MDCT domain error concealment

By identifying the MDCT coefficients of the tone spectrum interval and the noise spectrum interval, estimating and randomly specifying the coefficients of the noise spectrum interval, generating alternative packets to mask the erroneous packets in the MDCT domain, solving the problem of high complexity in the prior art. Effective audio signal error masking and decoding performance improvements are achieved.

CN112967727BActive Publication Date: 2024-11-01DOLBY INTERNATIONAL AB
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202110466890.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Priority Date
2014-12-09
Filing Date
2015-12-08
Publication Date
2024-11-01
Estimated Expiration
2035-12-08

AI Technical Summary

Technical Problem

In the existing audio signal decoding technology, the error masking method of the MDCT domain is highly complex, making it difficult to achieve effective error masking without significantly increasing the complexity.

Method used

By identifying the MDCT coefficients of the tonal spectral interval and the noise spectral interval, estimating and randomly specifying the coefficients of the noise spectral interval, generating alternative packets to mask the wrong packets, and generating decoded frames using IMDCT and overlapping addition operations to reduce complexity.

Benefits of technology

It is realized that error packets in the audio signal are effectively masked without significantly increasing the complexity, reduce distortion of the decoded audio signal, and improve error masking performance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN112967727B_ABST
    Figure CN112967727B_ABST
Patent Text Reader

Abstract

The present invention relates to MDCT domain error concealment. An error concealment audio decoding method comprises: receiving a packet including a set of MDCT coefficients encoding a frame of time domain samples of an audio signal; identifying the received packet as erroneous; generating estimated MDCT coefficients of the set of MDCT coefficients replacing the erroneous packet based on the corresponding MDCT coefficients associated with the received packet immediately preceding the erroneous packet; assigning the signs of a first subset of MDCT coefficients in the estimated MDCT coefficients to be consistent with the signs of the corresponding MDCT coefficients of the previous packet, wherein the first subset includes those MDCT coefficients associated with a tone-like spectral interval; randomly assigning the signs of a second subset of MDCT coefficients in the estimated MDCT coefficients, wherein the second subset includes MDCT coefficients associated with a noise-like spectral interval; replacing the erroneous packet with a concealed packet containing the estimated MDCT coefficients and the assigned signs.
Need to check novelty before this filing date? Find Prior Art

Description

[0001] This application is a divisional application of the invention patent application with application number 201580067028.X, application date December 8, 2015, and invention name “MDCT domain error concealment”. Technical Field

[0002] The invention disclosed herein relates generally to encoding and decoding of audio signals, and in particular, to methods and apparatus for concealing errors. Background Art

[0003] The Modified Discrete Cosine Transform (MDCT) and the corresponding Inverse Modified Discrete Transform (IMDCT) are used, for example, in audio coding and decoding techniques such as MPEG-2 and MPEG-4 audio layers, Advanced Audio Coding, MPEG-4 HE-AAC, MPEG-D USAC, Dolby Digital (Plus), and other proprietary formats.

[0004] In the application of such techniques, errors sometimes occur due to packet loss or packet errors associated with the transformation of the audio signal before or after the packet is received in the decoding system. Such errors include, for example, packet loss or packet distortion, and may result in audible distortion of the decoded audio signal.

[0005] Therefore, methods for concealing errors in the event of errors in packets have been provided. Error concealment methods are generally divided into estimation concealment methods, in which the error frame is replaced by an estimate, and non-estimation concealment methods, such as using silence, frame repetition or noise to replace the error frame.

[0006] The estimation masking method includes a method using estimation in the frequency domain (such as a method disclosed in US Pat. No. 8,620,644) and a method using estimation in the time domain (such as a method disclosed in International Patent Publication No. WO / 2014 / 052746).

[0007] All techniques for error concealment face problems related to the trade-off between the quality of the concealment and the complexity of the required estimation.Therefore, there is a need for other methods of error concealment. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Exemplary embodiments will now be described with reference to the accompanying drawings, in which:

[0009] Figure 1A and Figure 1B Describe the general block diagrams of MDCT and IMDCT respectively by way of example,

[0010] Figure 2 is a generalized block diagram of a first decoding system,

[0011] Figure 3 is a generalized block diagram of a second decoding system, and

[0012] Figure 4 is a generalized block diagram of a third decoding system.

[0013] All the figures are schematic and generally depict only parts which are necessary for the elucidation of the present disclosure, while other parts may be omitted or merely suggested. Unless otherwise indicated, similar reference numerals refer to the same parts in the different figures. DETAILED DESCRIPTION

[0014] In view of the above, it is an object to provide a decoder system and related methods that aim to provide the desired error concealment with insignificant complexity.

[0015] I. Overview - Aspect 1

[0016] According to a first aspect, exemplary embodiments propose a decoding method, a decoding system and a computer program product for decoding. The proposed method, decoding system and computer program product may generally have the same features and advantages.

[0017] According to an exemplary embodiment, a method for concealing errors in data packets to be decoded in an MDCT-based audio decoder is provided, the audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames. The method comprises: receiving a packet including a set of MDCT coefficients from an MDCT-based audio encoder arranged to encode an audio signal, the set of MDCT coefficients being associated with a frame including time domain samples of the audio signal; and identifying the received packet as an erroneous packet when the received packet includes one or more errors. The method also comprises generating an estimated MDCT coefficient of the set of MDCT coefficients that replaces the erroneous packet, the estimated MDCT coefficients being based on corresponding MDCT coefficients associated with a received packet immediately preceding the erroneous packet in the sequence of packets. The method also includes: assigning signs of a first subset of MDCT coefficients in the estimated MDCT coefficients equal to corresponding signs of corresponding MDCT coefficients of a received packet that immediately precedes an erroneous packet in a sequence of packets, wherein the first subset includes these MDCT coefficients associated with tonal-like spectral bins of the packet; and randomly assigning signs of a second subset of MDCT coefficients in the estimated MDCT coefficients, wherein the second subset includes these MDCT coefficients associated with noise-like spectral bins of the packet; generating a concealed packet based on the selected signs of the packet and the estimated MDCT coefficients; and replacing the erroneous packet with the concealed packet.

[0018] As used herein, an "erroneous packet" means comprising MDCT coefficients that differ in some way relative to the MDCT coefficients of the correct MDCT of a correct sample of the audio signal. This may mean that part of the entire packet is lost in the packet sequence, or part of the entire packet includes distortion.

[0019] The identification of the grouped pitch-like and noise-like spectral intervals may be performed using any suitable method.The order of identification of the pitch-like and noise-like spectral intervals is arbitrary and may, for example, depend on the method used.

[0020] It should be noted that the terms "first subset" and "second subset" are only used to distinguish the two subsets from each other in the text, and are not used to indicate the processing order associated with the two different subsets. The order in which the designation is performed is arbitrary. The designation may be performed first for the MDCT coefficients for the first subset and last for the MDCT coefficients for the second subset, or vice versa. In addition, in some exemplary embodiments, the designation may not be performed for the MDCT coefficients, so that all MDCT coefficients associated with the first subset are continuously designated, and all MDCT coefficients associated with the second subset are continuously designated. In some exemplary embodiments, one or more MDCT coefficients of one of the subsets may be designated first, and then one or more MDCT coefficients of the other subset may be designated. In addition, a grouping does not necessarily have MDCT coefficients associated with both noise-like spectral intervals and tone-like spectral intervals. In some exemplary embodiments, a grouping may have all MDCT coefficients associated with noise-like spectral intervals or all associated with tone-like spectral intervals, so that one of the subsets is empty. Finally, MDCT coefficients are generally identified as either belonging to the first subset or to the second subset.

[0021] Note that basing the estimate on MDCT coefficients and MDCT coefficient signs associated with a received packet that immediately precedes the erroneous packet in the sequence of packets does not exclude that the estimate may additionally be based on MDCT coefficients and MDCT coefficient signs associated with a received packet earlier in the sequence of packets than the packet that immediately precedes the erroneous packet.

[0022] As used herein, "generating estimated MDCT coefficients" involves assigning values ​​to the MDCT coefficients that are not necessarily the best approximation of the values ​​the MDCT coefficients would have if no errors were present in the error packet, but rather achieve the desired error concealment performance so that unwanted distortion of the decoded audio signal is avoided or reduced.

[0023] As used herein, "estimated MDCT coefficients" refer to the absolute values ​​of the estimated MDCT coefficients.

[0024] According to an exemplary embodiment, the method also includes: for each estimated MDCT coefficient, determining whether the MDCT coefficient is associated with a tone-like spectral interval or a noise-like spectral interval based on an approximate spectral peak detection of a power spectrum associated with an erroneous packet, wherein the approximate power spectrum is based on a power spectrum associated with a received packet that immediately precedes the erroneous packet in a sequence of packets.

[0025] According to some embodiments, the method further comprises determining, for each estimated MDCT coefficient, whether the MDCT coefficient is associated with a tone-like spectral interval or a noise-like spectral interval based on metadata associated with the packet, wherein the metadata is received in a bitstream comprising the packet sequence and the metadata.

[0026] As used herein, "metadata" refers to bitstream parameters used to control audio decoder processing.

[0027] The metadata may be sent in packets of a sequence of packets as well as outside packets in a bitstream comprising the sequence of packets and the metadata.

[0028] The metadata that can be used to determine whether the MDCT coefficient is associated with the tone shape spectrum interval or associated with the noise shape spectrum interval is the metadata that is used to control a certain audio decoder to process based on the audio content type. An example of this metadata is the metadata relevant to the compression and expansion tool used in AC-4. In certain embodiments, the compression and expansion tool can be turned off for the tone signal, so if the compression and expansion is closed, the signal is assumed to be a tone. As another example, if the longest MDCT is used, the audio content is most likely a tone signal.

[0029] According to some embodiments, the estimated MDCT coefficients are selected to be equal to corresponding MDCT coefficients of a received packet that immediately precedes the erroneous packet in the sequence of packets.

[0030] According to some embodiments, the estimated MDCT coefficients are selected to be equal to the corresponding MDCT coefficients of the received packet immediately preceding the erroneous packet in the packet sequence, energy-adjusted by the energy scaling factor at the scaling factor band resolution. For a detailed description of the scaling factor band resolution, refer to ETSI TS 103 190 V1.1.1 "Digital Audio Compression (AC-4) Standard", 2014-04, the contents of which are incorporated herein by reference.

[0031] According to some embodiments, the received packet includes N / 2 MDCT coefficients associated with N windowed time domain samples of the audio signal, further including: generating an intermediate frame including N windowed time domain aliasing samples from a masked frame through IMDCT; and modifying the windowed time domain aliasing samples of the intermediate frame based on the symmetric relationship between the windowed time domain aliasing samples of the intermediate frame.

[0032] As used herein, "N" is an even number.

[0033] As used herein, an "intermediate frame comprising N windowed time-domain aliased samples" refers to a sample frame obtained by IMDCT of MDCT coefficients received from an encoder in a decoder system. In some exemplary embodiments, an intermediate frame is a frame before overlap-addition is performed in a decoding system to generate a decoded frame in a sequence of decoded frames.

[0034] According to some embodiments, the modification uses a symmetrical relationship between the first half of the first half of the intermediate frame including N windowed time-domain aliasing samples and the second half of the first half of the intermediate frame including N windowed time-domain aliasing samples, and a symmetrical relationship between the first half of the second half of the intermediate frame including N windowed time-domain aliasing samples and the second half of the second half of the intermediate frame including N windowed time-domain aliasing samples.

[0035] As used herein, the "first half of an intermediate frame" means the first N / 2 samples of the intermediate frame. If the samples of the intermediate frame are numbered consecutively 0 to N-1, the first half will be samples 0 to N / 2-1. In addition, the "second half of an intermediate frame" means the last N / 2 samples of the intermediate frame. If the samples of the intermediate frame are numbered consecutively 0 to N-1, the second half will be samples 0 to N / 2 to N-1.

[0036] As used herein, “the first half of the first half of the intermediate frame” means a subset including the first N / 4 samples of the first half of the intermediate frame, the “second half of the first half of the intermediate frame” means a subset including the last N / 4 samples of the first half of the intermediate frame, the “first half of the second half of the intermediate frame” means a subset including the first N / 4 samples of the second half of the intermediate frame, and the “second half of the second half of the intermediate frame” means a subset including the last N / 4 samples of the second half of the intermediate frame.

[0037] According to some embodiments, the received packet includes N / 2 MDCT coefficients associated with N windowed time domain samples of the audio signal, further including: generating an intermediate frame including N windowed time domain aliasing samples from a masked frame by IMDCT; modifying the windowed time domain aliasing samples of the intermediate frame based on the relationship between the windowed time domain aliasing samples of the intermediate frame and the windowed time domain samples of the N time domain samples of the audio signal.

[0038] Exemplary embodiments provide that the previous decoded frame associated with the received packet immediately preceding the erroneous packet in the packet sequence can be used as an approximation of the relationship between the windowed time domain samples in the windowed time domain aliasing samples of the first subset and the N windowed time domain samples of the audio signal. This relationship can then be used to modify the generated intermediate frame in order to enhance error concealment performance.

[0039] According to an exemplary embodiment, a decoding system for concealing errors in a data packet to be decoded in an audio decoder based on a modified discrete cosine transform (MDCT), the MDCT-based audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames, the system comprising: a receiver part configured to receive a packet including a set of MDCT coefficients from an MDCT-based audio encoder arranged to encode an audio signal, the set of MDCT coefficients being associated with a frame including time domain samples of the audio signal; an error detection part configured to identify the received packet as an erroneous packet when the received packet includes one or more errors; and an error concealment part configured to: generate an estimated MDCT coefficient set for replacing the MDCT coefficient set of the erroneous packet , the estimated MDCT coefficients are based on corresponding MDCT coefficients associated with a received packet that immediately precedes the erroneous packet in the packet sequence; assigning signs of a first subset of MDCT coefficients in the estimated MDCT coefficients to be equal to corresponding signs of corresponding MDCT coefficients of a received packet that immediately precedes the erroneous packet in the packet sequence, wherein the first subset includes MDCT coefficients associated with a tone-like spectral interval of the packet; randomly assigning signs of a second subset of MDCT coefficients in the estimated MDCT coefficients, wherein the second subset includes MDCT coefficients associated with a noise-like spectral interval of the packet; generating a masked packet based on the estimated MDCT coefficients of the packet and the selected signs; and replacing the erroneous packet with the masked packet.

[0040] II. Overview - Second Aspect

[0041] According to a second aspect, exemplary embodiments propose a decoding method, a decoding system and a computer program product for decoding. The proposed method, decoding system and computer program product may generally have the same features and advantages.

[0042] According to an exemplary embodiment, a method for concealing errors in a data packet to be decoded in an MDCT-based audio decoder is provided, the audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames. The method comprises: receiving a packet including N / 2 MDCT coefficients associated with N windowed time domain samples of an audio signal from an MDCT-based audio encoder arranged to encode an audio signal; and identifying the packet as an erroneous packet when the packet includes one or more errors. The method further comprises: estimating a first subset, the first subset comprising N / 4 windowed time domain aliased samples of the first half of an intermediate frame, the intermediate frame comprising N windowed time domain aliased samples associated with the erroneous packet, the estimation being based on the relationship between the windowed time domain aliased samples of the first subset and the windowed time domain samples of the N windowed time domain samples of the audio signal; and estimating a second subset, the second subset comprising the remaining N / 4 windowed time domain aliased samples of the first half of the intermediate frame, the estimation being based on the symmetric relationship between the windowed time domain aliased samples of the second subset and the windowed time domain aliased samples of the first subset.

[0043] As used herein, "N" is an even number.

[0044] As used herein, an "erroneous packet" means comprising MDCT coefficients that differ in some way relative to the MDCT coefficients of the correct MDCT of a correct sample of the audio signal. This may mean that part of the entire packet is lost in the packet sequence, or part of the entire packet includes distortion.

[0045] As used herein, an "intermediate frame comprising N windowed time-domain aliased samples" refers to a sample frame obtained by inverse IMDCT of the MDCT coefficients received from the encoder in the decoder system. Therefore, an intermediate frame is a frame of windowed time-domain aliased samples before overlap-addition is performed in the decoding system to generate a decoded frame in a sequence of decoded frames.

[0046] As used herein, the "first half of an intermediate frame" means the first N / 2 samples of the intermediate frame. If the samples of the intermediate frame are numbered consecutively 0 to N-1, the first half will be samples 0 to N / 2-1.

[0047] As used herein, "a first subset comprising N / 4 windowed time-domain aliased samples" means a subset comprising N / 4 samples of the first half of the intermediate frame, which N / 4 samples do not need to be consecutive samples in the first half of the intermediate frame, but should be selected so as not to generate redundant information about the symmetrical relationship between the samples from the second subset and the samples of the first subset.

[0048] As used herein, “estimating a first subset” and “estimating a second subset” involve assigning the following values ​​to the windowed time-domain aliased samples of the first subset and the second subset: these values ​​are not necessarily the best approximation of the values ​​they would have if there were not yet any errors in the error packet, but achieve the desired error concealment performance so that unwanted distortion of the decoded audio signal is avoided or reduced.

[0049] According to an exemplary embodiment, the estimate of the first subset is based on a previous decoded frame associated with a received packet that immediately precedes the erroneous packet in the sequence of packets.

[0050] Note that basing the estimate on a previous decoded frame associated with a received packet immediately preceding the erroneous packet in the sequence of packets does not exclude that the estimate may additionally be based on an earlier decoded frame associated with a received packet earlier than the packet immediately preceding the erroneous packet in the sequence of packets.

[0051] In an exemplary embodiment, the estimate of the first subset based on the previous decoded frame can be combined with a first subset including N / 4 windowed time-domain aliasing samples (these N / 4 windowed time-domain aliasing samples are the first half of the first half of the intermediate frame), where, for n equals 0, 1, …, N / 4-1, the sample number n of the first subset is estimated as a windowed version of the sample number n of the previous decoded frame minus a windowed version of the sample number N / 2-1-n of the previous decoded frame.

[0052] Exemplary embodiments provide that the relationship between the windowed time-domain aliased samples of the first subset and the windowed time-domain samples in the N windowed time-domain samples of the audio signal can be reformulated by using the overlapping properties of the N windowed time-domain samples associated with the erroneous packet and the previous N windowed time-domain samples associated with the received packet immediately preceding the erroneous packet in the packet sequence. Therefore, the relationship between the windowed time-domain aliased samples of the first subset and the previous N windowed time-domain samples of the audio signal is derived. Exemplary embodiments also provide that the windowed time-domain samples in the previous N windowed time-domain samples of the audio signal can be approximated by a windowed version of the sample of the previous decoded frame.

[0053] Estimating a first subset based on a previous decoded frame, generating an estimated decoded frame, estimating a third subset, and estimating a fourth subset may, in an exemplary embodiment, be combined with a first subset including N / 4 windowed time-domain aliased samples (which N / 4 windowed time-domain aliased samples are the first half of the first half of an intermediate frame) and a third subset including N / 4 windowed time-domain aliased samples (which N / 4 windowed time-domain aliased samples are the first half of the second half of an intermediate frame), and wherein, for n equal to 0, 1, …, N / 4−1, the sample number n of the first subset is estimated as a windowed version of the sample number n of the previous decoded frame minus a windowed version of the sample number N / 2−1−n of the previous decoded frame, and wherein, for n equal to 0, 1, …, N / 4−1, the sample number n of the third subset is estimated as a windowed version of the sample number n of the estimated decoded frame plus a windowed version of the sample number N / 2−1−n of the estimated decoded frame.

[0054] Note that making the estimate based on an estimated decoded frame associated with a misgrouped packet does not exclude that the estimate may additionally be based on an earlier decoded frame associated with a received packet earlier in the packet sequence than the misgrouped packet.

[0055] An exemplary embodiment provides that windowed time-domain samples among the previous N windowed time-domain samples of an audio signal may be approximated by windowed versions of samples of a previous decoded frame and an estimated decoded frame.

[0056] In some exemplary embodiments, the estimation of the first subset is based on an offset set that includes a penultimate decoded frame associated with a received packet immediately preceding the packet associated with the previous decoded frame in the packet sequence and N / 2 samples of the previous decoded frame associated with a received packet immediately preceding the misgrouped packet in the packet sequence, the offset set including the last k samples of the penultimate decoded frame and all samples of the previous decoded frame except for the last k samples, wherein k < N / 2. In this exemplary embodiment, k may be set based on maximizing the self-similarity of the frame to be estimated with the previous frames, and k may depend on N, for example.

[0057] Instead of using only N / 2 samples of the previous decoded frame, N−k samples of the previous decoded frame are used together with k samples from the penultimate decoded frame. More specifically, the last k samples of the penultimate decoded frame and all samples of the previous decoded frame except for the last k samples are used. This requires k < N / 2.

[0058] Estimating a first subset based on a previous decoded frame, generating an estimated decoded frame, estimating a third subset, and estimating a fourth subset can be combined with an estimate of the first subset in an exemplary embodiment, the estimate of the first subset being further based on a previous decoded frame associated with a received packet that immediately precedes a packet associated with the previous decoded frame in a packet sequence, the first subset comprising N / 4 windowed time-domain aliasing samples being the first half of a first half of an intermediate frame, the third subset comprising N / 4 windowed time-domain aliasing samples being the first half of a second half of an intermediate frame, and for n equal to 0, 1, ..., k, the sample number n of the first subset is estimated to be a windowed version of the sample number N / 2-1+nk of the previous decoded frame minus the sample number of the previous decoded frame. The sample number n of the first subset is estimated as a windowed version of sample number nk-1 of the previous decoded frame minus a windowed version of sample number N / 2-1-nk of the previous decoded frame, and for n equals k+1,…,N / 4-1, the sample number n of the third subset is estimated as a windowed version of sample number N / 2-1+nk of the previous decoded frame minus a windowed version of sample number N / 2-1-nk of the estimated decoded frame, and wherein, for n equals k+1,…,N / 4-1, the sample number n of the third subset is estimated as a windowed version of sample number nk-1 of the estimated decoded frame plus a windowed version of sample number N / 2-1-nk of the estimated decoded frame, wherein k≤N / 4-1.

[0059] In an exemplary embodiment, there is provided a decoding system for concealing errors in a data packet to be decoded in an MDCT-based audio decoder arranged to decode a sequence of packets into a sequence of decoded frames, the system comprising: a receiver portion configured to receive a packet comprising N / 2 MDCT coefficients associated with N windowed time-domain samples of an audio signal from an MDCT-based audio encoder arranged to encode an audio signal; an error detection portion configured to identify the packet as an erroneous packet when the packet comprises one or more errors; and The error concealment part is configured to: estimate a first subset, the first subset includes N / 4 windowed time-domain aliasing samples of the first half of the intermediate frame, the intermediate frame includes N windowed time-domain aliasing samples associated with the error group, and the estimation is based on the relationship between the windowed time-domain aliasing samples of the first subset and the windowed time-domain samples in the N windowed time-domain samples of the audio signal; and estimate a second subset, the second subset includes the remaining N / 4 windowed time-domain aliasing samples of the first half of the intermediate frame, and the estimation is based on the symmetric relationship between the windowed time-domain aliasing samples of the second subset and the windowed time-domain aliasing samples of the first subset.

[0060] III. Overview - The Third Aspect

[0061] According to a third aspect, exemplary embodiments propose a decoding method, a decoding system and a computer program product for decoding. The proposed method, decoding system and computer program product may generally have the same features and advantages.

[0062] In some exemplary embodiments, a method for concealing errors in data packets to be decoded in an MDCT-based audio decoder is provided, the MDCT-based audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames. The method comprises: receiving a packet comprising N / 2 MDCT coefficients associated with N windowed time domain samples of an audio signal from an MDCT-based audio encoder arranged to encode an audio signal; and identifying the packet as an erroneous packet when the packet comprises one or more errors. The method also comprises estimating a decoded frame comprising the N / 2 samples associated with the erroneous packet to be equal to the second half of a previous intermediate frame comprising N non-windowed time domain samples associated with a received packet immediately preceding the erroneous packet in the sequence of packets.

[0063] As used herein, "N" is an even number.

[0064] As used herein, an "erroneous packet" means comprising MDCT coefficients that differ in some way relative to the MDCT coefficients of the correct MDCT of a correct sample of the audio signal. This may mean that part of the entire packet is missing in the packet sequence, or that part of the entire packet includes distortion.

[0065] As used herein, "estimating a decoded frame" involves assigning values ​​to samples of a decoded frame that are not necessarily the best approximation of the values ​​that these samples would have had if there were not already any errors in the error packet, but achieve the desired error concealment performance so that unwanted distortion of the decoded audio signal is avoided or reduced.

[0066] As used herein, the "second half of the previous intermediate frame" means the last N / 2 samples of the previous intermediate frame. If the samples of the intermediate frame are numbered consecutively 0 to N-1, the second half will be samples N / 2 to N-1.

[0067] In some exemplary embodiments, there is provided a method of estimating a subsequent decoded frame including N / 2 samples associated with a received packet that immediately follows an erroneous packet in a sequence of packets to be equal to a first half of a subsequent intermediate frame including non-windowed time domain samples associated with a received packet that immediately follows the erroneous packet in the sequence of packets.

[0068] In some exemplary embodiments, a decoding system for concealing errors in data packets to be decoded in an MDCT-based audio decoder is provided, the MDCT-based audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames, the system comprising: a receiver portion configured to receive a packet comprising N / 2 MDCT coefficients associated with N windowed time domain samples of an audio signal from an MDCT-based audio encoder arranged to encode an audio signal; an error detection portion configured to identify the packet as an erroneous packet when the packet comprises one or more errors; and an error concealment portion configured to estimate a decoded frame comprising N / 2 samples associated with the erroneous packet to be equal to the second half of a previous intermediate frame comprising non-windowed time domain samples associated with a received packet immediately preceding the erroneous packet in the sequence of packets.

[0069] In some exemplary embodiments, the method further comprises determining available complexity resources, and determining a method suitable for concealing the error based on the available complexity resources.

[0070] IV. Exemplary Embodiments

[0071] Figure 1A and Figure 1B The MDCT and inverse transform that the exemplary embodiment can be implemented with are described by way of example. In an audio encoding / decoding system, the audio signal is typically sampled at the encoder end and divided into a sequence of frames 101-105, wherein each frame in the sequence corresponds to a respective time interval t-2, t-1, t, t+1, t+2. Frames 101-105 each include N / 2 samples, wherein N may be 2048, 1920, 1536, etc., depending on the encoder type and the selected time-frequency resolution. Instead of applying the MDCT to frames 101-105, the MDCT is applied to a combination of two adjacent frames. Therefore, the MDCT uses overlap and is an example of a so-called overlapped transform. From a sequence of frames 101-105 (each frame comprising N / 2 time domain samples of the audio signal), the frames are combined two by two overlappingly in a consecutive order, so that, for example, the first frame 101 and the second frame 102 of the sequence of frames 101-105 are combined into a first combined frame 110, the second frame 102 and the third frame 103 are combined into a second combined frame 111, and so on, which means that the first combined frame 110 and the second combined frame 111 have an overlap because they both include the second frame 102. In order to smooth the transition between sequential frames, a window function w[n] (n=0, ..., N-1) is applied to each combination of two frames of the frame sequence to produce a combined frame 110-113 of N windowed time domain samples. As Figure 1AAs depicted, a first frame 101 and a second frame 102 corresponding to time intervals t-2 and t-1, respectively, are combined, and a windowing function is applied to the combination to generate a windowed time domain sample comprising N windowed time domain samples. The first combined frame 110, the second frame 102 and the third frame 103 corresponding to the time intervals t-1 and t are combined, and the windowing function is applied to the combination to generate a windowed time domain sample comprising N The third frame 103 and the fourth frame 104 corresponding to the time intervals t and t+1 are combined, and the windowing function is applied to the combination to generate a second combined frame 111 including N windowed time domain samples. The third combined frame 112 of FIG. 1 is composed of the fourth frame 104 and the fifth frame 105 corresponding to the time intervals t+1 and t+2, and the windowing function is applied to the combination to generate a windowed time domain sample comprising N The fourth combined frame 113.

[0072] The MDCT is then applied to the combined frames 110-113, resulting in a sequence of packets 120-123, each packet comprising N / 2 MDCT coefficients. Figure 1A As depicted, the MDCT is applied to the first combined frame 110 to produce a matrix comprising N / 2 MDCT coefficients. The MDCT is applied to the first group 120 of the second combined frame 111 to produce a second group 120 comprising N / 2 MDCT coefficients. The MDCT is applied to the third combined frame 112 to generate a second group 121 comprising N / 2 MDCT coefficients. The MDCT is applied to the fourth combined frame 113 to generate a third group 122 comprising N / 2 MDCT coefficients. The fourth group 123.

[0073] At the decoder end, the IMDCT is applied to the packets 120-123, each of which includes N / 2 MDCT coefficients, to produce intermediate frames 130-133 including N time-domain aliased samples. Figure 1B As depicted, the IMDCT is applied to the first group 120 to generate a signal including N time-domain aliased samples. The first intermediate frame 130, IMDCT is applied to the second group 121 to generate a time domain aliased sample comprising N The IMDCT is applied to the third group 122 to generate a second intermediate frame 131 including N time-domain aliased samples. The third intermediate frame 132, IMDCT is applied to the fourth group 123 to generate a time domain aliased sample comprising N The fourth intermediate frame 133.

[0074] To generate decoded frames 150-152 of decoded samples, overlap-add operations 140-142 are performed on the intermediate frames 130-133 taking into account the window function w[n]. Figure 1B As depicted, a first overlap-add operation 140 is performed between the first half of the second intermediate frame 131 and the second half of the first intermediate frame 130 to generate a first decoded frame 150 including N / 2 decoded samples corresponding to the time interval t-1, a second overlap-add operation 141 is performed between the first half of the third intermediate frame 132 and the second half of the second intermediate frame 131 to generate a second decoded frame 151 including N / 2 decoded samples corresponding to the time interval t, and a third overlap-add operation 142 is performed between the first half of the fourth intermediate frame 133 and the second half of the third intermediate frame 132 to generate a third decoded frame 152 including N / 2 decoded samples corresponding to the time interval t+1.

[0075] Errors may occur in packets including MDCT coefficients, or packets or parts of packets may be lost. Unless the errors are corrected or the lost packets are reconstructed, such errors or losses may affect the decoded frames in such a way that the decoded audio signal is impaired, such that information is lost or unwanted artifacts appear in the decoded audio signal. For example and with reference to Figure 1B , if an error is detected in the third packet 122 at the decoder end, the third intermediate frame 132 will generally be affected by the erroneous third packet 122. In this document, the packet including the error will be referred to as the erroneous packet, and the intermediate frame corresponding to the same time interval as the erroneous packet will be referred to as the intermediate frame associated with the erroneous packet or the intermediate frame including N time-domain aliased samples associated with the erroneous packet. In addition, when the third intermediate frame 132 is used in the overlap-add operation 141 to generate the second decoded frame 151, the second decoded frame 151 will generally be affected by the erroneous packet. In this document, the decoded frame corresponding to the same time interval as the erroneous packet will be referred to as the decoded frame associated with the erroneous packet. In addition, when the second intermediate frame 132 is also used in the overlap-add operation 142 to generate the third decoded frame 152, the third decoded frame 152 will also generally be affected by the erroneous packet.

[0076] Due to the overlapping nature of the combined frame, the relationship between the first N / 2 samples of the combined frame associated with time interval t and the last N / 2 samples of the combined frame associated with time interval t-1 can be derived according to Equation 1:

[0077]

[0078] Furthermore, a decoded frame is generated by using overlap-add between the first half of the intermediate frame and the second half of the previous intermediate frame. Thus, a decoded frame associated with time interval t is generated according to the following equation:

[0079]

[0080] The special properties between the windowed time domain samples of the intermediate frames can be used to estimate the intermediate frames affected by the error packets. More specifically, it can be proved that each intermediate frame has odd symmetry and even symmetry between the windowed time domain samples in the first half and the second half. For the time interval t, the following relationship can be proved:

[0081]

[0082] Furthermore, it can be shown that the windowed time-domain aliased samples can be explicitly derived for the original windowed samples of the audio signal according to the following equation (see V. Britanak et al., "Fast computational structures for an efficient implementation of the complete TDAC analysis / synthesis MDCT / MDST filter banks", Signal Processing, Vol. 89, No. 7 (July 2009), pp. 1379-1394, the contents of which are incorporated herein by reference):

[0083]

[0084] Using equation (1) in equation (4), the following relationship is derived:

[0085]

[0086] In another approximation, the decoded frame affected by the error packet can be calculated using the non-windowed time-domain aliased signal according to the following equation Frame estimation:

[0087]

[0088]

[0089] In equations (6) and (7), the notation a→b indicates that the variable b is assigned the value a.

[0090] Figure 2 By way of example there is depicted a generalized block diagram of a first decoding system 200. The decoding system 200 is arranged to conceal errors in data packets to be decoded in an MDCT based audio decoder arranged to decode a sequence of packets into a sequence of decoded frames.

[0091] The system comprises a receiver part 201 configured to receive a sequence of packets, wherein each packet comprises a set of MDCT coefficients associated with a frame comprising time domain samples of an audio signal. The sequence of packets is typically as described with respect to Figure 1A is generated as described by applying the MDCT to a combined frame of N windowed time-domain samples. Each packet of the sequence of packets comprises N / 2 MDCT coefficients.

[0092] The decoding system 200 also includes an error detection section (not shown) configured to identify whether the received packet is an error packet because the received packet includes one or more errors. The manner of detecting errors in the error detection section is arbitrary, and the position of the error detection section is also arbitrary, as long as an error packet requiring error concealment is detected and the detected error packet can be identified in the error concealment of the decoding system 200.

[0093] The decoding system 200 also includes an error concealment section 202, which is configured to estimate the MDCT coefficients of the error packet, assign symbols to the estimated MDCT coefficients, generate concealed packets, and replace the error packets with the concealed packets in the packet sequence. The concealed packets are generated as estimated MDCT coefficients with corresponding selected symbols of the error packet.

[0094] The decoding system 200 further comprises an IMDCT part 203 for applying an IMDCT to each packet of the sequence of packets including a concealed packet replacing an erroneous packet in the sequence of packets. The output of the IMDCT part 203 is a sequence of N intermediate frames of windowed time-domain aliased samples.

[0095] The decoding system 200 further comprises an overlap-add part 204 for performing an overlap-add operation between overlapping parts of consecutive intermediate frames in the intermediate frame sequence so as to generate a decoded frame of N / 2 samples.

[0096] In one embodiment, the estimated MDCT coefficients are based on the corresponding MDCT coefficients associated with the received packet immediately preceding the erroneous packet in the packet sequence. In other embodiments, the estimated MDCT coefficients are selected to be equal to the corresponding MDCT coefficients of the received packet immediately preceding the erroneous packet in the packet sequence. In addition, the signs of the first subset of the MDCT coefficients in the estimated MDCT coefficients are designated to be equal to the corresponding signs of the corresponding MDCT coefficients of the received packet immediately preceding the erroneous packet in the packet sequence. The first subset includes those MDCT coefficients associated with the tone-shaped spectrum interval of the packet. The signs of the second subset of the MDCT coefficients in the estimated MDCT coefficients are arbitrarily designated. The second subset includes those MDCT coefficients associated with the noise-shaped spectrum interval of the packet. Together with the signs of each MDCT coefficient, the error concealment portion 202 continuously receives the MDCT coefficients of each packet of the packet sequence from the receiving portion 201. The error concealment portion 202 also receives the identification of the error frame from the receiving portion. When an error frame is received, the error concealment section 202 can extract the MDCT coefficients and corresponding symbols of the previous packet received before the error packet in the packet sequence, generate the estimated MDCT coefficients of the error packet, and use the MDCT coefficients and symbols from the previous packet together to specify the symbols. When the coefficients and symbols have been estimated and specified, a concealed packet based on the estimated MDCT coefficients of the estimated packet and the selected symbol is generated, the error concealment section replaces the error packet with the concealed packet in the receiving section 201, and the concealed packet is forwarded from the receiving section 201 to the MDCT section 203.

[0097] It should be noted that when the estimated MDCT coefficients are mentioned with respect to the estimation and the assignment of a sign to each estimated MDCT coefficient, this implicitly refers to the absolute value of the estimated MDCT coefficients. Even if the sign assignment for the MDCT coefficients is disclosed to be first for the first subset and second for the second subset, the assignment of the signs may also be performed in the opposite order. Therefore, in an exemplary embodiment, the assignment may be performed first for the second subset and finally for the first subset. In fact, the MDCT coefficients may be assigned in any order. In an exemplary embodiment, it may not necessarily be necessary to continuously assign all MDCT coefficients associated with the pitch-shaped spectral interval and continuously assign all MDCT coefficients associated with the noise-shaped spectral interval. For example, one or more of the MDCT coefficients associated with the first subset may be assigned first, then one or more of the MDCT coefficients associated with the second subset may be assigned, then one or more of the MDCT coefficients associated with the first subset may be assigned, and so on. In addition, a grouping may not necessarily have MDCT coefficients associated with both the noise-shaped spectral interval and the pitch-shaped spectral interval. Instead, a grouping may have all MDCT coefficients associated with noise-like spectral intervals or all MDCT coefficients associated with tone-like spectral intervals, such that one of the first and second subsets is empty.Finally, MDCT coefficients are typically identified as either belonging to the first or second subset.

[0098] Estimating the signs of the MDCT coefficients based on content type may provide improved results with respect to error concealment performance compared to estimates using only randomly assigned estimates or estimates based only on the signs of the MDCT coefficients of previously received packets in a sequence of packets. MDCT coefficients associated with noise-like spectral intervals may be sufficiently accurate if estimated by random assignment, while MDCT coefficients associated with tone-like spectral intervals may provide improved results with respect to error concealment performance by assigning corresponding MDCT coefficients based on received packets that immediately precede the erroneous packet in the sequence of packets. Furthermore, when estimating the MDCT coefficients based on corresponding MDCT coefficients associated with received packets that immediately precede the erroneous packet in the sequence of packets, error concealment may be achieved using only data from previously received packets.

[0099] In some prior art, more complex methods including estimating the signs of all MDCT coefficients without using random assignments have been used. In other prior art, additional metadata has been provided for estimating the signs (which adds more complexity to the method) and requires changing the data stream from the encoder to the decoder. In addition, such metadata must be transmitted in packets following the erroneous packets, which causes a time delay in the symbol estimation to be performed in the decoding system.

[0100] By choosing the estimated MDCT coefficients to be equal to the corresponding MDCT coefficients of the previous packet, the complexity can be kept low, while if this is combined with the estimation of the sign of the MDCT coefficients based on the content type according to an exemplary embodiment, a concealed packet providing the desired error concealment performance can be achieved.

[0101] In other embodiments, the MDCT coefficients of a previous packet are energy adjusted at a scale factor band resolution by an energy scaling factor before the MDCT coefficients of the previous packet are selected as estimates of the MDCT coefficients of the erroneous packet.

[0102] By selecting the estimated MDCT coefficients to be equal to the corresponding MDCT coefficients of the previous packet energy-adjusted by the energy scaling factor with the scale factor band resolution, the error concealment performance achieved by concealing the packet may be enhanced with only a slight increase in complexity.

[0103] There are several alternative ways to determine whether the MDCT coefficients of a packet (e.g., an erroneous packet) in a packet sequence are associated with a tone-like spectral interval or a noise-like spectral interval. In one example, the determination is based on approximate spectral peak detection of the power spectrum associated with the erroneous packet, wherein the approximated power spectrum is based on the power spectrum associated with the received packet immediately preceding the erroneous packet in the packet sequence. In another example, the MDCT subband spectral flatness metric is used. If the value of the MDCT subband spectral flatness is above a certain threshold, the subband spectrum is flat, which implies that it is noisy. Otherwise, the spectrum is peaky, which implies that it is tonal. The MDCT subband flatness is estimated as the ratio between the geometric mean and the arithmetic mean of the amplitudes of the MDCT coefficients. It represents the deviation of the power spectrum of the signal from a flat shape. The metric is calculated band by band, wherein the term "band" is related to the set of MDCT coefficients, and the width of these bands is based on the perceptually relevant scaling factor band resolution. For a description of spectral flatness metrics, see N. Jayant and P. Noll, "Digital Coding of Waveforms, Principles and Applications to Speech and Video" in Englewood Cliffs, NJ: Prentice-Hall (1984). In other examples, the determination is based on metadata received in a packet or in a bitstream comprising a sequence of packets and metadata. The metadata to be used may, for example, be metadata for controlling processing by a certain audio decoder based on the type of audio content. For example, in AC-4, there is a compression and expansion tool that must be turned off for tonal signals. Therefore, if metadata is received indicating that compression and expansion is turned off, the signal can be assumed to be tonal. In addition, if, for example, the longest MDCT is used, the audio content is very likely to be a tonal signal.

[0104] In one embodiment, the symmetric relationship between the windowed time-domain aliased samples of the intermediate frame associated with the error frame of equation (3) is used to modify the windowed time-domain aliased samples of the intermediate frame associated with the error frame. When the error frame associated with the time interval t has been identified, a masked packet is generated in error concealment section 202, and the masked packet replaces the error frame. In IMDCT section 203, IMDCT is applied to the masked packet, which produces the intermediate frame associated with the error packet. The intermediate frame associated with the error packet generated is forwarded to error concealment section 202 from IMDCT section 203. Error concealment section 202 then modifies the windowed time-domain aliased samples of the intermediate frame generated so that the relationship of equation (3) is better satisfied.

[0105] The symmetric relationship that can be demonstrated between the windowed time-domain aliased samples of the intermediate frames can be used to modify the windowed time-domain aliased samples of the intermediate frames to enhance the error concealment performance. Thus, an enhancement in error concealment performance can be achieved with only a slight increase in complexity.

[0106] In other embodiments, the relationship between the windowed time-domain aliased samples of the intermediate frame associated with the error frame and the original data samples of equation (5) is used to modify the windowed time-domain aliased samples of the intermediate frame associated with the error frame. When the error frame associated with the time interval t has been identified, a masked packet is generated in the error concealment section 202, and the masked packet replaces the error frame. In the IMDCT section 203, the IMDCT is applied to the masked packet, which generates the intermediate frame associated with the error packet. The generated intermediate frame associated with the error packet is forwarded from the IMDCT section 203 to the error concealment section 202. The error concealment section 202 then modifies the windowed time-domain aliased samples of the generated intermediate frame so that the relationship of equation (5) is better satisfied. For example, the right hand side of the first relationship of equation (5) associated with the first half of the intermediate frame associated with the error packet is approximated with the past decoded frame associated with the time interval t-1 received from the overlap-add section 204 in the error estimation section 202. The result is an alternative estimate of the first half of the intermediate frame associated with the error packet, which alternative estimate can be used to modify the first half of the intermediate frame associated with the error packet generated by applying the IMDCT to the masked packet generated in the masking portion 202. In addition, the right hand side of the second relationship of equation (5) related to the second half of the intermediate frame associated with the error packet is approximated with the decoded frame associated with the time interval t (i.e., the decoded frame based on the modified first half of the intermediate frame associated with the error packet). The decoded frame associated with the time interval t is received from the overlap-add portion 204 in the error estimation portion 202. The result is an alternative estimate of the second half of the intermediate frame associated with the error packet, which alternative estimate can be used to modify the second half of the intermediate frame associated with the error packet generated by applying the IMDCT to the masked packet generated in the masking portion 202.

[0107] Figure 3 By way of example there is depicted a generalized block diagram of a second decoding system 300. The decoding system 300 is arranged to mask in data packets to be decoded in an MDCT based audio decoder arranged to decode a sequence of packets into a sequence of decoded frames.

[0108] The system comprises a receiver part 301 configured to receive a sequence of packets, wherein each packet comprises a set of MDCT coefficients associated with a frame comprising time domain samples of an audio signal. The sequence of packets is typically as described with respect to Figure 1A is generated as described by applying the MDCT to a combined frame of N windowed time-domain samples. Each packet of the sequence of packets comprises N / 2 MDCT coefficients.

[0109] The decoding system 300 also includes an error detection section (not shown) configured to identify whether the received packet is an error packet because the received packet includes one or more errors. The manner of detecting errors in the error detection section is arbitrary, and the position of the error detection section is also arbitrary, as long as an error packet requiring error concealment is detected and the detected error packet can be identified in the error concealment of the decoding system 200.

[0110] The decoding system 300 further includes an error concealment portion 302 configured to estimate windowed time-domain aliased samples of an intermediate frame including N windowed time-domain aliased samples associated with the erroneous packet.

[0111] The decoding system 300 further comprises an IMDCT part 303 for applying an IMDCT to each packet of the sequence of packets. The output of the IMDCT part 303 is a sequence of N intermediate frames of windowed time-domain aliased samples.

[0112] The error concealment portion 302 is further configured to replace the intermediate frame including the N windowed time-domain aliased samples associated with the erroneous packet with the estimated intermediate frame.

[0113] The decoding system 300 further comprises an overlap-add part 304 for performing an overlap-add operation between overlapping parts of consecutive intermediate frames in the intermediate frame sequence so as to generate a decoded frame of N / 2 samples.

[0114] In an embodiment, when an erroneous packet is identified within a time interval t, an intermediate frame associated with the erroneous packet can be estimated. The estimation is performed using the relationship between the windowed time-domain aliased samples of the intermediate frame associated with the time interval t and the original windowed samples of the audio signal of equation (5) and the symmetric relationship of equation (3). A first subset of the first N / 4 windowed time-domain aliased samples of the first half of the intermediate frame is estimated, and the intermediate frame includes N windowed time-domain aliased samples associated with the erroneous packet (i.e., associated with the time interval t). The estimation is performed by the first relationship of equation (5), wherein the right-hand side samples are approximated with samples of the previous decoded frame, wherein the previous decoded frame is associated with the time interval t-1. A decoded frame associated with the time interval t-1 is received from the overlap-add portion 304 in the error estimation portion 302. More specifically, for n=0, 1, ..., N / 4-1, the sample number n of the first subset is estimated as a windowed version of the sample number n of the previous decoded frame minus a windowed version of the sample number N / 2-1-n of the previous decoded frame. By the symmetric relationship of equation (3), a second subset is estimated, the second subset including the remaining (i.e., the last) N / 4 windowed time-domain aliased samples of the first half of the intermediate frame. An estimated decoded frame associated with the erroneous packet (i.e., associated with the time interval t) is generated in the overlap-add portion 304 by adding the first half of the estimated intermediate frame to the second half of the previous intermediate frame associated with the received packet immediately preceding the erroneous packet in the packet sequence (i.e., associated with the time interval t-1).

[0115] By estimating the second subset using the symmetric relationship between the windowed time-domain aliased samples of the second subset and the windowed time-domain aliased samples of the first subset, a reduction in the complexity of the estimation may be achieved while maintaining the achieved error concealment performance.

[0116] By using a previous decoded frame as an approximation of the relationship between windowed time-domain aliased samples of the first subset and the windowed time-domain samples of the N windowed time-domain samples of the audio signal to produce an estimate of the first subset, low complexity of the estimation can be achieved while achieving the desired error concealment performance.

[0117] A third subset is estimated, the third subset including the first N / 4 windowed time-domain aliased samples of the second half of the intermediate frame associated with the erroneous packet. The estimation is performed by the second relationship of equation (5), wherein the right-hand side samples are approximated by samples of an estimated decoded frame, wherein the estimated decoded frame is associated with the erroneous packet, i.e., associated with the time interval t. The estimated decoded frame associated with the time interval t is received from the overlap-add portion 304 in the error estimation portion 302. More specifically, for n=0, 1, ..., N / 4-1, the sample number n of the third subset is estimated as a windowed version of the sample number n of the estimated decoded frame plus a windowed version of the sample number N / 2-1-n of the estimated decoded frame. The fourth subset is estimated by the symmetric relationship of equation (3), the fourth subset including the remaining (i.e., the last) N / 4 windowed time-domain aliased samples of the second half of the intermediate frame. Note that for n=0, 1, ..., N / 4-1, the sample number n of the third subset is the sample number N / 2+n of the intermediate frame, because the third subset is the first half of the second half of the intermediate frame. The next estimated decoded frame associated with the received packet immediately following the erroneous packet (i.e., associated with the time interval t+1) is generated in the overlap-add portion 304 by adding the second half of the estimated intermediate frame associated with the time interval t to the first half of the next estimated intermediate frame.

[0118] In an alternative embodiment, the estimation of the first subset is based on an offset set that includes the penultimate decoded frame (not shown) associated with time interval t-2 and N / 2 samples of the previous decoded frame associated with time interval t-1, and the estimation of the third subset is based on an offset set that includes the previous decoded frame associated with time interval t-1 and N / 2 samples of the estimated decoded frame associated with time interval t. The offset set includes the last k samples of the penultimate decoded frame and all samples of the previous decoded frame except the last k samples, where k < N / 2. More specifically, for k ≤ N / 4-1, for n = 0, 1, …, k, the sample number n of the first subset is estimated as the windowed version of the sample number N / 2-1+n-k of the penultimate decoded frame (not shown) minus the windowed version of the sample number N / 2-1+n-k of the previous decoded frame. For n equal to k+1, …, N / 4-1, the sample number n of the first subset is estimated as the windowed version of the sample number n-k-1 of the previous decoded frame minus the windowed version of the sample number N / 2-1-n-k of the previous decoded frame. For n = 0, 1, …, k, the sample number n of the third subset is estimated as the windowed version of the sample number N / 2-1+n-k of the previous decoded frame minus the windowed version of the sample number N / 2-1-n-k of the estimated decoded frame. For n = k+1, …, N / 4-1, the sample number n of the third subset is estimated as the windowed version of the sample number n-k-1 of the estimated decoded frame plus the windowed version of the sample number N / 2-1-n-k of the estimated decoded frame.

[0119] The value of k can be calculated to maximize the self-similarity of the frame to be estimated using the previous frames, or it can be pre-calculated to save complexity. Additionally, k typically depends on N.

[0120] Regarding that only the windowed versions of the samples of the previous decoded frame are used to estimate the windowed time-domain aliased samples of the first subset, the error concealment performance can be improved. More specifically, the enhanced error concealment performance can be caused by using the offset of several samples or the offset of time in the estimation of the windowed time-domain aliased samples of the first subset.

[0121] Figure 4 A general block diagram of a third decoding system 400 is depicted by way of example. The decoding system 400 is arranged to conceal errors in data packets to be decoded in an MDCT-based audio decoder, which is arranged to decode a sequence of packets into a sequence of decoded frames.

[0122] The system comprises a receiver part 401 configured to receive a sequence of packets, wherein each packet comprises a set of MDCT coefficients associated with a frame comprising time domain samples of an audio signal. The sequence of packets is typically as described with respect to Figure 1A is generated as described by applying the MDCT to a combined frame of N windowed time-domain samples. Each packet of the sequence of packets comprises N / 2 MDCT coefficients.

[0123] The decoding system 400 also includes an error detection section (not shown) configured to identify whether the received packet is an error packet because the received packet includes one or more errors. The manner of detecting errors in the error detection section is arbitrary, and the position of the error detection section is also arbitrary, as long as an error packet requiring error concealment is detected and the detected error packet can be identified in the error concealment of the decoding system 400.

[0124] The decoding system 400 also includes an error concealment portion 402, which is configured to estimate a decoded frame including N / 2 samples associated with the erroneous packet to produce an estimated decoded frame. The decoded frame is estimated to be equal to the second half of the previous intermediate frame including N non-windowed time domain samples associated with the received packet immediately preceding the erroneous packet in the packet sequence.

[0125] The decoding system 400 further comprises an IMDCT part 403 for applying an IMDCT to each packet of the sequence of packets. The output of the IMDCT part 403 is a sequence of N intermediate frames of windowed time-domain aliased samples.

[0126] The decoding system 400 further comprises an overlap-add part 404 for performing an overlap-add operation between overlapping parts of consecutive intermediate frames in the intermediate frame sequence so as to generate a decoded frame of N / 2 samples.

[0127] The error concealment part 402 is also configured to estimate the next decoded frame including N / 2 samples associated with the received packet immediately following the error packet in the packet sequence to be equal to the first half of the next intermediate frame including the non-windowed time domain samples associated with the received packet immediately following the error packet in the packet sequence. The error concealment part 402 is also configured to replace the decoded frame associated with the error packet from the overlap-add part 404 with the estimated decoded packet and replace the next decoded frame associated with the error packet from the overlap-add part 404 with the estimated decoded packet.

[0128] Decoding system 400 uses approximations of equations (6) and (7).

[0129] Estimating samples of a decoded frame of samples associated with an erroneous packet using non-windowed time domain samples of a previous intermediate frame may provide a low complexity method for providing error concealment.

[0130] Furthermore, an adaptive method of determining available complexity resources may be provided, for example the method continuously determines the complexity level allowed for error concealment.For example, when an error packet is identified, the available complexity resources are determined, and a method for error concealment is selected based on the determined available resources.

[0131] V. Equivalence, Extension, Substitution and Others

[0132] After studying the above description, other embodiments of the present disclosure will become clear to those skilled in the art. Even though the present description and the accompanying drawings disclose embodiments and examples, the present disclosure is not limited to these specific examples. Many modifications and variations may be made without departing from the scope of the present disclosure as defined by the appended claims. Any reference symbols appearing in the claims are not to be construed as limiting their scope.

[0133] In addition, by studying the drawings, the present disclosure and the appended claims, the skilled person can understand and implement variations of the disclosed embodiments when implementing the present disclosure. In the claims, the word "comprising" does not exclude any elements or steps, and the indefinite article "a" or "an" does not exclude a plurality. The fact that only certain measures are recited in mutually different dependent claims does not indicate that a combination of these measures cannot be used to advantage.

[0134] The apparatus and method disclosed above can be implemented as software, firmware, hardware or a combination thereof. In hardware implementation, the task division between the functional units mentioned in the above description does not necessarily correspond to the division into physical units; on the contrary, a physical component can have multiple functions, and a task can be performed by several physical components in cooperation. Some components or all components can be implemented as software executed by a digital signal processor or a microprocessor, or can be implemented as hardware or a dedicated integrated circuit. Such software can be distributed on a computer-readable medium, which can include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). The software can be distributed on specially programmed devices, which can generally be referred to as "modules" in this article. The software component part of the module can be written in any computer language, and can be part of a large code base, or can be developed in more discrete code parts, such as usually in an object-oriented computer language. In addition, the module can be distributed on multiple computer platforms, servers, terminals, mobile devices, etc. A given module can even be implemented so that the described functions are performed by separate processors and / or computer hardware platforms. As is well known to those skilled in the art, the term computer storage media includes volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media includes but is not limited to RAM, ROM, EEPROM, flash memory or other memory technology, CD-ROM, digital versatile disk (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, magnetic disk storage or other magnetic storage devices, or any other medium that can be used to store the desired information and can be accessed by a computer. As used in this application, the term "portion" refers to all of the following: (a) a circuit implementation of hardware only (such as an implementation in analog and / or digital circuits only); (b) a combination of circuits and software (and / or firmware), such as (if applicable): (i) a combination of processor(s) or (ii) a portion of processor(s) / software (including digital signal processor(s) that work together to enable a device (such as a mobile phone or server) to perform various functions), software, and memory(s); and (c) circuits that require software or firmware for operation, such as microprocessor(s) or a portion of microprocessor(s), even if the software or firmware is not physically present. In addition, it is well known to technicians that communication media typically contains computer-readable instructions, data structures, program modules, or other data in a modular data signal (such as a carrier wave or other transmission medium), and includes any information delivery media.

Claims

1. A method for concealing errors in data packets to be decoded in an audio decoder based on a Modified Discrete Cosine Transform (MDCT), the MDCT-based audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames, the method comprising: receiving, from an MDCT-based audio encoder arranged to encode an audio signal, a N windowed time domain samples are associated with N / Grouping of 2 MDCT coefficients; identifying the packet as an error packet when the packet includes one or more errors; Estimate a first subset, the first subset comprising a first half of the intermediate frame N / 4 windowed time-domain aliased samples, the intermediate frame includes the N windowed time-domain aliased samples, the estimated first subset is based on the windowed time-domain aliased samples of the first subset and the audio signal N The relationship between the windowed time domain samples in the windowed time domain samples; Estimate a second subset, the second subset comprising the remainder of the first half of the intermediate frame N / 4 windowed time-domain aliased samples, wherein the estimated second subset is based on a symmetric relationship between the windowed time-domain aliased samples of the second subset and the windowed time-domain aliased samples of the first subset; as well as An estimated decoded frame associated with the erroneous packet is generated by adding a first half of the intermediate frame to a second half of a previous intermediate frame associated with a received packet that immediately precedes the erroneous packet in the sequence of packets.

2. The method of claim 1, wherein the estimate of the first subset is based on a previous decoded frame associated with a received packet that immediately precedes the erroneous packet in the sequence of packets.

3. The method according to claim 2, further comprising: estimating a third subset comprising a second half of the intermediate frame associated with the erroneous packet N / 4 windowed time-domain aliased samples, the estimated third subset being based on an estimated decoded frame associated with the erroneous packet; as well as Estimate a fourth subset, the fourth subset comprising the remainder of the second half of the intermediate frame N / 4 windowed time-domain aliasing samples, and the estimated fourth subset is based on a symmetric relationship between the windowed time-domain aliasing samples of the fourth subset and the estimated windowed time-domain aliasing samples of the third subset.

4. The method according to claim 3, further comprising: A subsequent estimated decoded frame associated with a received packet immediately following the erroneous packet in the packet sequence is generated by adding the second half of the intermediate frame to the first half of a subsequent intermediate frame, wherein the subsequent intermediate frame is associated with a received packet immediately following the erroneous packet in the packet sequence.

5. The method according to claim 2, wherein: include N / 4 windowed time-domain aliased samples is the first half of the first half of the intermediate frame, and wherein, for n is equal to 0, 1, ..., N / 4-1, the sample number of the first subset n Estimated to be the sample number of the previous decoded frame n The windowed version of , minus the sample number of the previous decoded frame N / 2-1- n A windowed version of .

6. The method according to claim 4, wherein: include N The first subset of the 4 windowed time-domain aliased samples is the first half of the first half of the intermediate frame, including N / 4 windowed time-domain aliased samples is the first half of the second half of the intermediate frame, and wherein, for n is equal to 0, 1, ..., N / 4-1, the sample number of the first subset n Estimated to be the sample number of the previous decoded frame n The windowed version of , minus the sample number of the previous decoded frame N / 2-1- n is a windowed version of , and where n is equal to 0, 1, ..., N / 4-1, the sample number of the third subset n The sample number of the estimated decoded frame is estimated n The windowed version of the estimated decoded frame plus the sample number N / 2-1- n A windowed version of .

7. The method according to claim 1, wherein: The estimation of the first subset is based on a set of offsets including a previous decoded frame associated with a received packet that immediately precedes an erroneous packet in the sequence of packets and a further previous decoded frame associated with a received packet that immediately precedes a packet associated with the previous decoded frame in the sequence of packets. N / 2 samples, the offset set includes the last sample of the previous decoded frame k samples and the last sample of the previous decoded frame except the last k All samples except samples, where k < N / 2.

8. The method according to claim 7, wherein: Set based on maximizing the self-similarity of the frame to be estimated with the previous frame k .

9. The method according to claim 7 or 8, wherein: k depending on N .

10. The method according to claim 3, in, the estimation of the first subset is further based on a further previous decoded frame associated with a received packet that immediately precedes the packet associated with the previous decoded frame in the sequence of packets, Among them, N The first subset of the 4 windowed time-domain aliased samples is the first half of the first half of the intermediate frame, including N / a third subset of 4 windowed time-domain aliased samples is the first half of the second half of the intermediate frame, Among them, for n is equal to 0, 1, ..., k , the sample number of the first subset n is estimated to be the sample number of the immediately previous decoded frame N / 2-1+ n - k The windowed version of the previous decoded frame minus the sample number of the previous decoded frame N / 2-1- n - k is a windowed version of n equal k+ 1,… N / 4-1, N / 4-1, the sample number of the first subset n is estimated to be the sample number of the previous decoded frame n - k -1 minus the sample number of the previous decoded frame N / 2-1- n - k A windowed version of Among them, for n is equal to 0, 1, ..., k , the sample number of the third subset n is estimated to be the sample number of the previous decoded frame N / 2-1+ n - k The sample number of the estimated decoded frame is subtracted from the windowed version of N / 2-1- n - k is a windowed version of , and where n equal k+ 1,…, N / 4-1, sample number of the third subset n The sample number of the estimated decoded frame is estimated n - k A windowed version of -1 plus the sample number of the estimated decoded frame N / 2-1- n - k A windowed version of , where k ≤ N / 4-1.

11. A decoding system for concealing errors in data packets to be decoded in an audio decoder based on a Modified Discrete Cosine Transform (MDCT), the MDCT-based audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames, the system comprising: A receiver portion configured to receive from an MDCT-based audio encoder arranged to encode an audio signal a signal including a N windowed time domain samples are associated with N / Grouping of 2 MDCT coefficients; an error detection section configured to identify the packet as an error packet when the packet includes one or more errors; The error concealment part is configured as: Estimate a first subset, the first subset comprising a first half of the intermediate frame N / 4 windowed time-domain aliased samples, the intermediate frame includes the N The estimated first subset is based on the windowed time-domain aliased samples of the first subset and the N The relationship between the windowed time domain samples in the windowed time domain samples; and Estimate a second subset, the second subset comprising the remainder of the first half of the intermediate frame N / 4 windowed time-domain aliased samples, the estimating the second subset is based on a symmetric relationship between the windowed time-domain aliased samples of the second subset and the windowed time-domain aliased samples of the first subset, and An overlap-and-add portion is configured to generate an estimated decoded frame associated with the erroneous packet by adding a first half of the intermediate frame to a second half of a previous intermediate frame associated with a received packet immediately preceding the erroneous packet in the sequence of packets.

12. A computer program product comprising a computer readable medium having instructions for performing the method of any of claims 1-10.

13. An apparatus for concealing errors in data packets to be decoded in an audio decoder based on a modified discrete cosine transform (MDCT), the MDCT-based audio decoder being arranged to decode a sequence of packets into a sequence of decoded frames, the apparatus comprising: one or more processors, One or more non-transitory computer-readable storage media storing instructions, which, when executed by the one or more processors, cause the apparatus to perform the method according to any one of claims 1-10.

14. A non-transitory computer-readable storage medium storing instructions, which, when executed by one or more processors, enable implementation of the method according to any one of claims 1-10.