Audio decoder, method and computer program using zero-input response for obtaining smooth transition

Abstract

To provide an audio decoder having an improved characteristic for switching between different encoding modes.SOLUTION: An audio decoder includes: a linear prediction area decoder for providing first decoded audio information based on an audio frame encoded in a linear prediction area; a frequency area decoder for providing second decoded audio information based on the audio frame encoded in a frequency area; and a transition processing device for obtaining a zero-input response of linear prediction filtering. An initial state of liner prediction filtering is defined according to first decoded audio information and second decoded audio information, and second decoded audio information is corrected in dependence on the zero-input response.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] 1.Technical Field An embodiment according to the present invention relates to an audio decoder for providing decoded audio information based on encoded audio information.

[0002] Another embodiment of the present invention relates to a method for providing decoded audio information based on encoded audio information.

[0003] Another embodiment of the invention relates to a computer program for carrying out the method described above.

[0004] Generally, embodiments according to the present invention relate to handling the transition from a CELP codec to an MDCT-based codec in switched audio coding. [Background technology]

[0005] 2. Background of the invention In recent years, there has been an increasing demand for transmitting and storing coded audio information. There has also been an increasing demand for audio encoding and decoding of audio signals comprising both speech and general audio (e.g., music, background noise, etc.).

[0006] In order to improve the coding quality and also to improve the bitrate efficiency, switched (or switching) audio codecs have been introduced to switch between different coding schemes, for example, such that a first frame is coded using a first coding concept (for example, a coding concept based on CELP) and a subsequent second audio frame is coded using a different second coding concept (for example, a coding concept based on MDCT). In other words, these can be switches between coding in the linear predictive coding domain (for example, using a coding concept based on CELP) and coding in the frequency domain (for example, coding based on a time domain-frequency domain or a frequency domain-time domain transform, such as an FFT transform, an inverse FFT transform, an MDCT transform or an inverse MDCT transform). For example, the first coding concept can be a coding concept based on CELP, an ACELP, a coding concept based on transform-coded excitation linear predictive domain, etc. The second coding concept can be, for example, a coding concept based on FFT, an MDCT, an AAC, or a coding concept that can be considered as a successor concept of the coding concept based on AAC.

[0007] In the following, some examples of conventional audio coders (encoders and / or decoders) will be described.

[0008] Switched audio codecs, such as MPEG USAC, are based on two main audio coding schemes. One coding scheme, e.g. the CELP codec, is intended for speech signals. The other coding scheme, e.g. a codec based on MDCT (hereafter simply called MDCT), is intended for all other audio signals (e.g. music, background noise). For mixed content signals (e.g. speech in music), the encoder (and consequently also the decoder) often switches between the two coding schemes. It is therefore necessary to avoid any artifacts (e.g. clicks due to discontinuities) when switching from one mode (or coding scheme) to another.

[0009] A switched audio codec may suffer from problems caused by, for example, the CELP-MDCT conversion.

[0010] The CELP-MDCT transform generally introduces two problems: aliasing can be introduced due to missing previous MDCT frames, and discontinuities can be introduced at the boundary between CELP and MDCT frames due to the non-perfect waveform coding nature of the two coding schemes operating at low / medium bit rates.

[0011] Several approaches already exist to solve the problems introduced by the CELP-MDCT transformation and will be discussed below.

[0012] A possible approach is described in the article "Efficient Cross-Fade Windows for Transitions between LPC-Based and Non-LPC-Based Audio Coding" by Jérémie Leconte, Philippe Goulnay, Ralph Geiger, Bruno Besete and Max Neuendorf (presented in paper 771 at the 126th AES Conference, May 2009). The article describes the approach in section 4.4.2 "ACELP-Non-LPD Mode". See also, for example, Figure 8 of the above article. Thus, the aliasing problem is first solved by increasing the MDCT length (here from 1024 to 1152) so that the MDCT left bending point is moved to the left of the boundary between the CELP and MDCT frames, by modifying the left part of the MDCT window so that the overlap is reduced, and finally by artificially introducing missing, aliasing, using CELP signals and overlap and add operations. The discontinuity problem is solved simultaneously by overlap and add operations.

[0013] This approach works well, but has the disadvantage of introducing a delay in the CELP decoder, which delay is equal to the overlap length (here 128 samples).

[0014] Another approach is described in U.S. Patent Publication US 8,725,503 B2, entitled "Elimination of Time Domain Aliasing in Transfer with Application in Weighting or Original Signal Domain," dated May 13, 2014 by Bruno Besete.

[0015] In this approach, the MDCT length (and MDCT window shape) is not changed. The aliasing problem is now solved by encoding the aliasing correction signal with a separate transform-based encoder. Additional side information bits are transmitted to the bitstream. The decoder reconstructs the aliasing correction signal and adds it for the decoded MDCT frame. In addition, the zero-input response (ZIR) of the CELP synthesis filter is used to reduce the amplitude of the aliasing correction signal and improve the coding efficiency. The ZIR also helps to significantly reduce the discontinuity problem.

[0016] Although this approach works well, the disadvantages are that it requires a significant amount of additional side information, and the number of bits required is generally variable, not appropriate for constant bit rate codecs.

[0017] Another approach is described in US Patent Publication US 2013 / 0289981 A1, entitled "Low Delay Sound Coding with Alternating Predictive and Transform Coding", dated October 31, 2013, by Stéphane Lagot, Balazs Kovesi and Pierre Berthet. According to the above mentioned approach, the MDCT is not modified, but the left part of the MDCT window is modified to reduce the overlap length. To solve the aliasing problem, the start of the MDCT frame is coded using the CELP codec, and thus the CELP signal is used to cancel the aliasing, either by completely replacing the MDCT signal or by artificially introducing the missing aliasing component (similar to the above mentioned article by Jérémie Leconte et al.). The discontinuity problem is solved by an overlap-add operation if an approach similar to that described by Jérémy Leconte et al. is used, otherwise it is solved by a simple cross-fade operation between the CELP and MDCT signals.

[0018] Like US 8,725,503 B2, this approach generally works well, but the disadvantage is that it requires a significant amount of side information introduced by the additional CELP.

[0019] In view of the above-mentioned conventional solutions, it would be desirable to have a concept with improved properties for switching between different coding modes (e.g., improved trade-off between bitrate overhead, delay and complexity). Summary of the Invention

[0020] 3. Overview of the Invention An embodiment according to the invention creates an audio decoder for providing decoded audio information based on encoded audio information. The audio decoder comprises a linear prediction domain decoder configured to provide first decoded audio information based on an audio frame encoded in a linear prediction domain and a frequency domain decoder configured to provide second decoded audio information based on an audio frame encoded in a frequency domain. The audio decoder further comprises a transition processing device. The transition processing device is configured to obtain a zero-input response of a linear predictive filtering, an initial state of the linear predictive filtering being defined depending on the first decoded audio information and the second decoded audio information. The transition processing device is further configured to modify the second decoded audio information, provided based on an audio frame encoded in the frequency domain following the audio frame encoded in the linear prediction domain depending on the zero-input response, in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0021] Provided that the initial state of the linear predictive filtering takes into account both the first decoded audio information and the second decoded audio information, this audio decoder is based on the discovery that a smooth transition between an audio frame coded in the linear predictive domain and a subsequent audio frame coded in the frequency domain can be achieved by using a zero-input response of the linear predictive filter to modify the second decoded audio information. Accordingly, the second decoded audio information can be adapted (modified) such that the start of the modified second decoded audio information is similar to the end of the first decoded audio information, helping to reduce or even avoid significant discontinuities between the first and second audio frames. Compared to the audio decoders described above, the concept is generally applicable even if the second decoded audio information does not comprise any aliasing. Moreover, it should be noted that the term "linear predictive filtering" can refer to both a single application of a linear predictive filter and multiple applications of a linear predictive filter. It should be noted that a single application of linear predictive filtering is typically equivalent to multiple applications of the same linear predictive filter. This is because linear prediction filters are typically linear.

[0022] In conclusion, the above described audio decoder makes it possible to obtain a smooth transition between a first audio frame coded in the linear prediction domain and a subsequent second audio frame coded in the frequency domain (or transform domain), without introducing delay and with a relatively small computational effort.

[0023] Another embodiment according to the invention creates an audio decoder for providing decoded audio information based on encoded audio information. The audio decoder comprises a linear prediction domain decoder configured to provide first decoded audio information based on audio frames encoded in a linear prediction domain (or equivalently, in a linear prediction domain representation). The audio decoder further comprises a frequency domain decoder configured to provide second decoded audio information based on audio frames encoded in a frequency domain (or equivalently, in a frequency domain representation). The audio decoder further comprises a transition processing device. The transition processing device is configured to obtain a first zero-input response of the linear prediction filter in response to a first initial state of the linear prediction filter defined by the first decoded audio information, and to obtain a second zero-input response of the linear prediction filter in response to a second initial state of the linear prediction filter defined by a modified version of the first decoded audio information provided with artificial aliasing (hereinafter referred to as artificial aliasing) and comprising a contribution of a portion of the second decoded audio information. Alternatively, the transition processing device is configured to obtain a joint zero-input response of the linear predictive filter in response to an initial state of the linear predictive filter defined by a combination of the first decoded audio information and a modified version of the first decoded audio information provided with artificial aliasing and comprising a contribution of a portion of the second decoded audio information. The transition processing device is further configured to modify the second decoded audio information, provided on the basis of an audio frame encoded in the frequency domain subsequent to an audio frame encoded in the linear predictive domain, in dependence on the first zero-input response and the second zero-input response or in dependence on the joint zero-input response, in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0024] This embodiment according to the invention is based on the discovery that a smooth transition between an audio frame coded in the linear prediction domain and a subsequent audio frame coded in the frequency domain (or in general in the transform domain) can be obtained by modifying the second decoded audio information based on a signal which is the zero-input response of a linear prediction filter, the initial state of which is defined by both the first decoded audio information and the second decoded audio information. The output signal of such a linear prediction filter can be used to adapt the second decoded audio information (e.g. an initial part of the second decoded audio information immediately following the transition between the first audio frame and the second audio frame) such that there is a smooth transition between the first decoded audio information (associated with the audio frame coded in the linear prediction domain) and the modified second decoded audio information (associated with the audio frame coded in the frequency domain or in the transform domain) without the need to amend the first decoded audio information.

[0025] It has been found that a zero-input response of the linear predictive filter is suitable for providing a smooth transition, since the initial state of the linear predictive filter is based on both the first decoded audio information and the second decoded audio information, and aliasing contained in the second decoded audio information is compensated for by artificial aliasing that is introduced into a modified version of the first decoded audio information.

[0026] It has also been found that no decoding delay is required by modifying the second decoded audio information based on the first zero-input response and the second zero-input response or depending on the combined zero-input response, while leaving the first decoded audio information unchanged. This is because the first zero-input response and the second zero-input response or the combined zero-input response are very well suited to smoothing the transition between an audio frame coded in the linear prediction domain and a subsequent audio frame coded in the frequency domain (or transform domain) without changing the first decoded audio information. This is because the first zero-input response and the second zero-input response or the combined zero-input response modify the second decoded audio information such that the second decoded audio information is substantially similar to the first decoded audio information, at least in the transition between an audio frame coded in the linear prediction domain and a subsequent audio frame coded in the frequency domain.

[0027] In conclusion, the above-described embodiments according to the present invention make it possible to provide a smooth transition between an audio frame coded in the linear predictive coding domain and a subsequent audio frame coded in the frequency domain (or transform domain). Since only the second decoded audio information (associated with the subsequent audio frame coded in the frequency domain) is modified, the introduction of additional delay is avoided. Also, a high quality of the transition (without substantial artifacts) can be achieved by using a first zero-input response and a second zero-input response or a joint zero-input response resulting from the consideration of both the first decoded audio information and the second audio information.

[0028] In a preferred embodiment, the frequency domain decoder is configured to perform an inverse lapped transform, so that the second decoded audio information comprises aliasing. It has been found that the above inventive concept works particularly well even in cases where the frequency domain decoder (or the transform domain decoder) introduces aliasing. It has been found that the above aliasing can be cancelled with moderate effort and good results by providing artificial aliasing in a modified version of the first decoded audio information.

[0029] In a preferred embodiment, the frequency domain decoder is configured to perform an inverse lapped transform such that the second decoded audio information comprises aliasing in a time portion that overlaps in time with the time portion for which the linear prediction domain decoder provides the first decoded audio information, and such that the second decoded audio information is free of aliasing for a time portion that follows the time portion for which the linear prediction domain decoder provides the first decoded audio information. This embodiment according to the invention is based on the idea that since no first decoded audio information is provided and there is no aliasing, it is beneficial to use a lapped transform (or an inverse lapped transform) and a windowing that preserves time portions. It has been found that, if there is no need to provide aliasing cancellation information over time because there is no first decoded audio information provided, the first zero-input response and the second zero-input response, or the combined zero-input response, can be provided with a small computational effort. In other words, it is preferred to provide the first zero-input response and the second zero-input response, or the combined zero-input response, based on an initial state in which the initial state of aliasing is substantially cancelled (e.g. by means of artificial aliasing). As a result, the first zero-input response and the second zero-input response, or the combined zero-input response, are substantially free of aliasing, so that it is desirable to have no aliasing in the second decoded audio information for a time period following the time period for which the linear prediction domain decoder provides the first decoded audio information. In this regard, it should be noted that the first zero-input response and the second zero-input response, or the combined zero-input response, are typically provided for the aforementioned time period following the time period for which the linear prediction domain decoder provides the first decoded audio information, since the first zero-input response and the second zero-input response, or the combined zero-input response, are substantially attenuated continuations of the first decoded audio information, taking into account the artificial aliasing that compensates for the aliasing contained in the second decoded audio information and, typically, for the "overlap" time period.

[0030] In a preferred embodiment, the part of the second decoded audio information used to obtain a modified version of the first decoded audio information comprises aliasing. By allowing some aliasing in the second decoded audio information, the windowing can be kept simple and an excessive increase in the information required to code an audio frame coded in the frequency domain can be avoided. In order not to seriously deteriorate the audio quality, the aliasing contained in the part of the second decoded audio information used to obtain a modified version of the first decoded audio information can be compensated by the above-mentioned artificial aliasing.

[0031] In a preferred embodiment, the artificial aliasing used to obtain a modified version of the first decoded audio information at least partially compensates for the aliasing contained in a part of the second decoded audio information and is used to obtain a modified version of the first decoded audio information, and accordingly, a good audio quality can be obtained.

[0032] In a preferred embodiment, the transition processing device is configured to apply a first windowing to the first decoded audio information to obtain a windowed version of the first decoded audio, and to apply a second windowing to the time-mirrored version of the first decoded audio information to obtain a windowed version of the time-mirrored version of the first decoded audio information. In this case, the transition processing device may be configured to combine the windowed version of the first decoded audio information and the windowed version of the time-mirrored version of the first decoded audio information to obtain a modified version of the first decoded audio information. This embodiment according to the invention is based on the idea that several windowings should be applied to obtain a proper cancellation of aliasing in the modified version of the first decoded audio information used as input for providing a zero-input response. Accordingly, a zero-input response (e.g. a second zero-input response or a combined zero-input response) may be achieved that is very favorable for a smoothness of the transition between the audio information coded in the linear predictive coding domain and the subsequent audio frame coded in the frequency domain.

[0033] In a preferred embodiment, the transition processing device is configured to linearly combine the second decoded audio information with the first zero-input response and the second zero-input response or the combined zero-input response for a time portion during which the first decoded audio information is not provided by the linear prediction domain decoder to obtain a modified second decoded audio information. It has been found that simple linear combinations (e.g. simple addition and / or subtraction, weighted linear combinations or cross-fading linear combinations) are suitable for providing a smooth transition.

[0034] In a preferred embodiment, the transition processing device is configured not to modify the first decoded audio information by the second decoded audio information when providing the decoded audio information for an audio frame encoded in the linear prediction domain, so that the decoded audio information provided for an audio frame encoded in the linear prediction domain is provided independently from the decoded audio information provided for a subsequent audio frame encoded in the frequency domain. It has been discovered that the concept according to the invention does not require modifying the first decoded audio information based on the second decoded audio information in order to obtain a sufficiently smooth transition. Therefore, by not modifying the first decoded audio information by the second decoded audio information, delays can be avoided. This is because the first decoded audio information can be provided as a result for rendering (e.g. to a listener) even before the decoding of the second decoded audio information (associated with a subsequent audio frame encoded in the frequency domain) is completed. On the other hand, the zero-input response (the first and second zero-input responses, or the joint zero-input response) can be calculated as soon as the second decoded audio information is available. Therefore, delays can be avoided.

[0035] In a preferred embodiment, the audio decoder is configured to provide complete decoded audio information for an audio frame coded in the linear prediction domain followed by an audio frame coded in the frequency domain prior to decoding (or prior to completion of decoding) the audio frame coded in the frequency domain. This concept is possible due to the fact that the first decoded audio information is not modified based on the second decoded audio information and helps to avoid any delay.

[0036] In a preferred embodiment, the transition processing device is configured to window the first and second zero-input responses, or the combined zero-input response, before modifying the second decoded audio information depending on the windowed first zero-input response and the windowed second zero-input response, or depending on the windowed combined zero-input response. Accordingly, the transition can be made particularly smooth, and any problems that would result from a very long zero-input response can be avoided.

[0037] In a preferred embodiment, the transition processor is configured to window the first zero-input response and the second zero-input response, or the combined zero-input response, with a linear window. The use of a linear window has been found to be a simple concept, yet still results in a good hearing impression.

[0038] An embodiment according to the invention creates a method for providing decoded audio information based on encoded audio information. The method comprises performing linear prediction domain decoding to provide first decoded audio information based on audio frames encoded in the linear prediction domain. The method further comprises performing frequency domain decoding to provide second decoded audio information based on audio frames encoded in the frequency domain. The method further comprises obtaining a second zero-input response of the linear predictive filtering in response to a second initial state of the linear predictive filtering defined by a modified version of the first decoded audio information provided with artificial aliasing and comprising a contribution of a portion of the second decoded audio information, and obtaining a first zero-input response of the linear predictive filtering in response to a first initial state of the linear predictive filtering defined by the first decoded audio information. Alternatively, the method comprises obtaining a combined zero-input response of the linear predictive filtering in response to an initial state of the linear predictive filtering defined by a combination of the first decoded audio information and a modified version of the first decoded audio information provided with artificial aliasing and comprising a contribution of a portion of the second decoded audio information. The method further comprises a step of modifying the second decoded audio information, provided on the basis of an audio frame coded in the frequency domain subsequent to an audio frame coded in the linear prediction domain, depending on the first zero-input response and the second zero-input response or depending on the joint zero-input response, in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information. This method is based on similar considerations and brings the same benefits as the audio decoder described above.

[0039] Another embodiment according to the invention provides a computer program which performs the above-mentioned method when the computer program runs on a computer.

[0040] Another embodiment according to the invention creates a method for providing decoded audio information based on encoded audio information. The method comprises a step of providing first decoded audio information based on an audio frame encoded in a linear predictive domain. The method further comprises a step of providing second decoded audio information based on an audio frame encoded in a frequency domain. The method further comprises a step of obtaining a zero-input response of the linear predictive filtering. An initial state of the linear predictive filtering is defined depending on the first decoded audio information and the second decoded audio information. The method further comprises a step of modifying the second decoded audio information, provided based on an audio frame encoded in a frequency domain following the audio frame encoded in a linear predictive domain, depending on the zero-input response in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0041] This method is based on similar considerations as the audio decoder described above.

[0042] Another embodiment of the present invention comprises a computer program for carrying out the method described above.

[0043] 4. Brief description of the drawings Embodiments of the present invention will be described hereinafter with reference to the accompanying drawings, in which: [Brief description of the drawings]

[0044] [Figure 1] FIG. 1 shows a schematic block diagram of an audio decoder according to an embodiment of the present invention. [Diagram 2] FIG. 2 shows a schematic block diagram of an audio decoder according to another embodiment of the present invention. [Diagram 3] FIG. 3 shows a schematic block diagram of an audio encoder according to another embodiment of the present invention. [Figure 4A]FIG. 4A shows a schematic diagram of a window at the transition from an MDCT-coded audio frame to another MDCT-coded audio frame. [Figure 4B] FIG. 4B shows a schematic diagram of a window used for the transition from a CELP coded audio frame to an MDCT coded audio frame. [Diagram 5] 5(A), (B) and (C) show graphical representations of an audio signal in a conventional audio decoder. [Figure 6] 6(A), (B), (C) and (D) show graphical representations of an audio signal in a conventional audio decoder. [Figure 7A] FIG. 7A shows a graphical representation of an audio signal obtained based on a previous CELP frame and on a first zero-input response. [Figure 7B] FIG. 7B shows a graphical representation of an audio signal, which is a second version of the previous CELP frame and of the second zero-input response. [Figure 7C] FIG. 7C shows a graphical representation of the audio signal obtained when the second zero-input response is subtracted from the audio signal of the current MDCT frame. [Figure 8A] FIG. 8A shows a graphical representation of an audio signal obtained based on a previous CELP frame. [Figure 8B] FIG. 8B shows a graphical representation of an audio signal, which is obtained as a second version of the current MDCT frame. [Figure 8C] FIG. 8C shows a graphical representation of an audio signal that is a combination of an audio signal obtained based on a previous CELP frame and on an audio signal that is a second version of the MDCT frame. [Figure 9] FIG. 9 shows a flowchart of a method for providing decoded audio information according to an embodiment of the present invention. [Figure 10] FIG. 10 shows a flowchart of a method for providing decoded audio information according to another embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0045] 5. Detailed Description of the Embodiments 5.1. Audio decoder according to FIG. 1 1 shows a schematic block diagram of an audio decoder 100 according to an embodiment of the present invention. The audio decoder 100 is configured to receive encoded audio information 110, which may comprise, for example, a first frame encoded in the linear prediction domain and a subsequent second frame encoded in the frequency domain. The audio decoder 100 is also configured to provide decoded audio information 112 based on the encoded audio information 110.

[0046] The audio decoder 100 comprises a linear prediction domain decoder 120 configured to provide first decoded audio information 122 based on audio frames encoded in a linear prediction domain. The audio decoder 100 further comprises a frequency domain decoder (or a transform domain decoder 130) configured to provide second decoded audio information 132 based on audio frames encoded in a frequency domain (or in a transform domain). For example, the linear prediction domain decoder 120 may be a CELP decoder, an ACELP decoder or a similar decoder that performs linear prediction filtering based on a source signal and based on a coded representation of linear prediction filter characteristics (or filter coefficients).

[0047] The frequency domain decoder 130 may be, for example, an AAC type decoder or any decoder based on AAC type decoding. For example, the frequency domain decoder (or transform domain decoder) may receive the coded representation of the frequency domain parameters (or transform domain parameters) and provide the second decoded audio information based thereon. For example, the frequency domain decoder 130 may decode (or transform) the frequency domain coefficients, scale (or transform) the frequency domain coefficients depending on a scale factor (the scale factor may be provided for different frequency bands and may be represented in different forms), and perform a frequency domain to time domain transform (or transform domain to time domain transform), such as an inverse fast Fourier transform or an inverse modified discrete cosine transform (inverse MDCT).

[0048] The audio decoder 100 further comprises a transition processing device 140. The transition processing device 140 is configured to obtain a zero-input response of a linear predictive filtering. An initial state of the linear predictive filtering is defined depending on the first decoded audio information and the second decoded audio information. Moreover, the transition processing device 140 is configured to modify the second decoded audio information 132, which is provided on the basis of an audio frame coded in the frequency domain following an audio frame coded in the linear predictive domain, depending on the zero-input response in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0049] For example, the transition processing device 140 may comprise an initial state determination 144, which receives the first decoded audio information 122 and the second decoded audio information 132 and provides initial state information 146 based thereon. The transition processing device 140 further comprises a linear predictive filtering 148, which receives the initial state information 146 and provides a zero-input response 150 based thereon. For example, the linear predictive filtering may be performed by a linear predictive filter that is initialized based on the initial state information 146 and is provided with zero input. In response, the linear predictive filtering provides the zero-input response 150. The transition processing device 140 further comprises a modification 152, which modifies the second decoded audio information 132 depending on the zero-input response 150 to thereby obtain modified second decoded audio information 142, which constitutes output information of the transition processing device 140. The modified second decoded audio information 142 is typically concatenated to the first decoded audio information 122 to obtain the decoded audio information 112.

[0050] Regarding the functionality of the audio decoder 100, it should be considered that in that case an audio frame (first audio frame) coded in the linear prediction domain is followed by an audio frame (second audio frame) coded in the frequency domain. The first audio frame, coded in the linear prediction domain, will be decoded by the linear prediction domain decoder 120. Accordingly, the first decoded audio information 122 is associated with the first audio frame and obtained. However, the decoded audio information 122 associated with the first audio frame typically remains unaffected by any audio information decoded based on the second audio frame, coded in the frequency domain. However, the second decoded audio information 132 is provided by the frequency domain decoder 130 based on the second audio frame coded in the frequency domain.

[0051] Unfortunately, the second decoded audio information 132 associated with the second audio frame typically does not comprise a smooth transition with the first decoded audio information 122 associated with the first decoded audio information.

[0052] However, it should be noted that the second decoded audio information is provided at a time period that further overlaps with the time period associated with the first audio frame. A part of the second decoded audio information (i.e. an initial part of the second decoded audio information 132) provided at the time of the first audio frame is evaluated by the initial state determination 144. Moreover, the initial state determination 144 further evaluates at least a part of the first decoded audio information. Accordingly, the initial state determination 144 obtains initial state information 146 based on a part of the first decoded audio information (a part of which is associated with the time of the first audio frame) and based on a part of the second decoded audio information (a part of the second decoded audio information 130 is also associated with the time of the first audio frame). Accordingly, the initial state information 146 is provided depending on the first decoded information 132 and also depending on the second decoded audio information.

[0053] It should be noted that the initial state information 146 may be provided as soon as the second decoded audio information 132 (or at least an initial part thereof as required by the initial state determination 144) is available. The linear predictive filtering 148 may further be performed as soon as the initial state information 146 is available. This is because the linear predictive filtering uses filtering coefficients that are known from the decoding of the first audio frame. Accordingly, the zero-input response 150 may be provided as soon as the second decoded audio information 132 (or at least an initial part thereof as required by the initial state determination 144) is available. Moreover, the zero-input response 150 may be used to modify a part of the second decoded audio information 132 associated with the time of the second audio frame (rather than the time of the first audio frame). Accordingly, a part of the second decoded audio information, typically located at the beginning of the time associated with the second audio frame, is modified. As a result, a smooth transition is achieved between the first decoded audio information 122 (which typically ends at the end of the time associated with the first audio frame) and the modified second decoded audio information 142 (the time portion of the second decoded audio information 132 having the time associated with the first audio frame is preferably truncated and thus preferably only used to provide initial state information for linear predictive filtering). Correspondingly, the overall decoded audio information 112 cannot be provided with a delay. This is because the provision of the first decoded audio information 122 is not delayed (because the first decoded audio information 122 is independent of the second decoded audio information 132) and because the modified second decoded audio information 142 can be provided as soon as the second decoded audio information 132 is available.Accordingly, even if there is switching from an audio frame coded in the linear prediction domain (first audio frame) to an audio frame coded in the frequency domain (second audio frame), a smooth transition between different audio frames can be achieved in the decoded audio information 112.

[0054] However, it should be noted that the audio decoder 100 may be supplemented with any of the features and functionality described herein.

[0055] 5.2. Audio decoder according to FIG. 2 Fig. 2 shows a schematic block diagram of an audio decoder according to another embodiment of the present invention. The audio decoder 200 is configured to receive encoded audio information 210, for example comprising one or more frames encoded in the linear prediction domain (or equivalently, in a linear prediction domain representation) and one or more audio frames encoded in the frequency domain (or equivalently, in a transform domain, or equivalently, in a frequency domain representation, or equivalently, in a transform domain representation). The audio decoder 200 is configured to provide decoded audio information 212 based on the encoded audio information 210. The decoded audio information 212 may be, for example, a time domain representation.

[0056] The audio decoder 200 comprises a linear prediction domain decoder 220, substantially identical to the linear prediction domain decoder 120, so that the above description applies. The linear prediction domain decoder 210 therefore receives audio frames encoded in a linear prediction domain representation contained in the encoded audio information 210, and provides, based on the audio frames encoded in the linear prediction domain representation, first decoded audio information 222, typically in the form of a time domain audio representation (and typically corresponding to the first decoded audio information 122). The audio decoder 200 further comprises a frequency domain decoder 230, substantially identical to the frequency decoder 130, so that the above description applies. Correspondingly, the frequency domain decoder 230 receives audio frames encoded in a frequency domain representation (or in a transform domain representation), and provides, based thereon, second decoded audio information 232, typically in the form of a time domain representation.

[0057] The audio decoder 200 further comprises a transition processing unit 240 configured to modify the second decoded audio information 232 thereby to derive modified second decoded audio information 242 .

[0058] The transition processor 240 is configured to obtain a first zero-input response of the linear predictive filter in response to an initial state of the linear predictive filter defined by the first decoded audio information 222. The transition processor is also configured to obtain a second zero-input response of the linear predictive filter in response to a second initial state of the linear predictive filter defined by a modified version of the first decoded audio information provided with artificial aliasing and comprising a contribution of a portion of the second decoded audio information 232. For example, the transition processor 240 comprises an initial state determination 242 that receives the first decoded audio information 222 and provides first initial state information 244 based thereon. For example, the first initial state information 244 may simply reflect a portion of the first decoded audio information 222, e.g. a portion adjacent to an end of a time portion associated for the first audio frame. The transition processing device 240 may further comprise a (first) linear predictive filtering 246 configured to receive first initial state information 244 as an initial linear predictive filter state and to provide a first zero-input response 248 based on the first initial state information 244. The transition processing device 240 further comprises a correction / aliasing addition / combination 250 configured to receive the first decoded audio information 222, or at least a part thereof (e.g. a part adjacent to an end of a time portion associated with the first audio frame) and also the second decoded information 232, or at least a part thereof (e.g. a time portion of the second decoded audio information 232 that is typically located at the end of a time portion associated with the first audio frame, the second decoded audio information being provided, for example, mainly for the time portion associated with the second audio frame but also to some extent at the end of a time portion associated with the first audio frame that is coded in a linear predictive domain representation). The modification / aliasing addition / combining may, for example, modify a time portion of the first decoded audio information, add artificial aliasing based on a time portion of the first decoded audio information, and further add a time portion of the second decoded audio information, thereby obtaining the second initial state information 252.In other words, the correction / aliasing addition / combination may be part of the second initial state determination. The second initial state information determines an initial state of the second linear predictive filtering 254, which is configured to provide a second zero-input response 256 based on the second initial state information.

[0059] For example, the first linear predictive filtering and the second linear predictive filtering may use filter settings (e.g., filter coefficients) provided by the linear predictive domain decoder 220 for the first audio frame (encoded in a linear predictive domain representation). In other words, the first and second linear predictive filtering 246, 254 may perform the same linear predictive filtering that is also performed by the linear predictive domain decoder 220 to obtain the first decoded audio information 222 associated with the first audio frame. However, the initial states of the first and second linear predictive filtering 246, 254 may be set to values ​​determined by the first initial state determination 244 and by the second initial state determination 250 (with correction / aliasing addition / combination). However, the input signals of the linear predictive filters 246, 254 may be set to zero. Accordingly, a first zero-input response 248 and a second zero-input response 256 are obtained such that the first zero-input response and the second zero-input response are based on the first decoded audio information and the second decoded audio information and are formed using the same linear prediction filter used by the linear prediction domain decoder 220.

[0060] The transition processing unit 240 further comprises a modification 258 which receives the second encoded audio information 232 and modifies the second decoded audio information 232 in dependence on the first zero input response 248 and in dependence on the second zero input response 256 to thereby obtain modified second decoded audio information 242. For example, the modification 258 may add and / or subtract the first zero input response 248 to or from the second decoded audio information 232 and add and / or subtract the second zero input response 256 to or from the second decoded audio information to obtain the modified second decoded audio information 242.

[0061] For example, the first zero-input response and the second zero-input response may be provided for a time period associated with the second audio frame such that only a portion of the second decoded audio information associated with the time period of the second audio frame is modified. Moreover, values ​​of the second decoded audio information 232 associated with a time portion associated with the first audio frame may be truncated in the final provision of the modified (based on the zero-input response) second decoded audio information.

[0062] Moreover, the audio decoder 200 is preferably configured to concatenate the first decoded audio information 222 and the modified second decoded audio information 242 to thereby obtain the overall decoded audio information 212.

[0063] Regarding the functionality of the audio decoder 200, reference is made to the above description of the audio decoder 100. Moreover, additional details will be described below with reference to other figures.

[0064] 5.3. Audio decoder according to FIG. 3 3 shows a schematic block diagram of an audio decoder 300 according to an embodiment of the present invention. The audio decoder 300 is similar to the audio decoder 200 so that only the differences will be described in detail. Otherwise, reference is made to the above description given with respect to the audio decoder 200.

[0065] The audio decoder 300 is configured to receive encoded audio information 310, which may correspond to the encoded audio signal 210. Additionally, the audio decoder 300 is configured to provide decoded audio information 312, which may correspond to the decoded audio information 212.

[0066] The audio decoder 300 comprises a linear prediction domain decoder 320, which may correspond to the linear prediction domain decoder 220, and a frequency domain decoder 330, which corresponds to the frequency domain decoder 230. The linear prediction domain decoder 320 provides first decoded audio information 322, for example based on a first audio frame encoded in the linear prediction domain. Moreover, the frequency domain audio decoder 330 provides second decoded audio information 332, for example based on a second audio frame (following the first audio frame) encoded in the frequency domain (or in the transform domain). The first decoded audio information 322 may correspond to the first decoded audio information 222, and the second decoded audio information 332 may correspond to the second decoded audio information 232.

[0067] The audio decoder 300 further comprises a transition processing device 340 which may correspond in terms of its overall functionality to the transition processing device 340 and which may provide modified second decoded audio information 342 based on the second decoded audio information 332.

[0068] The transition processing device 340 is configured to obtain a combined zero-input response of the linear prediction filter in response to a (combined) initial state of the linear prediction filter defined by a combination of the first decoded audio information provided with the artificial aliasing and a modified version of the first decoded audio information comprising a contribution of a portion of the second decoded audio information. Moreover, the transition processing device is configured to modify the second decoded audio information, provided on the basis of an audio frame encoded in the frequency domain subsequent to an audio frame encoded in the linear prediction domain, in dependence on the combined zero-input response in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0069] For example, the transition processing device 340 comprises a correction / aliasing addition / combination 342, which receives the first decoded audio information 322 and the second decoded audio information 332 and provides the combined initial state information 344 based thereon. For example, the correction / aliasing addition / combination may be considered as an initial state determination. It should further be noted that the correction / aliasing addition / combination 342 may perform the functionality of the initial state determination 242 and of the initial state determination 250. The combined initial state information 344 may for example be equal to (or at least correspond to) the sum of the first initial state information 244 and the second initial state information 252. Accordingly, the correction / aliasing addition / combination 342 may for example further combine a part of the first decoded audio information 322, the artificial aliasing and a part of the second decoded audio information 332. Moreover, the modification / aliasing addition / combining 342 may further modify a portion of the first decoded audio information and / or add a windowed copy of the first decoded audio information 322, as will be described in more detail below. Accordingly, combined initial state information 344 is obtained.

[0070] The transition processing unit 340 further comprises a linear predictive filtering 346 that receives the combined initial state information 344 and provides a combined zero-input response 348 for the modification 350 based thereon. The linear predictive filtering 346 may, for example, perform linear predictive filtering that is substantially identical to the linear predictive filtering performed by the linear predictive decoder 320 to obtain the first decoded audio information 322. However, an initial state of the linear predictive filtering 346 may be determined by the combined initial state information 344. Also, an input signal for providing the combined zero-input response 348 may be set to zero such that the linear predictive filtering 344 provides a zero-input response based on the combined initial state information 344. (The filtering parameters or filtering coefficients are, for example, identical to the filtering parameters or filtering coefficients used by the linear prediction domain decoder 320 to provide the first decoded audio information 322 associated with the first audio frame. Furthermore, the combined zero-input response 348 is used to modify the second decoded audio information 332 to thereby derive the modified second decoded audio information 342. For example, the modification 350 may add the combined zero-input response 348 to the second decoded audio information 332 or subtract the combined zero-input response from the second decoded audio information.)

[0071] However, for further details reference is also made to the description of the audio decoders 100, 200 and to the detailed description below.

[0072] Discussion of transition concepts In the following, some details regarding the transition from CELP frames to MDCT frames, which is applicable in the audio decoders 100, 200, 300, will be discussed.

[0073] Also, differences compared to conventional concepts will be discussed.

[0074] MDCT and Windowing - Overview In an embodiment according to the invention, the aliasing problem is solved by increasing the MDCT length (e.g. for an audio frame coded in the MDCT domain following an audio frame coded in the linear prediction domain) such that the left bending point (e.g. of a time domain audio signal reconstructed based on a set of MDCT coefficients using an inverse MDCT transform) is moved to the left of the boundary between the CELP and MDCT frames. The left part of the MDCT window (e.g. of the window applied to the time domain audio signal reconstructed based on a set of MDCT coefficients using an inverse MDCT transform) is further modified (e.g. when compared to a "standard" MDCT window) such that overlap is reduced.

[0075] As an example, Figures 4A and 4B show graphical representations of different windows: Figure 4A shows a window for a transition from a first MDCT frame (i.e. a first audio frame coded in the frequency domain) to another MDCT frame (i.e. a second audio frame coded in the frequency domain), while Figure 4B shows a window used for a transition from a CELP frame (i.e. a first audio frame coded in the linear prediction domain) to an MDCT frame (i.e. a second audio frame coded in the frequency domain in the following).

[0076] In other words, Fig. 4A shows a sequence of audio frames that can be considered as a comparative example, while Fig. 4B shows a sequence in which a first audio frame is coded in the linear prediction domain followed by a second audio frame coded in the frequency domain. The case according to Fig. 4B is handled in a particularly advantageous manner by an embodiment of the present invention.

[0077] Referring now to FIG. 4A, it should be noted that the horizontal axis 410 describes the time in milliseconds, and the vertical axis 412 describes the amplitude of the window (e.g., the normalized amplitude of the window) in arbitrary units. As can be seen, the frame length is equal to 20 ms, so that the time period associated with the first audio frame extends between t=-20 ms and t=0. The time period associated with the second audio frame extends from time t=0 to t=20 ms. However, it can be seen that the first window for windowing the time domain audio samples provided by the inverse modified discrete cosine transform based on the decoded MDCT coefficients extends between time t=-20 ms and t=8.75 ms. Hence, the length of the first window 420 is longer than the frame length (20 ms). Accordingly, even if a time between t=-20 ms and t=0 is associated for the first audio frame, time-domain audio samples are provided based on the decoding of the first audio frame for a time between t=-20 ms and t=-8.75 ms. Hence, there is an overlap of approximately 8.75 ms between the time-domain audio samples provided based on the first encoded audio frame and the time-domain audio samples provided based on the second decoded audio frame. It should be noted that the second window is indicated at 422 and extends between times t=0 and t=28.75 ms.

[0078] Moreover, it should be noted that the windowed time-domain audio signal provided to the first audio frame and provided to the second audio frame is not aliasing free. Rather, the windowed (second) decoded audio information provided to the first audio frame comprises aliasing between times t=-20 ms and t=-11.25 ms and also between times t=0 and t=8.75 ms. Similarly, the windowed decoded audio information provided to the second audio frame comprises aliasing between times t=0 and t=8.75 ms and also between times t=20 ms and t=28.75 ms. However, for example, the aliasing contained in the decoded audio information provided to the first audio frame is cancelled with the aliasing contained in the decoded audio information provided to the subsequent second audio frame in the time portion between times t=0 and t=8.75 ms.

[0079] Moreover, for windows 420 and 422, the time duration between MDCT knee points is equal to the frame length, which is equal to 20 ms.

[0080] 4B, there will now be described a different case, i.e., a window for a transition from a CELP frame to an MDCT frame, that may be used in the audio decoders 100, 200, 300 to provide the second decoded audio information. In FIG. 4B, the horizontal axis 430 describes the time in milliseconds, and the vertical axis 432 describes the amplitude of the window in arbitrary units.

[0081] As can be seen in Fig. 4B, the first frame extends between time t1 = -20 ms and time t2 = 0 ms. Hence, the frame length of the first audio frame, which is a CELP audio frame, is 20 ms. Furthermore, the second, subsequent audio frame extends between time t2 and t3 = 20 ms. Thus, the length of the second audio frame, which is an MDCT audio frame, is also 20 ms.

[0082] In the following, some details regarding window 440 will be described.

[0083] The window 440 comprises a first window slope 442 extending between time t4=-1.25 ms and time t2=0 ms. The second window slope 444 extends between time t3=20 ms and time t5=28.75 ms. It should be noted that the modified discrete cosine transform, which provides the (second) decoded audio information for (or associated with) the second audio frame, provides time domain samples between time t4 and t5. However, the modified discrete cosine transform (or, more precisely, the inverse modified discrete cosine transform, which may be used in the frequency domain decoder 130, 230, 330 when an audio frame coded in the frequency domain, e.g., the MDCT domain, follows an audio frame coded in the linear prediction domain) provides time domain samples with aliasing between time t4 and t2 and between time t3 and t5 based on the frequency domain representation of the second audio frame. Meanwhile, the inverse modified discrete cosine transform provides alias-free time domain samples in the time period between times t2 and t3 based on the frequency domain representation of the second audio frame. Hence, the first window slope 442 is associated with the time domain audio samples with some aliasing, and the second window slope 444 is further associated with the time domain audio samples with some aliasing.

[0084] It should also be noted that the time between MDCT knee points is equal to 25 ms for the second audio frame, which implies that the number of coded MDCT coefficients should be larger in the situation shown in FIG. 4B than in the situation shown in FIG. 4A.

[0085] In conclusion, when both a first audio frame and a second audio frame following the first audio frame are coded in the frequency domain (e.g., in the MDCT domain), the audio decoder 100, 200, 300 may apply the window 420, 422 (e.g., to window the output of the inverse modified discrete cosine transform in the frequency domain decoder). On the other hand, when a second audio frame following a first audio frame coded in the linear prediction domain is coded in the frequency domain (e.g., in the MDCT domain), the audio decoder 100, 200, 300 may switch the operation of the frequency domain decoder. For example, if the second audio frame is coded in the MDCT domain and follows a previous first audio frame coded in the CELP domain, an inverse modified discrete cosine transform using an increased number of MDCT coefficients may be used (which implies that an increased number of MDCT coefficients are included in the coding format in the frequency domain representation of the audio frame following the previous audio frame coded in the linear prediction domain, when compared to the frequency domain representation of the coded audio frame following the previous audio frame which is also further coded in the frequency domain). Moreover, in the case where the second (current) audio frame coded in the frequency domain follows an audio frame coded in the linear prediction domain (when compared to the case where the second (current) audio frame follows the previous audio frame which is also further coded in the frequency domain), a different window, i.e., window 440, is applied to window the output of the inverse modified discrete cosine transform (i.e., the time domain audio representation provided by the inverse modified discrete cosine transform), to obtain the second decoded audio information 132.

[0086] As a further consequence, an inverse modified discrete cosine transform with increased length (when compared to the standard case) may be applied by the frequency domain decoder 130 in the case where an audio frame coded in the frequency domain follows an audio frame coded in the linear prediction domain. Moreover, the window 440 may be used in this case (whereas the windows 420, 422 may be used in the "standard" case where an audio frame coded in the frequency domain follows a previous audio frame coded in the frequency domain).

[0087] With regard to the inventive concept, it should be noted that the CELP signal is modified so as not to introduce any additional delay, as will be shown in more detail below. Instead, the embodiment according to the invention creates a mechanism to remove any discontinuities that may be introduced at the boundary between the CELP and MDCT frames. This mechanism uses the zero-input response of the CELP synthesis filter (e.g., used by the linear prediction domain decoder) to smooth the discontinuities. Details are given below.

[0088] Step-by-step description In the following, a short step-by-step description will be provided, after which further details will be given.

[0089] Encoder side 1. If the previous frame (also sometimes denoted as "first frame") is CELP (or, in general, coded in the linear prediction domain), then the current MDCT frame (which may be considered as an example of a frame coded in the frequency domain or in the transform domain) (also sometimes denoted as "second frame") is coded with a different MDCT length and a different MDCT window. For example, window 440 may be used in this case (rather than the "standard" window 422).

[0090] 2. The MDCT length is increased (e.g., from 20 ms to 25 ms, see Figs. 4A and 4B) so that the left bending point is moved to the left of the boundary between the CELP and MDCT frames. For example, the MDCT length (which may be defined by the number of MDCT coefficients) may be selected so that the length of (or between) the MDCT bending points is equal to 25 ms (as shown in Fig. 4B) when compared to the "standard" length between MDCT bending points of 20 ms (as shown in Fig. 4A). It may also be seen that the "left" bending point of the MDCT transform is located between times t4 and t2 (rather than halfway between times t=0 and t=8.75 ms), as may be seen in Fig. 4B. However, the position of the right MDCT bending point may not be changed (e.g., halfway between times t3 and t5), which may be seen from a comparison of Figs. 4A and 4B (or, more precisely, windows 422 and 440).

[0091] 3. The left part of the MDCT window is modified so that the overlap length is reduced (e.g., from 8.75 ms to 1.25 ms). For example, the part with aliasing is located between times t4=-1.25 ms and t2=0 (i.e., before the time period associated with the second audio frame, which starts at t=0 and ends at t=20 ms) when the previous audio frame is coded in the linear prediction domain. On the other hand, the signal part with aliasing is located between times t=0 and t=8.75 ms when the previous audio frame is coded in the frequency domain (e.g., in the MDCT domain).

[0092] Decoder side 1. If the previous frame (also indicated as the first audio frame) is CELP (or generally coded in the linear prediction domain), the current MDCT frame (also indicated as the second audio frame) (which is an example of a frame coded in the frequency domain or transform domain) is decoded with the same MDCT length and the same MDCT window used at the encoder side. In other words, the windowing shown in Fig. 4B is applied in providing the second decoded audio information, and the above-mentioned properties of the inverse modified discrete cosine transform (corresponding to the properties of the modified discrete cosine transform used at the encoder side) can be further applied.

[0093] 2. In order to remove any discontinuities that may occur at the boundary between CELP and MDCT frames (eg at the boundary between the first and second audio frames mentioned above), the following mechanism is used. a) The first part of the signal is constructed by artificially introducing a missing aliasing of an overlap part (e.g., a signal part between times t4 and t2 of a time-domain audio signal provided by an inverse modified discrete cosine transform) of an MDCT signal (e.g., using the first decoded audio information) with a CELP signal, and an overlap-and-add operation. The length of the first part of the signal is, for example, equal to the overlap length (e.g., 1.25 ms). b) The second part of the signal is constructed by subtracting the first part of the signal (e.g. the part located immediately before the frame boundary between the first and second audio frames) for a corresponding CELP signal. c) The zero-input response of the CELP synthesis filter is generated by filtering a frame of zeroes and using the second part of the signal as the memory state (or as the initial state). d) The zero-input response is windowed, for example, so that it reduces to zero after a number of samples (eg, 64). e) The windowed zero-input response is added to the beginning of the MDCT signal (eg, the audio portion starting at time t2=0).

[0094] Step by step description - a detailed description of the decoder functionality In the following, the functionality of the decoder will be described in more detail.

[0095] JPEG2025032135000002.jpg42169

[0096] Detailed description of step 1 After step 1 on the decoder side (decoding the current MDCT frame with the same MDCT length and the same MDCT window as used on the encoder side), we obtain the current decoded MDCT frame (e.g., the time-domain representation of the "second audio frame" constituting the second decoded audio information described above). This frame (e.g., the second frame) does not contain any aliasing, because the left bending point has been moved to the left of the boundary between the CELP and the MDCT frame (e.g., using the concept as described in detail with reference to FIG. 4B). That is, we can obtain perfect reconstruction (e.g., between times t2=0 and t3=20 ms) in the current frame at a sufficiently high bit rate. However, at low bit rates, the signal does not necessarily match the input signal, and therefore discontinuities may be introduced at the boundary between the CELP and the MDCT (e.g., at time t=0, as shown in FIG. 4B).

[0097] For ease of understanding, this problem will be illustrated with reference to Figure 5. The top plot (Figure 5A) shows the decoded CELP signal S C (n), and the middle plot (Fig. 5B) shows the decoded MDCT signal S M 5(n), and the bottom plot (FIG. 5C) shows the output signal obtained by truncating the windowed overlap signal and concatenating the CELP frame and the MDCT frame. There is clearly a discontinuity in the output signal at the boundary between the two frames (e.g., at time t=0 ms) (shown in FIG. 5C).

[0098] Further processing comparison examples One possible solution to this problem is the approach proposed in the above-mentioned reference 1 ("Efficient Cross-Fade Windows for Transitions Between LPC-Based and Non-LPC-Based Audio Coding" by J. Leconte et al.), which describes a concept used in MPEG USAC. In the following, a brief description of the above-mentioned reference approach will be provided.

[0099] JPEG2025032135000003.jpg25169

[0100] Hence, missing aliasing is artificially introduced in the overlap region. JPEG2025032135000004.jpg16169

[0101] Finally, a second version of the decoded CELP signal is obtained using an overlap-and-add operation. JPEG2025032135000005.jpg13169

[0102] As can be seen in Figures 6A-6D, this comparison approach removes the discontinuity (see especially Figure 6D). The problem with this approach is that it introduces an additional delay (equal to the overlap length) because the past frames are modified after the current frame is decoded. In some applications, such as low-delay audio coding, it is desirable (or even required) to have as small a delay as possible.

[0103] Detailed description of the processing steps Compared with the conventional approaches mentioned above, the approach proposed here to remove the discontinuity does not have any additional delay: it does not modify the past CELP frame (also denoted as the first audio frame), but instead modifies the current MDCT frame (also denoted as the second audio frame coded in the frequency domain that follows the first audio frame coded in the linear prediction domain).

[0104] JPEG2025032135000006.jpg18169

[0105] JPEG2025032135000007.jpg19169

[0106] Hence, missing aliasing is artificially introduced in the overlap region. JPEG2025032135000008.jpg13169

[0107] Finally, a second version of the decoded CELP signal is obtained using an overlap-and-add operation. JPEG2025032135000009.jpg8164

[0108] However, contrary to Reference 1 ("Efficient Cross-Fade Windows for Transition Between LPC-Based and Non-LPC-Based Audio Coding" by J. Leconte et al.), the past decoded ACELP signal is not replaced by this version of the past ACELP frame in order not to introduce any additional delay: it is simply used as an intermediate signal to modify the current MDCT frame, as described in the next step.

[0109] JPEG2025032135000010.jpg64170

[0110] Step b) The concept further comprises generating the two signals by calculating a zero-input response (ZIP) of a CELP synthesis filter (which may generally be considered as a linear prediction filter) using two different memories (also denoted as initial states) for the CELP synthesis filter.

[0111] JPEG2025032135000011.jpg55170

[0112] JPEG2025032135000012.jpg61170

[0113] JPEG2025032135000013.jpg54170

[0114] JPEG2025032135000014.jpg25170

[0115] JPEG2025032135000015.jpg36170

[0116] JPEG2025032135000016.jpg24170

[0117] Now referring to FIG. 7, some details will be explained.

[0118] 7A shows a graphical representation of the previous CELP frame and of the first zero-input response, where the horizontal axis 710 describes time in milliseconds and the vertical axis 712 describes amplitude in arbitrary units.

[0119] JPEG2025032135000017.jpg30170

[0120] JPEG2025032135000018.jpg38170

[0121] JPEG2025032135000019.jpg17169

[0122] JPEG2025032135000020.jpg22170

[0123] Step c) The current MDCT signal (eg, the second decoded audio information 132, 232, 332) is replaced with a second version 142, 242, 342 of the current MDCT (ie, of the MDCT signal associated with the current, second audio frame). JPEG2025032135000021.jpg13170

[0124] JPEG2025032135000022.jpg22170

[0125] JPEG2025032135000023.jpg33170

[0126] For example, FIG. 8A shows a graphical representation of a signal for a previous CELP frame (e.g., of the first decoded audio information). The horizontal axis 810 describes time in milliseconds. Additionally, the vertical axis 812 describes amplitude in arbitrary units. As can be seen, the first decoded audio information begins at time t 81 (-20ms) and t 82 (0 ms) (e.g., by linear prediction domain decoding).

[0127] JPEG2025032135000024.jpg58170

[0128] FIG. 8C shows the transition between a previous CELP frame (as shown in FIG. 8A) and a second version of the current MDCT frame (as shown in FIG. 8B). The horizontal axis 830 describes time in milliseconds, and the vertical axis 832 describes amplitude in arbitrary units of time. As can be seen, 81 and 82 (between time t 824B)。 Therefore, there is a substantially continuous transition between the first frame (coded in the linear prediction domain) and the second frame (coded in the frequency domain) of the current MDCT frame (starting at time t5 and ending at time t6, e.g., as shown in FIG. 4B). Therefore, audible distortions are avoided in the transition from the first frame (coded in the linear prediction domain) to the second frame (coded in the frequency domain).

[0129] JPEG2025032135000025.jpg34170

[0130] Step d) Optionally, a window can be applied to the two ZIRs in order not to affect the entire current MDCT frame, which can be useful, for example, to reduce complexity or if the ZIR is not close to 0 at the end of the MDCT frame.

[0131] One example of a window is a simple linear window v(n) of length P. JPEG2025032135000026.jpg17170For example, P=64.

[0132] For example, the window may process the zero-input response 150, the zero-input response 248, 256, or the combined zero-input response 348.

[0133] 5.8. Method according to FIG. 9 Fig. 9 shows a flow chart of a method for providing decoded audio information based on encoded audio information. The method 900 comprises a step 910 of providing first decoded audio information based on an audio frame encoded in a linear predictive domain. The method 900 further comprises a step 920 of providing second decoded audio information based on an audio frame encoded in a frequency domain. The method 900 further comprises a step 930 of obtaining a zero-input response of the linear predictive filtering. An initial state of the linear predictive filtering is defined depending on the first decoded audio information and the second decoded audio information.

[0134] The method 900 further comprises a step 940 of modifying the second decoded audio information, provided on the basis of an audio frame encoded in the frequency domain subsequent to an audio frame encoded in the linear prediction domain, depending on the zero-input response in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0135] The method 900 may also be supplemented with respect to an audio decoder by any of the features and functionality described herein.

[0136] 5.10. Method according to FIG. 10 FIG. 10 shows a flow chart of a method 1000 for providing decoded audio information based on encoded audio information T.

[0137] The method 1000 comprises a step 1010 of performing linear prediction domain decoding to provide first decoded audio information based on an audio frame encoded in the linear prediction domain.

[0138] The method 1000 further comprises a step 1020 of performing frequency domain decoding to provide second decoded audio information based on the audio frame encoded in the frequency domain.

[0139] The method 1000 further comprises a step 1030 of obtaining a first zero-input response of the linear predictive filtering in response to a first initial state of the linear predictive filtering defined by the first decoded audio information, and a step 1040 of obtaining a second zero-input response of the linear predictive filtering in response to a second initial state of the linear predictive filtering defined by a modified version of the first decoded audio information provided with artificial aliasing and comprising a contribution of a portion of the second decoded audio information.

[0140] Instead, the method 1000 comprises a step 1050 of obtaining a combined zero-input response of the linear predictive filtering in response to an initial state of the linear predictive filtering defined by a combination of the first decoded audio information and a modified version of the first decoded audio information provided with the artificial aliasing and comprising a contribution of a portion of the second decoded audio information.

[0141] The method 1000 further comprises a step 1060 of modifying the second decoded audio information, provided on the basis of an audio frame encoded in the frequency domain subsequent to an audio frame encoded in the linear prediction domain, depending on the first zero-input response and the second zero-input response, or depending on the joint zero-input response, in order to obtain a smooth transition between the first decoded audio information and the modified second decoded audio information.

[0142] It should be noted that the method 1000 may also be supplemented with any of the features and functionality described herein in relation to an audio decoder.

[0143] 6. Conclusion In conclusion, embodiments according to the present invention concern CELP-MDCT transforms. These transforms generally present two problems. 1. Aliasing due to missing previous MDCT frames, and 2. Discontinuities at the boundaries between CELP and MDCT frames due to the imperfect waveform coding nature of the two coding schemes operating at low / medium bit rates.

[0144] In an embodiment according to the invention, the aliasing problem is solved by increasing the MDCT length so that the left bending point is moved to the left of the boundary between the CELP and MDCT frames. The left part of the MDCT window is further altered so that the overlap is reduced. Contrary to conventional solutions, the CELP signal is not modified to introduce any additional delay. Instead, a mechanism is created to remove any discontinuities that may be introduced at the boundary between the CELP and MDCT frames. This mechanism uses the zero-input response of the CELP synthesis filter to smooth the discontinuities. Additional details are now described.

[0145] 7. Alternative Implementations Although some aspects have been described in the context of an apparatus, it will be apparent that these aspects also express a description of a corresponding method, where a block or apparatus corresponds to a method step or a feature of a method step. Analogously, aspects described in the context of a method step also express a description of a corresponding block or item or feature of a corresponding apparatus. Some or all of the method steps may be performed by (or using) a hardware apparatus, such as, for example, a microprocessor, a programmable computer, or an electronic circuit. In some embodiments, one or more of the most important method steps may be performed by such an apparatus.

[0146] The encoded audio signal according to the invention may be stored on a digital recording medium or may be transmitted over a transmission medium, such as a wireless transmission medium or a wired transmission medium, such as the Internet.

[0147] Depending on the specific implementation requirements, embodiments of the present invention may be implemented in hardware or in software. The implementation may be performed using a digital storage medium, such as a floppy disk, DVD, Blu-Ray, CD, ROM, PROM, EPROM, EEPROM or FLASH memory, having electronically readable control signals stored thereon, which cooperates (or is capable of cooperating) with a programmable computer system such that the respective methods are performed. Thus, the digital storage medium may be computer readable.

[0148] Some embodiments of the present invention comprise a data carrier having electronically readable control signals capable of cooperating with a programmable computer system to cause one of the methods described herein to be performed.

[0149] In general, embodiments of the present invention may be implemented as a computer program product with program code that operates to perform one of the methods when the computer program product runs on a computer. The program code may, for example, be stored on a machine-readable carrier.

[0150] Other embodiments comprise the computer program for performing one of the methods described herein, stored on a machine readable carrier.

[0151] In other words, one embodiment of the inventive method is, therefore, a computer program having a program code for performing one of the methods described herein, when the computer program runs on a computer.

[0152] A further embodiment of the method according to the invention is therefore a data carrier (or digital storage medium, or computer readable medium) comprising a computer program for performing one of the methods described herein, recorded thereon. The data carrier, digital storage medium or recording medium is typically tangible and / or non-transitory.

[0153] A further embodiment of the inventive method is therefore a data stream or a sequence of signals representing the computer program for performing one of the methods described herein, the data stream or the sequence of signals being for example adapted to be transmitted via a data communication connection, for example via the Internet.

[0154] A further embodiment comprises a processing means, for example a computer, or a programmable logic device, configured to or adapted to perform one of the methods described herein.

[0155] A further embodiment comprises a computer having installed thereon a computer program for performing one of the methods described herein.

[0156] Further embodiments according to the invention comprise a device or a system configured to transmit (e.g. electronically or optically) a computer program for performing one of the methods described herein for a receiver. The receiver may be, for example, a computer, a mobile device, a memory device, etc. The device or system may comprise, for example, a file server for transmitting the computer program for the receiver.

[0157] In some embodiments, a programmable logic device (e.g., a field programmable gate array) may be used to perform some or all of the functionality of the methods described herein. In some embodiments, a field programmable gate array may cooperate with a microprocessor to perform one of the methods described herein. In general, the methods are preferably performed by any hardware apparatus.

[0158] The apparatus described herein may be implemented using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.

[0159] The methods described herein may be performed using a hardware apparatus, or using a computer, or using a combination of a hardware apparatus and a computer.

[0160] The above-described embodiments are merely illustrative for the principles of the present invention. It is understood that modifications and variations of the arrangements and details described herein will be apparent to others skilled in the art. It is therefore intended to be limited only by the scope of the appended claims, and not by the specific details provided by way of description and illustration of the embodiments herein.

Claims

1. An audio decoder (100; 200; 300) for providing decoded audio information (112; 212; 312) on the basis of encoded audio information (110; 210; 310), said audio decoder comprising: A first decoded audio information (122; 222; 322; S) is generated based on an audio frame encoded in the linear prediction domain. C a linear prediction domain decoder (120; 220; 320) configured to provide a linear prediction domain decoder (120; 220; 320) A second decoded audio information (132; 232; 332; S) is generated based on the audio frames encoded in the frequency domain. M a frequency domain decoder (130; 230; 330) configured to provide a A transition processing device (140; 240; 340), the transition processing device is configured to obtain a zero-input response (150; 256; 348) of a linear predictive filtering (148; 254; 346), an initial state (146; 252; 344) of the linear predictive filtering being defined in dependence on the first decoded audio information and the second decoded audio information; and

2. 2. An audio decoder as claimed in claim 1,

3. 3. An audio decoder (100; 200; 300) according to claim 1 or 2, wherein the frequency domain decoder (130; 230; 330) is configured to perform an inverse lapped transform such that the second decoded audio information (132; 232; 332) contains aliasing.

4. 4. An audio decoder (100; 200; 300) according to claim 1, wherein the frequency domain decoder (130; 230; 330) is configured to perform an inverse lapped transform such that, for the linear prediction domain decoder (120; 220; 320) to provide a first decoded audio information (122; 222; 322), the second decoded audio information (132; 232; 332) comprises aliasing in a time portion which overlaps in time with a time portion, and such that, for the linear prediction domain decoder to provide a first decoded audio information, the second decoded audio information is free of aliasing for the time portion which follows the time portion.

5.

6.

7. An audio decoder (100; 200; 300) according to one of the preceding claims, characterized in that the transition processing device (140; 240; 340) m denotes a running variable, M is the filter length of the linear prediction filter, a m denotes the filter coefficients of the linear prediction filter, S C (n) denotes a previously decoded value of the first decoded audio information (122; 222; 322) for time index n, N indicates the processing length of the audio decoder.

8. An audio decoder (100; 200; 300) according to any one of claims 1 to 7, wherein the transition processing device (140; 240; 340) performs a first window processing ((w(-n-1)w(-n-1)) on the first decoded audio information (122; 222; 322; S C (n)) to obtain a windowed version of the first decoded audio information, and applying a second windowing (w(n+L)w(-n-1)) to the first decoded audio information (122; 222; 322; S C (n)) C (-n-L-1)) to obtain a windowed version of the time-mirrored version of the first decoded audio information; and

9. An audio decoder according to one of claims 1 to 8, The transition processing device (140; 240; 340) n indicates the time index, w(-n-1) denotes the value of the window function for time index (-n-1); w(n+L) denotes the value of the window function for time index (n+L); S c (n) denotes a previously decoded value of the first decoded audio information (122; 222; 322) for a time index (n), S C (-n-L-1) denotes a previously decoded value of the first decoded audio information for time index (-n-L-1), S M (n) indicates the decoded value of the second decoded audio information (132; 232; 332) for time index n; and L denotes the window length of the audio decoder.

10. An audio decoder (100; 200; 300) according to one of the claims 1 to 9, characterized in that the transition processing device (140; 240; 340) m denotes the execution variable, M denotes the filter length of the linear prediction filter (148; 254; 346), a m denotes the filter coefficient of the linear prediction filter, N indicates the processing length of the audio decoder.

11. 11. An audio decoder (100; 200; 300) according to one of claims 1 to 10, wherein the transition processing device (140; 240; 340) is configured for linearly combining the second decoded audio information (132; 232; 332) to the first zero-input response (248) and to the second zero-input response (256), or to the combined zero-input response (150; 348), for each time portion for which no first decoded audio information (122; 222; 322) is provided by the linear prediction domain decoder (120; 220; 320), in order to obtain the modified second decoded audio information.

12. An audio decoder (100; 200; 300) according to one of the claims 1 to 11, characterized in that the transition processing device (140; 240; 340) n indicates the time index, S M (n) indicates a value of the second decoded audio information for time index n, v(n) denotes the value of the window function, N indicates the processing length of the audio decoder.

13. 13. An audio decoder (100; 200; 300) according to one of claims 1 to 12, wherein the transition processing device (140; 240; 340) is configured for not modifying the first decoded audio information (122; 222; 322) by the second decoded audio information (132; 232; 332) when providing decoded audio information for an audio frame encoded in the linear prediction domain such that the decoded audio information provided for an audio frame encoded in the linear prediction domain is provided independently from decoded audio information provided for a subsequent audio frame encoded in the frequency domain.

14. 14. An audio decoder (100; 200; 300) according to one of claims 1 to 13, wherein the audio decoder is configured to provide complete decoded audio information (122; 222; 322) for an audio frame coded in the linear prediction domain, followed by an audio frame coded in the frequency domain, prior to decoding the audio frame coded in the frequency domain.

15. 15. An audio decoder (100; 200; 300) according to one of claims 1 to 14, wherein the transition processing device (140; 240; 340) is configured to window the first zero-input response (248) and the second zero-input response (256), or the combined zero-input response (150; 348), before modifying the second decoded audio information (132; 232; 332) in dependence on a windowed first zero-input response and a windowed second zero-input response, or in dependence on a windowed combined zero-input response.

16. 16. An audio decoder (100; 200; 300) according to claim 15, wherein the transition processor is configured to window the first zero-input response and the second zero-input response or the combined zero-input response with a linear window.

17. A method (900) for providing decoded audio information based on encoded audio information, the method comprising: A first decoded audio information (S) is generated based on an audio frame encoded in the linear prediction domain. C (n)) (910); A second decoded audio information (S) is generated based on the audio frames encoded in the frequency domain. M (n)) (920); obtaining (930) a zero-input response of a linear predictive filtering, an initial state of said linear predictive filtering being defined depending on said first decoded audio information and said second decoded audio information;

18. A computer program for carrying out the method according to claim 17 when the computer program runs on a computer.

Images