencies Consideration that is clearly caused by sources of noise or interference were.

In an alternative development of the above method variant, the Selection of the signal used to generate the simulated noise signal Part of the frequency spectrum such that only discrete frequencies of the Spectrum are considered, and that the distance of the discrete frequencies steadily larger in the direction of higher frequencies, preferably after one logarithmic function is selected. This is the frequency resolution to the Perception of the human ear better adapted.

Can be further improved these developments in that the selected part of the frequency spectrum in predetermined frequency groups divided and that in each frequency group only the frequency or the frequency band with the largest signal energy within the frequency group selected and used to generate the simulated noise signal becomes. With this selection, a large reduction of the to be calculated Frequencies at a constant audible or perceptible Quality achieved, which further reduces the processing power for the process and the quality of the output signal is further increased.

It is particularly advantageous if the selection of the frequency or the frequency band with the largest signal energy within the frequency group Step (c) or before step (d) takes place. By selecting a specific frequency From a frequency group, differences in the signal energy are particularly easily detectable.

Also advantageous is a method variant in which the frequency spectrum in step (b) the branched TK signal only in a predetermined frequency range is produced. If the source of interference only a limited frequency spectrum can, in turn, with this measure, considerable computing power be saved. For example, in motor vehicles with sources of interference in a frequency range only up to a maximum of 1 KHz to be expected, since the Interference signal mainly due to low-frequency sound generation (engine, Gear, rolling noise, etc.) is formed.

Particularly simple is a method variant, which is characterized in that in step (b) and / or in step (d) a discrete Fourier transform or an inverse discrete Fourier transform is applied, wherein the incoming TK signal with discrete time amplitude values a sampling frequency f _{T are} sampled.

In a preferred embodiment of the method variant, in step (b) a fast Fourier transform (= FFT) is applied. If a large frequency area with simultaneous high frequency resolution, With this approach you can analyze with the lowest computing power To run. The FFT is particularly useful if, for example must be expected with more than 128 frequency lines.

Advantageously, in step (d) an inverse discrete Fourier transform (= IDFT). This allows a signal synthesis with Perform the least amount of processing if a selected spectrum is processed, since the disadvantage of an equidistant frequency distribution in the FFT is avoided. The IDFT can therefore be advantageous for a defined frequency range be applied. The distribution of frequencies can be individual respectively. From a frequency resolution of less than 128 frequency lines is a saving of the computing power compared to the FFT possible.

In the application are also savings in computing power or quality improvements achievable if in step (d) an inverse fast Fourier transform (= IFFT) is applied. In combination with an FFT In step (b) broadband interferers can be processed particularly economically become.

As an alternative to the latter method variant, an embodiment is selected in which only that part of the generated frequency spectrum which is below half the sampling frequency f _T / 2. This in turn can be achieved in savings in computing power, but also in storage space.

Also particularly advantageous is a variant of the method according to the invention, wherein in step (c) a frequency spectrum is buffered, the by averaging the frequency spectrum currently generated in step (b) previously generated frequency spectra is obtained. By averaging Spectral lines found with great energy and random or sporadic Error systematically suppressed.

It is particularly advantageous if the averaging with different relative Weighting of the currently generated frequency spectrum in different Frequency ranges takes place. Generally, with such different Consider the natural transient response of interferers. For example, the speed of an engine in a motor vehicle usually not change abruptly. Low frequency interferers have one higher transient time than high frequency. The proposed weighting helps to make the adaptivity of a system stable and fast.

It is again particularly advantageous if the weighting according to criteria psychoacoustics according to the means of the perception spectrum of the human hearing. As discussed above, psychoacoustic Weighting the frequency dependent settling times to the adapted to human hearing. This achieves an optimization of the system in terms of naturalness, stability and adaptation time.

To avoid overcompensation in the noise treatment is in a particularly preferred variant of the method according to the invention in step (e) according to predetermined criteria with a weighting factor a <1 weighted simulated noise signal from the currently arriving one TK signal deducted.

In an advantageous development, the weighting factor a is used as one of Faults of the TK system dependent constant value selected. this makes possible an inexpensive and simple optimization of the invention Procedure to the errors of the respective telecommunications system. The errors become automatic recorded, the weighting can also take place during operation.

Alternatively, the weighting factor a may be determined as one after one by the user of the TK-Systems selectable quality level adjustable value can be selected. One such user-defined weighting factor allows an individual, custom adaptation of the method according to the invention to the individual Needs. Is the system of the invention in an existing integrated parent concept can be a user-provided statistical value, such as the error probability or recognition rate used to control the weighting factor. For applications in the automotive sector, the weighting factor, for example also be derived from the speed or speed.

This can be further improved by the fact that the weighting factor a is adaptive is adapted to the current incoming TK signal. The adaptive weighting allows automatic optimization of noise reduction during of the operation. The weighting factor may vary from statistical values such as probability of error, Mean value, state changes, etc. are derived. With the adaptive weighting are particularly easy and quick adjustments of the inventive method to individual circumstances in the acoustic environment of the telecommunications terminal possible.

A further advantageous variant of the method according to the invention is characterized characterized in that the simulated noise signal generated in step (d) before step (e) a synthetic noise signal is added. The Admixture of an artificial noise signal with constant power density can mask dynamic, non-stationary interferers in the output signal serve.

A further variant of the method according to the invention provides that the currently arriving TK signal before step (e) of a defined time delay is subjected, which is preferably designed so that the phase angle of the incoming TK signal with the phase position of the simulated noise signal matches before withdrawal.

In an alternative variant of the method it is provided that the currently arriving TK signal is fed without delay to the trigger in step (e), and that the simulated noise signal in its phase position before step (e) to the phase angle of the currently arriving TK signal is adjusted. Will the Phase angle of the reproduced noise signal in the frequency range before Corrected inverse transformation, the subtraction from the instantaneous signal take place in the time domain. Disturbing signal delays can thus be dispensed with. These inevitably occur in all procedures in which the useful signal (Language) makes the detour via two transformations, such as in the known spectral subtraction discussed above.

Particularly preferred is a variant of the method according to the invention in addition to the detection and reduction of noise signals presence echo signals are detected and / or predicted and the echo signals be suppressed or reduced. An additional echo cancellation is however, only possible if the received original signal from the remote TK participant is included in the echo calculation. This means that the noise reproduction also includes a echo reproduction that with a connected to the remote TK subscriber signal is connected.

This process variant can be improved by the fact that the control the reduction of noise signals and the reduction of echo signals done separately.

It is also advantageous if, during the period of an echo reduction to User signal in addition an artificial noise signal is added, as it already is discussed above to give the subjective impression of a "dead Line "to avoid.

In particular, the artificial noise signal may be psychoacoustically pleasing perceived acoustic signal sequence (= comfort noise) include.

Alternatively, the artificial noise signal may be pre-selected during the current one TK connection recorded sound signal, which includes the current reproduce the acoustic environment particularly "lifelike".

The scope of the present invention also includes a server line, a Processor assembly and a gate array assembly to support the method described above and a computer program to carry out the process. The method can be used both as a hardware circuit, as well as in the form of a computer program. Nowadays, software programming for powerful DSP's preferred because new knowledge and additional functions easier by a Modification of the software can be implemented on an existing hardware basis are. However, methods can also be used as hardware components, for example in TK terminals or telephone systems are implemented.

Further advantages of the invention will become apparent from the description and the Drawing. Likewise, the above and those listed further Features according to the invention each individually or for several find use in any combination. The shown and described Embodiments are not to be exhaustive but rather have an exemplary character for the description the invention.

Fig. 1

ein stark schematisiertes Diagramm der Funktionsweise einer Einrichtung zur Durchführung des erfindungsgemäßen Verfahrens;

Fig. 2

eine detailliertere schematische Darstellung einer Einrichtung zur Durchführung des erfindungsgemäßen Verfahrens;

Fig. 3

ein Schema für ein spektrales Subtraktionsverfahren nach dem Stand der Technik;

Fig. 4

eine Ausführungsform der Erfindung mit schneller Fourier-Transformation und schneller Rücktransformation sowie blockweise überlappender Bearbeitung des eingegebenen Zeitsignals im Frequenzbereich;

Fig. 5

Ein Schema einer Ausführungsform mit gleichzeitiger Echo-reduktion;

Fig. 6a

ein Beispiel eines mit FFT berechneten Geräuschsignals im Frequenzraum;

Fig. 6b

ein mit einer diskreten Fourier-Transformation und nur bis f_s/2 berechneten Geräuschsignals; und

Fig. 6c

ein Geräuschsignal im Frequenzbereich bis f_s/2 als Ergebnis einer modifizierten Fourier-Transformation mit höherer Auflösung.

The invention is illustrated in the drawing and will be explained in more detail with reference to embodiments. Show it:

Fig. 1

a highly schematic diagram of the operation of a device for carrying out the method according to the invention;

Fig. 2

a more detailed schematic representation of a device for carrying out the method according to the invention;

Fig. 3

a scheme for a spectral subtraction method according to the prior art;

Fig. 4

an embodiment of the invention with fast Fourier transform and fast back propagation and block overlapping processing of the input time signal in the frequency domain;

Fig. 5

A schematic of an embodiment with simultaneous echo reduction;

Fig. 6a

an example of a FFT calculated noise signal in frequency space;

Fig. 6b

a noise signal calculated with a discrete Fourier transform and only up to f _s / 2; and

Fig. 6c

a noise signal in the frequency range up to f _s / 2 as a result of a modified Fourier transform with higher resolution.

In Fig. 1 it is shown how from an incoming original signal x, which contains a voice portion s and a noise n, on the one hand in a device 1, a noise signal y _n in the frequency domain is simulated and on the other hand, the original signal x _{s + n} separated from the noise simulation of a Noise subtraction is supplied, wherein optionally a time delay time delay τ can be made. The noise-reduced signal y _s is then forwarded in the TK system.

In Fig. 2, a simple embodiment is shown in which in the device 1 a for noise simulation a virtually always required speech pause detector 2 is provided, which determines when the incoming Signal may contain speech signals or when there is a speech break. In parallel, the incoming TK signal of a Fourier transform FT subjected to generating a Frequenzsprektrums and each of them resulting frequency spectrum stored in a buffer 3. The time sequentially stored frequency spectra can help with a means 4 are averaged.

Once the speech pause detector 2 determines that a speech pause is over is and in the incoming original signal and speech signals can be present, becomes the last stored in the buffer memory 3 frequency spectrum (possibly averaged with previously recorded spectra) of an inverse Fourier transformation IFT subjected and in a Subtraktionsglied 5 of Original signal, which was possibly subjected to a time delay τ deducted, to get a noise-free or at least noise-reduced signal.

In contrast, in known methods of spectral subtraction the incoming original signal, as shown in Fig. 3, directly a Fourier transform FT, a replicated noise signal in the frequency domain in a subtractor 5 'from the Fourier-transformed Original signal subtracted and the resulting new, noise-reduced Signal in the frequency domain of an inverse Fourier transform

IFT subjected and forwarded as a noise-reduced TK signal in the time domain. It thus finds in the known methods in the prior art basically always a change of the original signal even before actual sound reduction instead.

FIG. 4 shows a further embodiment of the invention, in which the original signal x _{s + n,} which is initially received in the time domain, is processed block-by-block in the device 1 b for noise simulation. In this case the time signal before the transformation into the frequency range is subjected to a windowing (eg according to Hamming) in a correspondingly upstream device 4 'or 4 "In order to compensate for the errors caused by the windowing during the inverse transformation, in addition to the processing in a first Path is made parallel processing in another path with the same fenestration, wherein only the signal is offset by half the window length and otherwise the simulated noise signal is calculated by the same means, whereby a compensation of the errors generated by the fenestration can be achieved.

Specifically, in the example shown, in the first path the windowing is performed in a device 4 ', then the time signal is subjected to a fast Fourier transformation FFT and the resulting spectrum is stored in an intermediate memory 3'. The same happens in the second path via a window device 4 "and an intermediate storage of the Fourier-transformed signal in a buffer 3". An inverse fast Fourier transformation IFFT is connected to the latches 3 ', 3 ", and the resulting spectra in the time domain are combined to form a simulated noise signal Yn in an overlap device 6. Subsequently, the simulated noise signal in the subtraction element 5 is converted by a optionally subtracted by a time τ time-offset original signal x _{s + n in} order to obtain the noise-corrected output signal y _S. The subtraction of the noise signal from the original signal in the subtraction element 5 can be phase-adjusted.

A further embodiment is shown in Fig. 5, where the branched incoming TK signal x _{s + n + e in} addition to speech and noise signals also contains echo signals. In a device 1c for noise and echo replica also an echo signal e is input, which is further treated in a processing path parallel to the noise training path.

The incoming original signal X _{s + n + e} is first subjected to a windowing in a device 4a, then a fast Fourier transform FFT and the obtained frequency spectrum are buffered in a buffer 3a. In parallel, the echo signal e in a device 4b is also subjected to a windowing and then Fourier-transformed. The frequency spectra of both paths are buffered in a buffer 3b and possibly subjected to averaging. Thereafter, a fast inverse Fourier transformation IFFT is performed separately on both paths. Finally, in a device 6a, the simulated noise signal and the simulated echo signal are overlapped into a total signal y _{n + e to be} subtracted, which is subtracted in the subtraction device 5 from the original signal x _{s + n + e} delayed or delayed by a time τ in order to record the noise and echo-reduced TK signal y _s .

The Fign. Finally, FIGS. 6a to 6c show examples of noise signals calculated in the frequency domain according to the method of the invention. In this case, in the example according to FIG. 6a, the noise signal to be reproduced has been obtained from a fast Fourier transformation FFT. The typical mirror symmetry can be seen around half the frequency value f _s / 2.

However, it is also sufficient if only the first half of the simulated noise signal in the frequency space up to the frequency f _s / 2 is used, which is illustrated in FIG. 6 _b by means of an example whose result was obtained by means of a discrete Fourier transformation.

Finally, Fig. 6c shows the result of using a modified discrete Fourier transform with higher resolution, again processing only half of the frequency spectrum up to the frequency f _s / 2.

Claims (14)

1. Process for reducing noise signals in telecommunications (TC) systems for the transmission of acoustic useful signals, in particular human speech, with the following steps:

   (a) Determining by means of speech pause detection when a speech signal is contained in the mixture of useful signals and interference signals to be transmitted, or when a speech pause is present;

   (b) Branching the incoming TC signal from the main signal path and using a Fourier transformation on the branched TC signal to generate a frequency spectrum of the branched TC signal;

   (c) Storing in a buffer memory (3) the last frequency spectrum recorded during the last speech pause;

   (d) Using an inverse Fourier transformation on the last respective recorded frequency spectrum to generate a simulated noise signal;

   (e) Subtracting the simulated noise signal in the time domain from the current incoming TC signal.
2. Process according to Claim 1, characterised in that in step (d) only one selected part of the generated frequency spectrum is utilised for the generation of the simulated noise signal.
3. Process according to Claim 2, characterised in that the selection of the part of the frequency spectrum used for the generation of the simulated noise signal is made in accordance with psycho-acoustic criteria implementing the mean values of the perception spectrum of the human ear.
4. Process according to Claim 2, characterised in that the selection of the part of the frequency spectrum used for the generation of the simulated noise signal is made in such a way that only discrete frequencies of the spectrum are considered, and that the spacing between the discrete frequencies is made to steadily increase towards the higher frequencies and preferably in accordance with a logarithmic function.
5. Process according to Claim 2, characterised in that the selected part of the frequency spectrum is divided into previously determined frequency groups, and that in each frequency group only the frequency or frequency band, respectively, having the highest signal energy within the frequency group is selected and further utilised for the generation of the simulated noise signal.
6. Process according to Claim 5, characterised in that the selection of the frequency or frequency band, respectively, having the highest signal energy within the frequency group is made prior to step (c) or step (d), respectively.
7. Process according to Claim 1, characterised in that in step (b) the frequency spectrum of the branched TC signal is generated only in a predetermined frequency band.
8. Process according to Claim 1, characterised in that a frequency spectrum that is obtained by averaging the current frequency spectrum generated in step (b) and the previously generated frequency spectra, is temporarily stored in step (c).
9. Process according to Claim 8, characterised in that the averaging with a different relative weighting of the currently generated frequency spectrum is realised in different frequency bands.
10. Process according to Claim 9, characterised in that the weighting is realised in accordance with psycho-acoustic criteria implementing the mean values of the perception spectrum of the human ear.
11. Process according to Claim 1, characterised in that a simulated noise signal weighted with a weighting factor a < 1 in accordance with predetermined criteria is subtracted from the current incoming TC signal in step (e).
12. Process according to Claim 1, characterised in that prior to step (e) a synthetic noise signal is mixed with the simulated noise signal generated in step (d).
13. Process according to Claim 1, characterised in that prior to step (e) the current incoming TC signal undergoes a specified time delay that is preferably designed so that the phase of the incoming TC signal coincides with the phase of the simulated noise signal prior to subtraction.
14. Process according to Claim 1, characterised in that the current incoming TC signal is fed for immediate subtraction in step (e) and that prior to step (e) the phase of the simulated noise signal is matched to the phase of the current incoming TC signal.

Images