Audio decorrelation remover, processing system and method for decorrelation audio signals

Abstract

The decorrelation comprises a plurality of delay units, each configured to receive a portion of a frequency representation based on the audio signal, each delay unit configured to delay the received portion to provide a delayed portion. The decorrelation comprises an envelope shaper configured to receive and combine a signal based on the delayed portions of the frequency representation. The envelope shaper is configured to receive the frequency representation of the audio signal and adjust the energy of the delayed portions with respect to the frequency representation of the audio signal. The envelope shaper is configured to provide a synthetic shaped frequency representation. The transient signal portion is processed by an adaptive operation of the decorrelation.

Description

[Technical Field]

[0001] The present invention relates to a decorrelation device for audio signals, a processing system having such a decorrelation device, a decorrelation method, and a computer program product. The present invention particularly relates to an audio signal decorrelation device. [Background technology]

[0002] In perceptual audio coding, decorrelators are a crucial component of parametric spatial audio coding. Known solutions relate to decorrelators known from parametric stereo or parametric spatial audio coding such as MPEG surround. Decorrelators such as those described in [1] or [2] use computationally expensive time-domain reverb filters with long impulse responses. Decorrelators such as those described in [3] or [4] require the use of orthogonal mirror filter banks (QMFs) with considerable processing delays and computationally expensive grid filters.

[0003] Therefore, there is a need for a decorrelator for decorrelating portions of an audio signal that enables low processing delay and / or low computational complexity decorrelation, a processing system having such a decorrelator, and a method. [Overview of the project]

[0004] The object of the present invention is to provide a decorrelator, processing system, and decorrelator method that enable decorrelation with low processing delay and / or low complexity and high perceptual quality, particularly when processing signals including transient phenomena. This objective is achieved by the subject matter defined in the independent claim.

[0005] The discovery of the present invention is that by dividing the frequency representation into a plurality of parts and processing, that is, delaying each part by separate delay units, calculations of different parts can be executed in parallel, enabling a low processing delay. At the same time, such frequency domain operations require a low computational complexity.

[0006] According to an embodiment, the correlation remover includes a plurality of delay units, each delay unit being configured to receive a part of the frequency representation based on the audio signal, and each delay unit being configured to delay the received part to provide a delayed part. The correlation remover receives a composite signal based on the delayed part of the frequency representation, receives the frequency representation of the audio signal, and is configured with an envelope shaper that adjusts the energy of the delayed part with respect to the frequency representation of the audio signal and provides a synthesized shaped frequency representation.

[0007] According to an embodiment, different parts of the frequency representation include the same or different numbers of frequency bins. The same number of frequency bins may enable the same processing time, and different numbers of frequency bins may enable adaptation to application requirements.

[0008] According to an embodiment, the correlation remover includes a phase shifter configured to phase shift the frequency representation of the audio signal or to phase shift the audio signal in the time domain to obtain a phase-shifted audio signal. The phase shift may enable a perceived reverberation and thus a high audio quality.

[0009] According to an embodiment, the phase shifter is configured to phase shift the frequency representation of the audio signal and includes a plurality of all-pass filters, each all-pass filter being configured to phase shift a relevant part of the frequency representation of the audio signal. That is, the all-pass filters can be associated and adapted to each part of the audio signal that may enable a high overall audio quality.

[0010] According to an embodiment, one of the plurality of all-pass filters comprises a set of all-pass filter structures connected in series with each other, i.e., an all-pass filter structure using a shredder IIR filter. The all-pass filter structures are adapted to provide different time delays. Alternatively or additionally, the all-pass filter structure comprises an nested all-pass filter structure.

[0011] According to an embodiment, the number of all-pass filter structures and / or the circuitry of the all-pass filter structures are equal or different between different all-pass filters. This allows for high flexibility of the correlation canceler.

[0012] According to an embodiment, the different time delays are based on prime multiples of the local sampling rate used to obtain the frequency representation of the audio signal. This allows for high perceptual audio quality.

[0013] According to an embodiment, the set of all-pass filter structures comprises some four all-pass filter structures and is adapted to provide delays of 1, 2, 3, and 5 time units. Such time units may be based on the block size of the conversion to the frequency domain. For example, using a block size of 256 with 50% overlap, the time unit may result in 128 samples at 48 kHz = 2.7 ms. Other reasonable time units may be, for example, 32 or 64 samples or other values. The time unit is preferably short enough to allow for sufficient time resolution in subsequent time / frequency envelope shaping. In an alternative solution, delays of 1, 3, 5, and 7 are provided by four all-pass filter structures. This makes it possible to avoid overlap in the time domain.

[0014] According to an embodiment, the gain factor of the all-pass filter is adapted to a magnitude of 0.7 within an allowable range, i.e., a value having a positive or negative value. The allowable range is, for example, 20%, 10%, or 5%.

[0015] According to the embodiment, the phase shifter is configured to phase-shift an audio signal in the time domain, and the phase shifter comprises a set of all-pass filter structures connected in series with each other, the all-pass filter structures being adapted to provide different time delays. Alternatively or additionally, the all-pass filter structures comprise nested all-pass filter structures.

[0016] According to the embodiment, different all-pass time delays are based on prime multiples of the reciprocal of the sampling rate used to obtain the frequency representation of the audio signal. Corresponding advantages can be obtained in the time domain as well as in the frequency domain. In the time domain, different time delays may be based on primes obtained by multiplying each of the smallest set of primes, e.g., 1, 2, 3, and 5 as one exemplary set, or 1, 3, 5, and 7 as another exemplary set, by a downsampling coefficient used to generate a portion of the frequency representation of the audio signal to obtain an intermediate result and to use the next prime for the intermediate result. As the next prime, for example, the closest distance to obtain the next larger or next smaller prime value can be understood. In a given example, the values ​​131, 257, 383, and 641 may be obtained for the first set, and 131, 383, 641, and 907 may be obtained for the second example set, where one time unit may be one sample. A sample may relate to a sampling frequency, e.g., 48 kHz. In other embodiments, the sampling frequency may be 44.1 kHz or 32 kHz or other values.

[0017] According to the embodiment, the correlation remover comprises a first conversion unit for obtaining the frequency representation of an audio signal from the audio signal of the envelope shaper, and a second conversion unit for obtaining the frequency representation from the reverberation audio signal, wherein the frequency representation portion is formed from the reverberation audio signal. This makes it possible to generate a used signal that is directly formed in the correlation remover.

[0018] According to the embodiment, the correlation remover is adapted to further implement the same predetermined delay for a subset or all of the frequency representation. That is, the delay equal to each part or delay line can also be commonly applied to a common delay module that allows for a simple delay unit for each delay line of the relevant part.

[0019] According to one embodiment, a delay unit associated with a spectral portion among a plurality of delay units is configured to delay the relevant portion of the frequency representation differently compared to delay units associated with other spectral portions. This enables high perceptual quality by treating different frequency portions differently.

[0020] According to one embodiment, the delay unit is configured to delay a portion of the frequency representation containing lower frequencies by a higher time delay compared to a portion of the frequency representation containing higher frequencies.

[0021] According to the embodiment, the relationship between different time delays is based on linear, logarithmic, and / or subband sample rounding. This enables high perceptual quality.

[0022] According to the embodiment, the decorrelator comprises a transform unit for receiving to transform an audio signal or a reverberation version of an audio signal into parts by performing a discrete Fourier transform, DFT, or short-time Fourier transform, STFT on a time-block basis, wherein the transform unit is configured to transform blocks having a 50% overlap within an acceptable range. Such block-by-block transforms allow for short delays in each resulting part and parallel processing of different parts.

[0023] According to one embodiment, the envelope shaper is configured to operate with a time resolution of less than 4 milliseconds in the subband region.

[0024] According to one embodiment, the correlation remover includes a signal processing stage configured to receive a signal based on a synthesized and shaped frequency representation, for example, as a monaural signal, and to process the monaural signal into at least a stereo signal. This enables improved perception by the listener.

[0025] According to one embodiment, the correlation remover includes a signal processing stage configured to process the synthesized and shaped frequency representation into at least a stereo signal, and, for example, in the frequency domain, to perform source expansion modeling based on at least the stereo signal.

[0026] According to one embodiment, the processing system comprises a correlation remover as described herein and a processing stage for converting a central / side-decomposed signal into a left / right-decomposed signal.

[0027] According to the embodiment, the processing system may perform transient suppression to suppress echoes, such as pre-echoes and / or post-echoes caused by transients. Such transient processing may include muting the output of a decorrelator and, correspondingly, amplifying the output of a delay compensation unit that provides a portion of the left / right separated signals, is in parallel with the decorrelator, and is connected to the processing stage.

[0028] According to the embodiment, the method includes receiving multiple parts of a frequency representation based on an audio signal, delaying each of the received parts to provide multiple delay parts, and receiving and combining signals based on the delay parts of the frequency representation. The method also includes receiving a frequency representation of an audio signal and adjusting the energy of the delay parts with respect to the frequency representation of the audio signal. A synthesized and shaped frequency representation is provided.

[0029] According to one embodiment, when executed on a computer, a computer program or computer program product or non-temporary storage medium is provided that stores instructions for executing each instruction in order to perform such a method.

[0030] Further advantageous embodiments are defined in the dependent claims. Advantageous embodiments will be described in more detail with reference to the attached drawings. [Brief explanation of the drawing]

[0031] [Figure 1] A schematic block diagram of the correlation remover according to the embodiment is shown. [Figure 2] A schematic block diagram of a correlation remover comprising a conversion unit for generating a frequency representation of a time-domain signal according to the embodiment is shown. [Figure 3] A schematic block diagram of correlation removal, including a prior delay, according to the embodiment is shown. [Figure 4] A schematic block diagram of the all-pass filter according to the embodiment is shown. [Figure 5] This shows a schematic block diagram of a nested all-pass filter structure according to the embodiment. [Figure 6] A schematic block diagram of a correlation remover comprising a phase shifter configured to operate in the time domain according to the embodiment is shown. [Figure 7] This shows a schematic block diagram of a correlation remover connected to a source extension model according to the embodiment. [Figure 8] This shows a schematic block diagram of the processing system according to the embodiment. [Figure 9] This shows a schematic block diagram of the processing system configured for handling transient phenomena according to the embodiment. [Figure 10] A schematic block diagram of the method according to the embodiment is shown. [Modes for carrying out the invention]

[0032] Equal or equivalent elements, or elements having equivalent or similar functions, are indicated in the following description by equivalent or similar reference numerals, even if they occur in different figures.

[0033] The following description includes several details to provide a more complete description of embodiments of the present invention. However, it will be apparent to those skilled in the art that embodiments of the present invention can be carried out without these specific details. In other examples, well-known structures and apparatus are shown in block diagram form rather than in detail to avoid obscuring embodiments of the present invention. Furthermore, features of the different embodiments described below can be combined with each other unless otherwise specified.

[0034] Figure 1 shows a schematic block diagram of a correlation remover 10 according to an embodiment. The correlation remover 10 consists of at least two delay units 121 to 12 n>1 n The system comprises two delay units 12, but the number is preferably more, for example, 4, 8, 16, or other values ​​obtained by powers of 2, and embodiments are not limited to such a number. That is, embodiments may also comprise 3, 5, 7, or 9 delay units 12. Each delay unit comprises the relevant portion 141 to 14 of the frequency representation 14 based on the audio signal. n It is configured to receive the following. For example, the frequency representation 14 may be or may include the spectrum obtained by a Fourier transform such as the Discrete Fourier Transform, DFT, or Short-Term Fourier Transform, STFT. Parts 141 to 14 n This can be obtained, for example, as a subband of the spectrum, i.e., a part of the frequency domain representation. For example, such parts 141 to 14 n This can be achieved by using the appropriate window.

[0035] Each delay unit 121 to 12 n The delayed portion is from 14'1 to 14' n To provide, that is, to have a delay in the time domain, receiving parts 141 to 14 n It was configured to delay it.

[0036] The correlation remover 10 controls the delay portion 14'1 to 14' nIt further includes an envelope shaper 16 configured to receive a signal based on such. Such a signal may be the delay parts 14’1 to 14’ itself or its processed variant. The envelope shaper 16 is configured to synthesize the received signal. Further, the envelope shaper is configured to receive the frequency representation 14 of the audio signal. The envelope shaper 16 is configured to adjust the energy of the delay parts 14’1 to 14’ with respect to the frequency representation 14 of the audio signal. The envelope shaper 16 is configured to provide a synthesized and shaped frequency representation 18. In the synthesized and shaped frequency representation 18, for each part 141 to 14, the resulting signals can be decorrelated with respect to each other and / or with respect to the frequency representation 14. n Although the envelope shaper 16 is shown as receiving the synthesized frequency representation 14, alternatively, the envelope shaper 16 may receive the information for each by receiving parts 141 to 14 that may not be delayed or are generally processed. n The envelope shaper 16 is configured to adjust the energy of the delay parts 14’1 to 14’ with respect to the frequency representation 14 of the audio signal. The envelope shaper 16 is configured to provide a synthesized and shaped frequency representation 18. In the synthesized and shaped frequency representation 18, for each part 141 to 14, the resulting signals can be decorrelated with respect to each other and / or with respect to the frequency representation 14. n 、それぞれその結果生じる信号は、互いに関しておよび / または周波数表現14に関して相関除去され得る。

[0037] Although the envelope shaper 16 is shown as receiving the synthesized frequency representation 14, alternatively, the envelope shaper 16 may receive the information for each by receiving parts 141 to 14 that may not be delayed or are generally processed. n を受信することによってそれぞれの情報を受信してもよい。

[0038] Figure 2 shows a schematic block diagram of a decorrelator 20 according to an embodiment. The decorrelator 20 is configured to receive an audio signal 22. The decorrelator 20 may include a conversion unit 24 configured to generate the frequency representation 14 shown in FIG. 1. The conversion unit 24 may provide parts 141 to 14 obtained by an exemplary STFT. For example, the frequency representation may include a total of 129 frequency bins. Alternatively, 128 bins may be used. For example, two types of digital Fourier transforms (DFTs), so-called “even stacking” and “odd stacking”, may be used. For example, as a “standard” DFT, the even stacking version may be considered to have 129 bands (127 complex numbers, 1 real number, and 1 imaginary number) in the provided example. The odd stacking may include 128 (complex number) bands. In the embodiments described herein, both conversions can be used. Parts 141 to 14 16 を提供し得る。例えば、周波数表現は、合計129個の周波数ビンを含み得る。あるいは、128個のビンが使用されてもよい。例えば、いわゆる「偶数積み重ね」および「奇数積み重ね」の2種類のデジタルフーリエ変換(DFT)が使用され得る。例えば、「標準」DFTとして、偶数積み重ねバージョンは、提供される例では129バンド(127複素数、1つの実数および1つの虚数)を有すると見なされ得る。奇数積み重ねは、128個の(複素数)バンドを含み得る。本明細書に記載の実施形態では、両方の変換が使用されることができる。部分141から1416 The parts may contain the same or different number of bins, partially or completely. For example, part 141 may contain bins 1 through 9, e.g., 9 bins. Part 142 may contain bins 10 through 19, and thus 10 bins. The adaptation or selection regarding the number of bins depends on the sampling frequency, which is 48 kHz in the illustrated example, the overlap, e.g., 50%, and / or the resulting parts 141 through 14 16 It can be obtained based on the number of parts 141 to 14 16 some or all parts 141 to 14 16 It may contain an equal or different number of frequency bins, so that it can be generated to contain the same number of frequency bins.

[0039] The correlation remover 20 connects delay lines 121 to 12 16 Further comprising a delay section 25 having each delay line 121 to 12 16 This refers to specific parts 141 to 14 16 It is associated with and configured to receive the aforementioned portion and its processed version, respectively. Delay units 121 to 12 16 These are the respective spectral parts 141 to 14 16 It may be associated with such delay units 121 to 12 16 This can be configured to delay the relevant portion of the frequency representation 14 differently compared to the delay units associated with other spectral portions. Alternatively or additionally, the relationship between the different time delays may be linear, logarithmic, and / or based on superband sample rounding.

[0040] The correlation remover 20 further comprises a phase shifter 26 coupled to the delay section 25, the phase shifter 26 from the delay section 14'1 to 14' 16 It is configured to receive. Phase shifting using phase shifter 26 can enable reverberation in the signal portion. However, according to the embodiment, the sequence of delay section 25 and reverberation section 26 is also divided into parts 141 to 14 16It may be modified so that it is first subjected to the reverberation filter and then delayed.

[0041] The phase shifter 26 may be configured to phase shift the frequency representation 14 of the processed, for example, delayed version of the audio signal. The phase shift may also be performed before converting the audio signal 22 to the frequency domain, and the corresponding phase shifter may be configured to phase shift the audio signal 22 in the time domain to obtain the phase-shifted audio signal. In a short configuration in which the phase shifter 26 is configured to phase shift the frequency representation of the audio signal 14 and its delayed version, the phase shifter is configured to phase shift multiple all-pass filters 281 to 28 16 It may be equipped with. In the illustrated example, all-pass filters 281 to 28 16 The delayed portion is from 14'1 to 14' 16 It is configured to receive. The term all-pass filter means that the frequency range that is passed is divided into parts 141 to 14 16 It should be understood that this corresponds to the frequency range of 281. Here, this corresponds to all-pass filters 281 to 28 16 Each of these may include an example that passes through the full frequency range provided in frequency representation, and different all-pass filters 281 to 28 16 The passing bands are also divided into sections 141 to 14. 16 They may differ from each other based on the different frequency bins they contain.

[0042] All-pass filters 281 to 28 16 Each of these is configured to phase-shift the relevant portion of the frequency representation of the audio signal.

[0043] In other words, the number of all-pass filter structures and / or the circuits of the all-pass filter structures may be the same, i.e., equal or equivalent, or different all-pass filters 281 to 28 16 They may differ between them.

[0044] Delay line 121 to 12 16 The time delay provided by is different from part 141 to 14 16 This may be the same or different for each other. As shown in Figure 2, the portion of the frequency representation containing lower frequencies may be delayed with a higher time delay compared to the portion of the frequency representation containing higher frequencies. The represented frequencies may increase from bin 1 to higher bins. As shown in the z domain, the time delay may decrease with increasing frequency.

[0045] Signals 321 to 32 16 For example, all-pass filters 281 to 28 16 The output may include the results of delay and phase shift.

[0046] The envelope shaper 16 controls signals 321 to 32 16 and its unfiltered and undelayed version, i.e., parts 141 to 14 16 , that is, it may be configured to receive the frequency representation of the audio signal 22. Parts 141 to 14 16 This can be understood as a subband. The envelope shaper 16 may be configured to operate in the subband region. For example, the time resolution of the envelope shaper 16 may be at most 4 milliseconds or less, for example, 4 milliseconds, 3.5 milliseconds, or 3 milliseconds or less.

[0047] The decorrelation unit 20 may include another conversion unit 35 that can provide an inverse operation when compared to the conversion unit 24. For example, the conversion rate 34 may perform an inverse short-term Fourier transform iSTFT. The composite shape-frequency representation 18 may include information about the frequency domain present in each bin so that the composite shape-frequency representation 18 can be processed in accordance with the output of the conversion unit 24. That is, the conversion unit 34 includes parts 141 to 14 of the frequency representation 14. 16 The processed version is received, and processed version 14'1 to 14' is based on the overlap addition procedure, for example. 16A combined signal 36 can be synthesized from these signals. The signal 36 may be provided, for example, at the interface 38 of the correlation remover 20.

[0048] The envelope shaper 16 may be configured to shape the spectral bins in time and / or frequency. Shaping can be performed by the envelope shaper 26 by implementing, for example, interdependent or at least common shaping processes for individual bins and / or groups of bins.

[0049] Referring again to the conversion unit 24, it receives the audio signal 22 or its reverberation version and parts 141 to 14 16 It may be configured to convert to, and 16 is just one example. The reverberation version of the audio signal 22 may also be the input when the phase shifter 26 operates in the time domain, and therefore may be placed upstream of the conversion unit 24. The conversion unit 24 may perform a discrete Fourier transform, DFT, or short-time Fourier transform, STFT, on a time-block basis. The conversion unit may be configured to convert blocks with, for example, a 50% overlap within a tolerance range. For example, the tolerance range may be 0% if possible, 5% at most, 10% at most, 15% at most, or more.

[0050] The block may include block lengths of, for example, 128, 256, or 512 samples, with 256 being a preferred value.

[0051] Figure 3 shows a schematic block diagram of the decorrelation block 30. Compared to the decorrelation block 20, the decorrelation block 30 may further include a pre-delay 42, the term of which does not limit the delay to any delay implemented immediately before or after any particular block. The pre-delay 42 may be placed when operating in the frequency domain at any stage before the envelope shaper 16, preferably after the conversion unit 24. That is, for example, the sequence between the all-pass filter of the reverberation or phase shifter 26 and the pre-delay 42 may be swapped when compared to the diagram in Figure 3. The pre-delay 42 or delay block 42 is part of the frequency representation, from part 141 to part 14 16 It may be configured to further implement the same predetermined delay for a subset or all of them. This combines the processing at this stage for each part 141 to 14 16 Alternatively, delay lines 121 to 12 could be used to add a separate delay, possibly different from the common delay implemented in block 42, which could allow for the same delay implementation for that group. 16 This can be used. In one example, the pre-delay 42 is configured to allow a constant pre-delay for all spectral bands.

[0052] Figure 4 shows the filters 281 to 28 of the correlation remover 20 and / or 30. 16 A schematic block diagram of an all-pass filter 40 according to an embodiment that may operate as at least part of one of the all-pass filters 40 is shown. The all-pass filter 40 may include, for example, the structure of a Schrader IIR filter and may include a forward branch 46 combined with a reverse branch 48 combined with a delay block 52 to provide respective output signals 54 based on the input signals 44 of the all-pass filter 40. The all-pass filters 28 of the correlation removers 20 and / or 30 may include one or more such all-pass filters 40 connected in series with each other. Different all-pass filters 281 to 28 16 To provide different time delays, different numbers of all-pass filter structures 14 can be connected in series. In other words, Figure 4 shows the all-pass filter stage.

[0053] Figure 5 shows a schematic block diagram of an all-pass filter structure 50, which is a nested all-pass filter structure. Alternatively, or in addition to the all-pass filter structure 40, one or more all-pass filter structures 50 are all-pass filters 281 to 28 of the correlation remover 20 and / or 30. 16 This can form at least a part of the following. Although two delay blocks 521 and 522 are shown, there may be different, in particular more delay blocks 52, and as a result, the number of forward branches 46 and / or reverse branches 48 may increase in some cases. Furthermore, gains g1 / -g1 and / or g2 / -g2 may be employed.

[0054] For example, if we consider connecting the delay block 52 in series with one or more all-pass filter structures 40 and / or one or more all-pass filter structures 50, then different all-pass filters 281 to 28 16 It may be implemented to include different time delays compared to other all-pass filters. For example, different all-pass filter structures and / or different delays of circuits of all-pass filter structures may be based on prime multiples of the local sampling rate used to obtain the frequency representation 14 of the audio signal 22, e.g., 48 kHz. For example, a set of all-pass filter structures forming at least part of an all-pass filter may comprise four all-pass filter structures, e.g., all-pass filter structure 40. Different delay blocks within it may be adapted to provide delays of 1, 2, 3, and 5. According to another example, the number of four all-pass filter structures may provide delays of 1, 3, 5, and 7 units in the z domain. These values ​​may also form a set of prime numbers, i.e., a number of prime numbers of 2, 3, 4, 5 or more may be grouped together.

[0055] In this embodiment, when a set of prime numbers is transferred to the possible operation of the all-pass filter in the time domain, the time delay is based on a prime multiple of the reciprocal of the sampling rate used to obtain the frequency representation of the audio signal in this embodiment. For example, different time delays may be based on primes obtained by multiplying each of the above-mentioned set of primes, e.g., 1, 2, 3, and 5 or 1, 3, 5, and 7, by a downsampling coefficient used to generate a portion of the frequency representation of the audio signal to obtain an intermediate result. Instead of the intermediate result, the following primes may be used for the intermediate result. For example, referring to a downsampling coefficient of 128 and considering the above set of primes, such results could be delays of 131, 257, 383, and 641 on the one hand, and 131, 383, 641, and 907 on the other hand, with each delay relating to multiplication by one sample at a sampling rate that is approximately 20.8 μs for a sampling rate of 48 kHz. Other sets of primes are possible without limitation.

[0056] For example, referring to Figure 4, the gain coefficient g of the all-pass filter can be fitted to a value of 0.7 within a tolerance range of, for example, ±20%, ±10%, or ±5%. However, the gain value may also have a negative value of, for example, -0.7 within the aforementioned tolerance range. That is, the gain coefficient can be fitted to a value of magnitude 0.7 within the tolerance range.

[0057] In other words, in addition to the serial outpass configuration shown in Figure 4, nested configurations in which the delay element of the outer Schrader allpass is replaced with that of other inner allpass configurations, or combinations of both, may be implemented. Figure 5 shows a simple nested allpass filter stage.

[0058] Figure 6 shows a schematic block diagram of a correlation remover 60 according to an embodiment. The correlation remover 60 comprises a phase shifter 26 configured to operate in the time domain. The all-pass filter structure 28' may be configured to use the next prime number when compared to a set of prime numbers, as described in relation to correlation removers 20 and / or 30. To ensure the accurate operation of the correlation remover 60, it may be provided with conversion units 241 and 242. Conversion unit 241 may provide a frequency representation of the audio signal, while conversion unit 242 may receive the reverberation audio signal or phase-shifted audio signal 22' provided by the phase shifter 28'. Obtained parts 14''1 to 14'' 16 This allows for time-domain-based reverberation, while delay units 121 to 12 reach equivalent inputs to the envelope shaper 16 when compared to the correlation removers 20 and / or 30. 16 This can be delayed. That is, the frequency representation portion can be formed from the reverberation audio signal 22'.

[0059] According to the embodiments, the correlation remover described herein may be combined with further functions, i.e., the output signal may be further processed. In other words, Figure 6 shows an alternative implementation of the correlation remover for Figure 2.

[0060] Furthermore, the correlation decorruptor of the present invention may be combined with a transient processing process. Transients can cause artifacts in the decorrupted stereo signal, such as aftermath echoes or undesirable panning effects. To mitigate this, transient processing can be combined with the correlation decorruptor described herein. Transient processing may mute the correlation decorruptor output to preserve the direct start waveform and suppress aftermath echoes caused by pre-delays.

[0061] Figure 7 shows a schematic block diagram of a correlation remover 70 according to an embodiment. The correlation remover 70 comprises at least a portion of the correlation remover 10, and alternatively or additionally, at least a portion of the correlation removers 20, 30 and / or 60 may be included. The correlation remover 70 may include a signal processing stage 56 configured to process a synthesized and shaped frequency representation 18 or a signal based thereon. The synthesized and shaped frequency representation 18 may be considered a monaural signal, i.e., it may represent a single channel. From the received monaural signal, the processing stage may provide signals 581 and 582 representing at least a stereo signal.

[0062] A source expander 58 may be coupled to the decorrelator 70 to model the perceptual effect of a spatially extended source from a monaural signal of a point source and its decorrelated version. The source expander 58 may comprise filters 641 to 642 that enable source expansion modeling based on a stereo signal having signals 581 and 582. Source expansion modeling may be performed, for example, in the frequency domain to yield stereo output signals 641, e.g., the left channel and 642, e.g., the right channel. It should be noted that the source expander 58 may also form part of the decorrelator 70. In other words, Figure 7 shows a schematic block diagram of the source extension process.

[0063] Figure 8 shows a schematic block diagram of a processing system 80 according to an embodiment. The processing system 80 may include a decorrelation 10. Alternatively or additionally, decorrelation 20, 30, 60 and / or 70 may be provided. The processing system 80 includes a processing stage 66 configured to convert a center / side-decomposed signal 68 into a left / right-decomposed signal 72. That is, the center / side-decomposed signal 68 may include, for example, a first signal 741 representing one of the center / intermediate or side portions and a second signal 742 representing the other. The processing stage 66 may be configured to convert signals 741 to 742 and optionally additional signals into signals 761 to 762 representing at least the left channel and the right channel. One channel, for example the left channel L, may be obtained by adding, for example, an intermediate component M and a side component M+S. The other channel, for example the right channel, may be obtained by subtracting one component from the other, for example by MS. According to different methods, both channels can be obtained by using 50% or 0.5 times their values, i.e., 0.5(M+S) and 0.5(MS). Other coefficients and / or decision rules are also possible.

[0064] According to one embodiment, signal 741 is provided by a decorrelator in the processing system 80. Another signal 742 may be provided by a delay compensation unit 78, which is connected in parallel to the decorrelator 10 and configured to also receive the audio signal 22. Thus, the delay compensation unit 78 is connected to the processing stage 66. The delay compensation unit 78 may be configured to provide a time delay comparable to that of the decorrelator. Preferably, in the frequency domain embodiment, the delay is equal to the processing delay introduced by the STFT analysis / synthesis of the decorrelator. However, the decorrelator 10 may provide additional signal processing leading to decorrelation so that signal 742 may include a similar delay when compared to signal 741. According to one embodiment, signal 742 may be untouched except for the time delay.

[0065] The decorrelation analyzer 10 within the processing system 80 may provide the processing stage 66 with a synthesized frequency representation as at least a portion of the center / side-decomposed signal. The processing stage 66 may convert the synthesized frequency representation, together with the delayed signal 742, into left / right-decomposed signals in the frequency domain. The output of the processing stage 66 may be an L / R signal 72. The decorrelation analyzer 10 itself may generate a monaural signal S (side, component 18), and in that respect, it is only a part of it. In transient processing, the signal S is muted and "replaced" by an amplified M signal (signal 74'2), so the direct component M (742; 74'2) and the decorrelation analyzer output S (signal 18) can be closely coupled. As a result, both units, the decorrelation analyzer and the "upmixing unit" 66, are closely coupled, and therefore the processing stage 66 ultimately provides a decorrelation stereo signal. For example, if the decorrelation converter operates independently on the monaural output without the processing stage 66, the delay-compensated direct signal must be added directly to the monaural output without any scaling to fill the muted gap and provide a "complete" signal.

[0066] In other words, Figure 8 shows a correlation remover in an M / S to L / R setup with delay compensation for mono (intermediate signal) input.

[0067] Figure 9 shows a schematic block diagram of a processing system 90 according to an embodiment. Compared to processing system 80, processing system 90 includes a transient suppressor 82 configured to detect transients in the audio signal 22 or its frequency representation 14 at the input of the decorrelation analyzer. The transient suppressor may include a transient detection unit 84 configured to receive the audio signal 22 or its frequency representation. The transient detection unit 84 may detect transients in the audio signal, for example, by processing the audio signal 22. The transient suppressor 82 may further include a mute unit 86 configured to receive the synthesized and shaped frequency representation 18 and mute it based on a control signal. However, it should be noted that the same or equivalent effect can be obtained by controlling the decorrelation analyzer included in processing system 90 to mute the decorrelation analyzer 10 or the output of the decorrelation analyzer. That is, the mute unit 86 may also constitute part of the decorrelation analyzer. However, the signal 741 forming the input of processing stage 66 may be muted based on transients detected in the audio signal 22. The transient suppressor 82 may be configured to temporarily mute the portion provided by the decorrelation remover in order to suppress echoes in the processing stage 66, and the echoes may be related to pre-echoes and / or post-echoes. When operating in the time domain, a window for soft muting may be used to avoid additional transients caused by muting. When performed in the frequency domain, the STFT windowing process described in relation to the decorrelation removers 20, 30, and 60 may automatically, i.e., synergistically, provide such an effect.

[0068] With respect to the processing stage 66, muting the output of the correlation rejector 10 may result in an undesirable shift in the input energy of the signal processing stage 66. To avoid this adverse effect, an amplifier 82 may be connected between the delay compensation unit 78 and the signal processing stage 66 to temporarily amplify the signal 742 to obtain an amplified signal 74'2. The amplification of the signal 742 may be conditional on muting the output of the correlation rejector 10. That is, the transient suppressor 82 may be configured to amplify the portion of the delay compensation unit 78 that corresponds to muting the portion of the correlation rejector.

[0069] The amplification level may be fixed or controlled. For example, if applicable, the amplification factor of amplifier 82 is compared to the unmuted portion of the correlation rejector. The rate may be TIFF0007832956000001.tif126. That is, when the output of the correlation remover is muted, amplifier 88 is Signal 742 can be amplified by TIFF0007832956000002.tif126, but it will not be amplified when the mute is off, i.e., when g=1.

[0070] Optionally, and to avoid undesirable effects during transient suppression, the transient suppressor 82 may be configured to suppress detected transients in the audio signal and to suppress subsequent transients that do not occur earlier than a predetermined suppression time. For example, the transient suppressor 82 may include a control unit 92 configured to control and / or apply hold time, hysteresis, and / or suppression time. For example, the hold time may be shorter than the inhibition time. The hold time may relate to the time during which the output of the correlation remover 10 is muted in response to the detected transient, i.e., the characteristics determined by the transient detection unit 84. The inhibition time may be longer than the hold time to avoid undesirable effects. For example, the hold counter, i.e., the time for muting, may be 1, 2, 4, 6, 7, or 8 blocks, while the inhibition time may be at least twice that time, e.g., at least 14, at least 20, at least 30, or 56 blocks, or any other arbitrary duration.

[0071] For example, the control unit 92 may also provide hysteresis to mitigate the on / off toggle of transient suppression of audio signals, such as low-rate pulse trains. That is, the suppression time provided by the control unit 92 may be a first suppression time. The transient suppressor 82 may be configured to restart the suppression time as a second suppression time that is longer than the first suppression time if a transient occurs during the first suppression time. That is, if the hold time has elapsed but the suppression time has not yet elapsed and a new transient is detected (regardless of whether the hold time has elapsed or not), the suppression timer may be restarted. Optionally, the restarted suppression timer may be longer than the canceled suppression timer. In other words, when a very first transient is detected, both the hold counter and the suppression counter are started. The transient may be muted until the hold counter reaches its stop count, for example, 8 blocks. Then the hold counter may be reset and the mute may be stopped. The suppression counter may reach its stop count / reset with a considerable time delay, for example, at 56 blocks. If a new transient is detected during the ongoing inhibit counting process, only the inhibit counter is restarted, but the stop count value is higher, for example, 64 blocks. Thus, hysteresis is implemented by conditional switching and stop count modification; that is, new triggers for transient suppression or muting may be disabled while the inhibit counter is running.

[0072] The transient suppressor 82 may be configured to operate in the frequency domain. Alternatively or additionally, the transient suppressor 82 may be configured to mute a portion of the correlation rejector for a longer time compared to the prior delay of the correlation rejector. That is, if a transient is detected in the audio signal 22, the mute should still be effective when the transient reaches the output of the correlation rejector.

[0073] In other words, the decorrelator according to the embodiment operates in the Short-Time Fourier Transform (STFT) domain on overlapping transform blocks with short durations. This allows for small processing delays of a few milliseconds, e.g., 2.7 milliseconds for transform sizes at sample rates of 256 and 48 kHz, in contrast to the high delays of the PS / MDS decorrelators described in [2] or [3], which can reach delay times of 13.3 milliseconds at a sample rate of 48 kHz. Furthermore, the decorrelator described can be implemented using a very low computational all-pass filter and is therefore computationally much more efficient than the time-domain decorrelators described in [1] or [2]. If further downstream spectral processing, e.g., source extension modeling, is required or desired, the decorrelator described can interface directly with this processing stage in the STFT domain to achieve low computational complexity.

[0074] Therefore, the decorrelators described herein may offer short processing latency and moderate computational complexity. The decorrelators can be combined with additional downstream processing for modeling audio objects having spatial dimensions, such as spatially augmented sound sources (SESS) with perceptual properties of so-called "sound source augmentation."

[0075] In other words, Figures 2 and 9 illustrate preferred embodiments of the present invention. First, an input signal or audio signal (e.g., sound from a point source) can be fed into a correlation remover 20 which includes a DFT per time block having, for example, a 256-sample block length and, for example, 50% overlap. Next, the spectral bins of the DFT are time-delayed over frequency-dependent durations, with lower frequencies having higher delays and higher frequencies having lower delays. For example, the delay could be 16 subband samples at low frequencies (42.7 milliseconds at 48 kHz), and the highest bin, i.e., z -1The number of subband samples can then be reduced to one. The reduction in delay over time can be linear, logarithmic, or rounding of the subband samples to an integer. Each bin is then passed through an all-pass filter, preferably comprising a chain of simple all-pass filters or a nested all-pass filter structure. An example of an all-pass filter is shown in Figure 4. A different structure is shown in Figure 5. With respect to Figure 4, one possible chain may include or consist of four such all-pass filters. The parameter g may be chosen to be, for example, 0.7, and the delay M i M may be a prime number. Note that Figure 4 shows the very first part of the chain, namely M1. Since these filters can operate on downsampled spectral bands, for example with a downsampling factor of 128, the delay is very low and can be, for example, prime numbers 1, 2, 3, and 5, or as another example, 1, 3, 5, and 7. Subsequently, time / frequency envelope shaping may be applied. The input signal to envelope shaping may be the direct DFT bins and their delayed and filtered versions. Finally, IDFT with overlap summation may synthesize the output signal. The output signal may be further processed in the time domain to obtain a left / right stereo signal from a monaural input signal in a configuration such as that shown in Figure 8. Alternatively, the left / right stereo signal may be assembled in the DFT frequency domain and further processed in the frequency domain, for example, for source expansion / SESS modeling by fast convolution, if it is beneficial to the overall computational efficiency. A configuration for source expansion modeling is shown in Figure 7. In contrast to other embodiments, delay M iAlternative embodiments may be selected as prime numbers that are about 128 times larger than those selected in the subband region, for example, 131, 257, 383, and 641 (for the set of prime numbers 1, 2, 3, and 5) or 131, 383, 641, and 907 (for the set of prime numbers 1, 3, 5, and 7). For different sets of prime numbers and / or prime numbers having different prime numbers, corresponding values ​​may be selected. Furthermore, alternative embodiments may require an additional STFT to obtain the direct signal input to the time / frequency envelope shaper.

[0076] Figure 9 shows an exemplary correlation remover in an M / S to L / R setup with transient handling. The aspects of these embodiments are as follows: Transient phenomenon detection detects the presence of isolated transient phenomena.

[0077] If a transient is detected, the decorrelated sound is muted for the "hold time," and the delay-compensated direct signal is amplified accordingly. To compensate for the effect of coherent summation, a coefficient of 2 / sqrt(2) is applied to amplify the direct signal, thereby replacing the decorrelated signal.

[0078] To avoid triggering high-speed pulse trains perceived as tones, inhibition prevents triggering by the next transient during a specific "inhibition time." During the "holding time," the inhibition time is restarted each time a new transient is detected.

[0079] Hysteresis prevents toggling of transient detection (for example, by increasing the "inhibition time" in the case of re-induced inhibition).

[0080] Transient detection, muting, direct sound amplification, detection inhibition, and hysteresis may be advantageously implemented in the STFT domain as follows: ○STFT block overlap provides a smooth crossfade. ○ The mute time is longer than the pre-delay of the correlation remover. ○ Mute block counter for muting the correlation-removed signal and amplifying the direct signal. ○ Inhibition block counter to inhibit transient phenomenon detection ○ Hysteresis to avoid toggling in transient detection

[0081] Embodiments of the present invention relate to the following: Apparatus / method for decorrelation of audio signals, • A correlation remover, ○DFT / IDFT pair (optional if directly interfaced with SESS processing in the frequency domain), ○ A delay in the subband region, preferably having a higher delay at lower frequencies and a lower delay at higher frequencies, with a frequency-aligned delay distribution such as linear or logarithmic. A correlation remover comprising: an all-pass filter in the subband region, wherein the lower frequencies can optionally have a higher delay / order, the higher frequencies can have a lower delay / order, and the higher-order all-pass filter can be realized by a stage of lower-order all-pass filters; A short-Schrader IIR filter in the DFT subband domain (downsampled) that uses a small integer delay prime in combination with a frequency-variable delay, A short-Schrader IIR filter operates in the subband region, with a high-time-resolution (<4ms) T / F envelope tuner that measures the energy before and after delay / all-pass processing and adjusts the energy of the subband signal to match (as much as possible) the energy of the original subband signal. • A low-latency decorrelation analyzer (in contrast to an MPEG surround decorrelation analyzer) as part of "source extension" modeling / processing, • Interfaces to downstream source extension processing in the time or DFT frequency domain for computational efficiency, Alternative implementation: An apparatus / method comprising an all-pass filter before the delay ("after the delay").

[0082] Figure 10 shows a schematic block diagram of Method 1000 according to an embodiment that may be implemented by, for example, a correlation remover described herein. Method 1000 includes step 1010, in which multiple parts based on an audio signal are received. In 1020, each received part is delayed to provide multiple delayed parts. 1030 includes receiving and combining signals based on the delayed parts of the frequency representation. 1040 includes receiving the frequency representation of the audio signal. 1050 includes adjusting the energy of the delayed parts with respect to the frequency representation of the audio signal. 1060 includes providing a synthesized and shaped frequency representation, for example, using an envelope shaper 16.

[0083] While several embodiments have been described in the context of the apparatus, it is clear that these embodiments also represent descriptions of the corresponding methods, where blocks or apparatus correspond to method steps or features of method steps. Similarly, embodiments described in the context of method steps also represent descriptions of the corresponding blocks, items, or functions of the corresponding apparatus.

[0084] The encoded audio signal of the present invention can be stored in a digital storage medium or transmitted over a transmission medium such as a wireless transmission medium or a wired transmission medium such as the Internet.

[0085] Depending on specific implementation requirements, embodiments of the present invention may be implemented in hardware or software. Implementation can be carried out using digital storage media such as floppy disks, DVDs, CDs, ROMs, PROMs, EPROMs, EEPROMs, or flash memory, which store electronically readable control signals and cooperate (or can cooperate) with a computer system that is programmable to perform each method.

[0086] Some embodiments of the present invention include a data carrier having an electronically readable control signal that can cooperate with a programmable computer system so that one of the methods described herein is performed.

[0087] Generally, embodiments of the present invention can be implemented as a computer program product comprising program code, which operates to perform one of the methods when the computer program product is executed on a computer. The program code may be stored, for example, in a machine-readable carrier.

[0088] Other embodiments include a computer program stored in a machine-readable carrier for performing one of the methods described herein.

[0089] In other words, embodiments of the method of the present invention are, therefore, computer programs having program code for performing one of the methods described herein when the computer program is executed on a computer.

[0090] Therefore, a further embodiment of the method of the present invention is a data carrier (or digital storage medium, or computer-readable medium) on which a computer program for performing one of the methods described herein is recorded.

[0091] Therefore, a further embodiment of the method of the present invention is a data stream or sequence of signals representing a computer program for performing one of the methods described herein. The data stream or sequence of signals may be configured to be transmitted over a data communication connection, such as the Internet.

[0092] Further embodiments include processing means configured or adapted to perform one of the methods described herein, such as a computer or a programmable logic device.

[0093] Further embodiments include a computer on which a computer program for performing one of the methods described herein is installed.

[0094] In some embodiments, a programmable logic device (e.g., a field-programmable gate array) may be used to perform some or all of the functions of the method described herein. In some embodiments, a field-programmable gate array may cooperate with a microprocessor to perform one of the methods described herein. Generally, the method is preferably performed by any hardware device.

[0095] The embodiments described above are merely illustrative of the principles of the present invention. Modifications and variations of the configurations and details described herein will be apparent to those skilled in the art. Therefore, the invention is intended to be limited only by the imminent claims and not by the descriptions of the embodiments herein or by the specific details presented herein. [Prior art documents] [Non-patent literature]

[0096] [Non-Patent Document 1] [1] W. Oomen, E. Schuijers, B. den Brinker, and J. Breebaart, "Advances in Parametric Coding for High-Quality Audio," Paper 5852, (2003 March.)

[0097] [Non-Patent Document 2] [2] J. Breebaart, S. van de Par, A. Kohlrausch, and E. Schuijers, "High-quality Parametric Spatial Audio Coding at Low Bitrates," Paper 6072, (2004 May.)

[0098]

Table 3

[0099]

Fashion 4

Claims

1. It is a correlation remover, A plurality of delay units (12), each delay unit (12) is a portion of the frequency representation (14) based on the audio signal (22). １ ~14 ｎ ) is configured to receive, and each delay unit (12) receives the receiving portion (14 １ ~14 ｎ ) delays the delayed portion (14' １ ~14' ｎ Multiple delay units configured to provide, The delay part (14’ １ ~14’ ｎ ) of the frequency representation provided by the plurality of delay units (12), receive the frequency representation of the audio signal (22), and combine the signals based on the delay part (14’ １ ~14’ ｎ ) of the frequency representation of the audio signal (22), an envelope shaper (16) configured to adjust the energy of the synthesized shaped frequency representation and provide the synthesized shaped frequency representation. The system includes a phase shifter (26) configured to phase shift the frequency representation (14) of the audio signal (22), or to phase shift the audio signal (22) in the time domain to obtain a phase-shifted audio signal (22), Correlation remover.

2. The portion of the frequency representation that differs (14 １ ~14 ｎ The correlation remover according to claim 1, wherein the ) includes the same or different number of frequency bins.

3. The phase shifter (26) is configured to phase shift the frequency representation of the audio signal (22), and comprises a plurality of all-pass filters, each all-pass filter (28) shifts the relevant portion (14) of the frequency representation of the audio signal (22). １ ~14 ｎ The correlation remover according to claim 1 or 2, configured to phase-shift )

4. One of the plurality of all-pass filters (28) comprises a set of all-pass filter structures (40; 50) connected in series with each other, and the all-pass filter structures (40; 50) are adapted to provide different time delays, or The correlation remover according to claim 3, wherein the all-pass filter structure (40; 50) comprises a nested all-pass filter structure.

5. Correlation remover according to claim 4, wherein the number of all-pass filter structures (40; 50) and / or the circuits of the all-pass filter structures are equal or different among different all-pass filters (28).

6. The decorrelation remover according to claim 4 or 5, wherein the different time delays are based on prime multiples of the local sampling rate used to obtain the frequency representation of the audio signal (22).

7. Correlation remover according to any one of claims 4 to 6, wherein the set of all-pass filter structures (40; 50) comprises four all-pass filter structures (40; 50), each adapted to provide delays for samples 1, 2, 3, and 5 or for samples 1, 3, 5, and 7.

8. The correlation remover according to any one of claims 3 to 7, wherein the gain coefficient of the all-pass filter (28) is adapted to a value having a magnitude of 0.7, for example, within a tolerance range of 20%.

9. The phase shifter (26) is configured to phase shift the audio signal (22) in the time domain, and the phase shifter (26) comprises a set of all-pass filter structures (40; 50) connected in series with each other, and the all-pass filter structures (40; 50) are adapted to provide different time delays, or The correlation remover according to claim 1 or 2, wherein the all-pass filter structure (40; 50) comprises a nested all-pass filter structure.

10. The decorrelation remover according to claim 9, wherein the different all-pass time delays are based on prime multiples of the reciprocal of the sampling rate used to obtain the frequency representation of the audio signal (22).

11. Each of the aforementioned different time delays corresponds to the portion of the frequency representation (14) of the audio signal (22) in order to obtain an intermediate result. １ ~14 ｎ A decorrelation remover according to claim 9 or 10, based on a prime number obtained by multiplying it by a downsampling coefficient 128 used to generate and use the following prime number for the intermediate result.

12. The envelope shaper (16) comprises a first conversion unit (24) for obtaining the frequency representation of the audio signal (22) from the audio signal (22), and a second conversion unit (34) for obtaining the frequency representation from the audio signal (22), and the portion (14) of the frequency representation. １ ~14 ｎ ) is the portion of the frequency representation (14) from the audio signal (22). １ ~14 ｎ A correlation remover according to any one of claims 9 to 11, which forms a correlation remover.

13. The portion of the frequency representation (14 １ ~14 ｎ A correlation remover according to any one of claims 1 to 12, wherein the frequency bins include an equal or different number of frequency bins.

14. The 16 parts of the frequency representation (14 １ ~14 ｎ A correlation remover according to any one of claims 1 to 13, which is adapted to obtain ).

15. A correlation remover according to any one of claims 1 to 14, which is adapted to obtain the frequency representation having 128 or 129 frequency bins.

16. A correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), The system includes an envelope shaper (16) configured to provide a synthesized and shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22), so as to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12), The correlation remover removes a subset or all of the frequency representation (14 １ ~14 ｎ A correlation remover adapted to further implement the same predetermined delay for ).

17. A correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), The system includes an envelope shaper (16) configured to provide a synthesized and shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22), so as to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12), Of the plurality of delay units (12), the spectral portion (14 １ ~14 ｎ The delay unit (12) associated with ) other spectral portions (14 １ ~14 ｎ When compared with the delay unit (12) associated with the frequency representation, the relevant portion (14 １ ~14 ｎ A correlation remover configured to delay ) in different ways.

18. A correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), The system includes an envelope shaper (16) configured to provide a synthesized and shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22), so as to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12), The plurality of delay units (12) include a portion of the frequency representation (14) that includes a higher frequency. １ ~14 ｎ The portion of the frequency representation (14) includes lower frequencies that have a higher time delay when compared to the portion of the frequency representation (14) １ ~14 ｎ A correlation remover configured to delay ).

19. Correlation remover according to claim 18, wherein the relationship between different time delays is one of linear, logarithmic, and / or based on rounding of subband samples to integers.

20. By performing a discrete Fourier transform (DFT) or short-time Fourier transform (STFT) on each time block, the audio signal (22) or the reverberation version of the audio signal (22) is received and the portion (14 １ ~14 ｎ Correlation remover according to any one of claims 1 to 19, comprising a conversion unit (24) for converting to ), wherein the conversion unit (24) is configured to convert blocks having an overlap of 50% within an acceptable range.

21. By performing a discrete Fourier transform (DFT) or short-time Fourier transform (STFT) on each time block, the audio signal (22) or the reverberation version of the audio signal (22) is received and the portion (14 １ ~14 ｎ Correlation remover according to any one of claims 1 to 20, comprising a conversion unit (24) for converting to ), wherein the block includes a block length of 256 samples.

22. A correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), The system includes an envelope shaper (16) configured to provide a synthesized and shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22), so as to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12), A decorrelation device comprising a decorrelation unit (34) for receiving a processed version of the portion of the frequency representation (14) and for synthesizing a composite signal from the processed version based on an overlap-adding procedure.

23. Correlation remover according to any one of claims 1 to 22, wherein the envelope shaper (16) is configured to operate with a time resolution of less than 4 ms in the subband region.

24. Correlation remover according to any one of claims 1 to 23, comprising an interface (38) for providing a signal (36) based on the synthesized and shaped frequency representation.

25. A correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), The system includes an envelope shaper (16) configured to provide a synthesized and shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22), so as to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12), Correlation remover wherein the envelope shaper (16) is for shaping spectral bins individually or as a group in time and / or frequency, for example, by implementing interdependent or at least group-based common shaping processes.

26. Correlation remover according to any one of claims 1 to 25, comprising a signal processing stage (66) configured to receive a signal based on the synthesized and shaped frequency representation as a monaural signal and process the monaural signal into at least a stereo signal.

27. ​​A correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), The system includes an envelope shaper (16) configured to provide a synthesized and shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22), so as to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12), A decorrelation device comprising a signal processing stage (66) configured, for example, for source extension modeling based on at least a stereo signal in the frequency domain, to process the synthesized and shaped frequency representation into at least a stereo audio signal.

28. A processing system, A correlation remover according to any one of claims 1 to 27, A processing system comprising a processing stage (66) that converts a center / side-decomposed signal into a left / right-decomposed signal.

29. A processing system, It is a correlation remover, A plurality of delay units (12), each delay unit (12) configured to receive a portion (141 to 14n) of the frequency representation based on an audio signal (22), and each delay unit (12) configured to delay the received portion (141 to 14n) to provide a delayed portion (14'1 to 14'n), Correlation remover comprising: an envelope shaper (16) configured to provide a composite-shaped frequency representation, such as receiving the frequency representation of the audio signal (22) and adjusting the energy of the delay portion (14'1 to 14'n) with respect to the frequency representation of the audio signal (22) in order to receive and combine signals based on the delay portion (14'1 to 14'n) of the frequency representation provided by the plurality of delay units (12); The system includes a processing stage (66) that converts a center / side decomposed signal into a left / right decomposed signal, One portion of the aforementioned central / side-resolved signal (74 １ ) is provided by the correlation remover, and the other part (74 ２ A processing system in which the processing is provided by a delay compensation unit (78) connected in parallel with the correlation remover and connected to the processing stage (66).

30. The correlation remover includes a transient suppressor (82) configured to detect transient phenomena in the audio signal (22) or its frequency representation (14) at its input, The transient suppressor (82) is provided by the correlation remover to suppress echoes in the processing stage (74 １ The processing system according to claim 29, configured to temporarily mute ).

31. The processing system according to claim 30, wherein the transient suppressor (82) is configured to amplify the portion of the delay compensation unit that corresponds to muting the portion of the correlation remover.

32. When the transient suppressor (82) compares the portion of the delay compensation unit with the unmuted portion of the correlation remover, The processing system according to claim 31, configured to amplify by a certain rate.

33. The processing system according to any one of claims 30 to 32, wherein the transient phenomenon suppressor (82) is configured to suppress the detected transient phenomenon and to suppress subsequent transient phenomena that do not occur earlier than a predetermined inhibition time.

34. The processing system according to any one of claims 30 to 33, wherein the inhibition time is a first inhibition time, and the transient phenomenon suppressor (82) is configured to restart the inhibition time as a second inhibition time that is longer than the first inhibition time if a transient phenomenon occurs during the first inhibition time.

35. The processing system according to any one of claims 30 to 34, wherein the transient suppressor (82) is configured to operate in the frequency domain.

36. The processing system according to any one of claims 30 to 35, wherein the transient suppressor (82) is configured to mute the portion of the correlation remover for a longer period of time compared to the prior delay of the correlation remover.

37. The processing system according to any one of claims 28 to 36, wherein the decorrelation unit provides the composite shaped frequency representation to the processing stage as part of the center / side decomposed signal, and the processing stage converts the composite shaped frequency representation and a delayed version of the audio signal (22) into left / right decomposed signals in the frequency domain.

38. It is a method, Receiving multiple parts of the frequency representation based on the audio signal (1010), In order to provide multiple delay portions, each of the received portions is delayed (1020), The signals based on the plurality of delay portions of the frequency representation obtained by delaying (1020) are received (1030) and combined, Receiving the frequency representation of the audio signal (1040), (1050) Adjusting the energy of the delay portion with respect to the frequency representation of the audio signal, (1060) provides a synthesized and shaped frequency representation, and This includes phase-shifting the frequency representation of the audio signal, or phase-shifting the audio signal in the time domain to obtain a phase-shifted audio signal, or This includes further implementing the same predetermined delay for a subset or all of the frequency representation, This includes delaying the relevant portion of the frequency representation differently when the delay portion associated with a spectral portion among the plurality of delay portions is compared with the delay portions associated with other spectral portions, or The plurality of delay portions include delaying the portion of the frequency representation containing lower frequencies which has a higher time delay when compared with the portion of the frequency representation containing higher frequencies, or This includes receiving a processed version of the portion of the frequency representation and synthesizing a composite signal from the processed version based on an overlap-adding procedure, or Providing the aforementioned synthesized and shaped frequency representation is for shaping spectral bins individually or as a group, temporally and / or frequency-wise, by, for example, implementing interdependent or at least group-based common shaping processes, or A method comprising, for example, performing source extension modeling based on at least a stereo signal in the frequency domain, so as to process the synthesized and shaped frequency representation into at least a stereo audio signal.

39. A computer program for performing the method described in claim 38, when executed on a computer or processor.

Images