JP2009134187A - Encoder, decoder and method thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve high prediction gain, while compressing information required for transmitting an inter-channel prediction (ICP) parameter. <P>SOLUTION: A monaural signal synthesis section 101 generates a monaural signal M by using a left channel signal L and a right channel signal R. A monaural encoding section 102 encodes the monaural signal M by monaural CODEC. A local decoding section 103 obtains a monaural reproduction signal M' by decoding an encoded data of the monaural signal M. An ICP analysis section 104 performs ICP analysis of the left channel signal L and the monaural reproduction signal M', and generates a left channel ICP parameter. An ICP analysis section 105 performs the ICP analysis of a right channel signal R and the monaural reproduction signal M', and generates a right channel ICP parameter. Here, the ICP analysis sections 104 and 105 implement ICP process by a filter processing using a filter coefficient for each sample string, which is generated by grouping an input multiple samples in the past. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an encoding device, a decoding device, and a method for realizing scalable stereo encoding using inter-channel prediction.

  Conventionally, speech coding is used for communication applications that use narrowband speech in the telephone band (200 Hz to 3.4 kHz). Monaural audio narrowband audio codecs are widely used in communications applications such as mobile telephones, teleconferencing equipment and recently voice communications over packet networks (eg, the Internet).

  One of the steps toward a more realistic audio communication system is the transition from monaural audio expression to stereo audio expression. Broadband stereo voice communication provides a more natural acoustic environment. Scalable stereo speech coding is a core technology for realizing high-quality and highly useful speech communication.

  One common method of encoding stereo speech signals is by using signal prediction techniques based on monaural speech. That is, the basic channel signal is transmitted using a known monaural audio codec, and the left channel or the right channel is predicted from the basic channel signal using additional information and parameters. In many applications, a mixed monaural signal is selected as the basic channel signal.

  Conventional methods for encoding a stereo signal include intensity stereo coding, binaural cue coding, and inter-channel prediction (hereinafter abbreviated as “ICP”). These parametric stereo coding schemes have different strengths and weaknesses and are suitable for coding different source materials.

  Non-Patent Document 1 discloses a technique for predicting a stereo signal based on a monaural signal using these encoding methods. Specifically, a monaural signal is generated using channel signals constituting a stereo signal, for example, a left channel signal and a right channel signal, and the obtained monaural signal is encoded / decoded using a known audio codec. Further, a difference signal (side signal) between the left channel and the right channel is predicted from the monaural signal using a prediction parameter. In such an encoding method, the encoding side models the relationship between the monaural signal and the side signal using a time-dependent adaptive filter, and transmits the filter coefficient calculated for each frame to the decoding side. On the decoding side, the high-quality monaural signal transmitted by the monaural audio codec is filtered to regenerate the difference signal, and the left channel signal and the right channel signal are calculated from the regenerated difference signal and monaural signal.

  Also, Non-Patent Document 2 discloses an encoding method called cross-channel correlation canceller, and when applying the inter-channel correlation canceller technique in the ICP encoding method, The other channel can be predicted from one channel.

ICP uses channel-to-channel correlation that is unique between channels. FIG. 9 shows a general block diagram of the ICP. In FIG. 9, x (n) and y (n) represent an input signal and a reference signal, respectively. FIG. 10 shows the structure of the adaptive filter H (z). In this figure, H (z) is H (z) = b ₀ + b ₁ z ⁻¹ + b ₂ z ⁻² +... + B _k z ^−k and is an adaptive filter, FIR (Finite Impulse Response) filter in this figure. The model (transfer function) is shown. Using these input and reference signals, ICP parameters can be determined by minimizing the mean square error (MSE) of e (n) with respect to the filter parameters.

The adaptive filter has an adaptive filter parameter (ICP parameter) b = [b ₀ , b ₁ ,... That minimizes the mean square error (MSE) between the prediction signal and the reference signal according to the following equation (1). b _k ] is obtained and output. In equation (1), k represents the filter order, E represents a statistical expectation operator, and E {. } Indicates an ensemble average operation, and e (n) indicates a prediction error.

  In Non-Patent Document 1 and Non-Patent Document 2, the conventional FIR filter shown in FIG. 10 is used.

  Linear prediction (LP) has been widely used in various speech processing. LP analysis is used when extracting a spectrum envelope from a speech sequence. This spectral envelope is represented by LP coefficients. Conventional LP analysis predicts the nth sample x (n) of the speech signal using k coefficients from the previous k samples. At this time, the LP coefficient is obtained by minimizing the energy of the linear prediction error (also referred to as LP residual signal).

  Modern speech codecs use this LP residual signal in the encoding process. So far, over the last 20 years, variations of conventional LP analysis have been devised. In Non-Patent Document 3, the spectrum of voiced speech is better modeled using a method called linear prediction with linear extrapolation (LPLE).

In this method in which 2k samples past x (n) are grouped and extrapolated, an all-pole filter of order 2k is generated, and there are only k unknown coefficients in the normal equation. In this LPLE method, since the order of the filter is high, an all-pole model that is more accurate than the conventional LP can be obtained for the speech spectrum (particularly, the formant having the lowest frequency). Furthermore, in this LPLE, because the filter order is high, more resonances of the spectrum can be modeled. According to Non-Patent Document 3, as an interesting point, the energy in the low frequency part of the LP residual signal when using LPLE is smaller than in the case of the conventional LP analysis.
Extended AMR-Wideband (AMR-WB +) codec; Transcoding functions, 3GPP TS 26.290 V6.3.0 (2005-06), pp.44-46. S. Minami and O. Okada, “Stereophonic ADPCM voice coding method,” in Proc. ICASSP'90, Apr. 1990. Susanna Varho and Paavo Alku, “A new predictive method for all-pole modeling of speech spectra with a compressed set of parameters,” in Proc. IEEE Int. Symp. On Circuits and Systems, ISCAS '99, May 1999. AMR Wideband Speech Codec; General Description, 3GPP TS 26.171 V5.0.0 (2001-03), pp.5-7.

Consider a configuration in which monaural signals are used to individually predict left and right channel signals by ICP. The monaural signal is usually calculated by the following equation (2).

In this equation (2), L (n) represents a left channel signal and R (n) represents a right channel signal. n is a time index. An index representing the effectiveness of ICP is a prediction gain that can be defined as the following equation (3).

  In this equation (3), e (n) = y (n) −y ′ (n), y (n) represents the reference signal, and y ′ (n) represents the predicted signal. A higher prediction gain means better prediction performance.

  In the conventional ICP process, ICP performance is satisfactory when the correlation between the two channels is high. However, when the correlation is very low, more orders of adaptive filter coefficients are required to provide comparable ICP performance, and in some cases the cost to increase the prediction gain (necessary for transmission of adaptive filter information). Large amount of information, and further, the amount of calculation necessary for calculating the filter information) is too large. Therefore, it is necessary to adopt a method that can improve the performance of ICP.

  The present invention has been made in view of this point, and an object thereof is to obtain a high-quality stereo signal while achieving high prediction gain while compressing information necessary for transmitting ICP parameters.

  The encoding apparatus of the present invention groups monaural reproduction signal generating means for generating a monaural reproduction signal using the first channel signal and the second channel signal of a stereo signal, and groups a plurality of samples of the monaural reproduction signal inputted in the past Between the first channels, with the first filter coefficient generated using the signal obtained by the first filter processing using the converted first sample sequence and the first channel signal as the first inter-channel prediction parameter First inter-channel prediction analysis means for performing prediction analysis, a signal obtained by second filter processing using a second sample sequence in which a plurality of samples of the monaural reproduction signal input in the past are grouped, and the second channel A second inter-channel prediction analysis in which a second filter coefficient generated using the signal is used as a second inter-channel prediction parameter. A configuration having a, a prediction analysis unit between Yaneru.

  The decoding apparatus according to the present invention includes a monaural decoding unit that decodes encoded data of a monaural signal generated by the encoding apparatus to generate a monaural reproduction signal, and an encoding-side monaural reproduction signal that is generated by the encoding apparatus. The first inter-channel prediction parameter generated by the first inter-channel prediction analysis using the first sample stream on the encoding side obtained by grouping a plurality of samples and the first channel signal of the stereo signal on the encoding side is represented by the first filter. A first synthesis filter that can be used as a coefficient, and a first sample sequence in which a plurality of samples of the monaural reproduction signal are grouped is input by the first synthesis filter using the input inter-channel prediction parameter First inter-channel prediction synthesis means for generating a first channel reproduction signal of a stereo reproduction signal by filtering; Between the second channels using the encoding-side second sample sequence obtained by grouping a plurality of samples of the encoding-side monaural reproduction signal and the second-channel signal of the encoding-side stereo signal made in the encoding device The second inter-channel prediction parameter generated by the prediction analysis has a second synthesis filter that can be used as a second filter coefficient, and a second sample sequence in which a plurality of samples of the monaural reproduction signal are grouped is input. And a second inter-channel prediction synthesis unit that generates a second channel reproduction signal of the stereo reproduction signal by filtering with the second synthesis filter using the second inter-channel prediction parameter.

  According to the present invention, it is possible to achieve a high prediction gain and obtain a high-quality stereo signal while compressing information necessary for transmitting ICP parameters.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. Note that the present invention is performed during scalable stereo coding using ICP, and all information and parameters are generated in units of frames. In the following description, the left channel signal, the right channel signal, and the monaural signal are respectively represented as L, R, and M, and their regenerated signals are represented as L ′, R ′, and M ′, respectively.

(Embodiment 1)
FIG. 1 is a block diagram showing the configuration of the encoding apparatus according to the present embodiment. For example, a stereo signal including a left channel signal and a right channel signal in a PCM (Pulse Code Modulation) format is input to the encoding apparatus 100 illustrated in FIG. 1 for each frame. Also, the encoding apparatus 100 shown in FIG. 1 outputs a bit stream of encoded data for the monaural signal and the left and right channel ICP parameters.

  The monaural signal synthesis unit 101 generates the monaural signal M using the left channel signal L and the right channel signal R, for example, according to the above equation (2), and outputs the monaural signal M to the monaural encoding unit 102.

  The monaural encoding unit 102 encodes the monaural signal M using a monaural audio codec (for example, AMR-WB, see Non-Patent Document 4), and outputs the obtained encoded data to the local decoding unit 103 and the multiplexing unit 108. .

  The local decoding unit 103 decodes the encoded data of the monaural signal M, and outputs the obtained monaural regeneration signal M ′ to the ICP analysis units 104 and 105.

  The ICP analysis unit 104 includes the adaptive filter shown in FIG. 9, performs ICP analysis using the left channel signal L and the monaural regeneration signal M ′, generates a left channel ICP parameter, and performs ICP parameter quantization. To the unit 106. The ICP analysis unit 105 includes the adaptive filter shown in FIG. 9, performs ICP analysis using the right channel signal R and the monaural regenerated signal M ′, generates a right channel ICP parameter, and performs ICP parameter quantization. Output to the unit 107.

  The ICP parameter quantization unit 106 quantizes the left channel ICP parameter and outputs the obtained encoded data of the left channel ICP parameter to the multiplexing unit 108. The ICP parameter quantization unit 107 quantizes the right channel ICP parameter and outputs the obtained encoded data of the right channel ICP parameter to the multiplexing unit 108.

  The multiplexing unit 108 multiplexes the encoded data of the monaural signal M, the encoded data of the left channel ICP parameter, and the encoded data of the right channel ICP parameter, and outputs the obtained bit stream.

  FIG. 2 is a diagram illustrating the structure of the adaptive filter H (z) of the ICP analysis units 104 and 105. In the example of FIG. 2, the input signal x (n) is grouped by two samples to generate x ′ (n). The calculation of the adaptive filter parameter is the same as the above formula (1) except that x (n) is replaced with x ′ (n).

FIG. 3 is a diagram illustrating a method for grouping input samples. In the example shown in FIG. 3, in each of S _p (1 ≦ p ≦ k), two adjacent input samples are grouped according to the following equation (4) to generate one x ′ (n).

  As described above, the ICP analysis units 104 and 105 group a plurality of input past samples, and perform an ICP process (hereinafter referred to as “sample grouping”) by performing filter processing using a filter coefficient for a sample sequence in which the plurality of samples are grouped. ICP process with technique ”). There are a plurality of methods for grouping input samples, and the methods shown in FIGS. 2 and 3 are merely examples.

  FIG. 4 is a block diagram showing a configuration of the decoding apparatus according to the present embodiment. The bit stream transmitted from the encoding device 100 shown in FIG. 1 is received by the decoding device 400 shown in FIG.

  Separating section 401 separates the bit stream received by the decoding apparatus, the encoded data of monaural signal M to monaural decoding section 402, the encoded data of the left channel ICP parameter to ICP parameter decoding section 403, and the right channel ICP The parameter encoded data is output to ICP parameter decoding section 404, respectively.

  The monaural decoding unit 402 decodes the encoded data of the monaural signal M, and outputs the obtained monaural regeneration signal M ′ to the ICP synthesis filters 405 and 406. ICP parameter decoding section 403 decodes the encoded data of the left channel ICP parameter, and outputs the decoded left channel ICP parameter to ICP synthesis filter 405. The ICP parameter decoding unit 404 decodes the encoded data of the right channel ICP parameter and outputs the decoded right channel ICP parameter to the ICP synthesis filter 406.

The ICP synthesis filter 405 has the same structure as the adaptive filter H (z) of the ICP analysis unit 104, and filters the monaural regenerated signal M ′ using an ICP synthesis filter composed of left channel ICP parameters (formula (5) )). Thereby, the left channel regeneration signal L ′ can be obtained.

The ICP synthesis filter 406 has the same structure as the adaptive filter H (z) of the ICP analysis unit 105, and filters the monaural regenerated signal M ′ by an ICP synthesis filter configured with right channel ICP parameters (formula (6) )). Thereby, the right channel regeneration signal R ′ can be obtained.

Here, b _i ^L (n) in Expression (5) and b _i ^R (n) in Expression (6) are elements of the ICP parameters of the left channel and the right channel, respectively. Further, x ′ (n) in the above formulas (5) and (6) is calculated by formula (4) after replacing x (n) with M ′ (n).

  Thus, in this embodiment, in a stereo audio codec including an ICP process, ICP parameters are calculated by an ICP process with a sample grouping method. Thus, in the conventional ICP, the (k + 1) ICP coefficients are used to construct a k-th order analysis / synthesis filter, whereas in the ICP of the present embodiment, (k + 1) ICP coefficients are used. Higher order (eg 2k) analysis / synthesis ICP filters can be constructed. Thus, the predicted order of the ICP filter can be substantially extended, higher prediction gains can be achieved, and the information required to transmit higher order ICP parameters can be compressed. Become. Further, the prediction gain of ICP can be increased by further using the older sample having a correlation with the current sample to be predicted. Since a higher prediction gain is obtained, a high-quality stereo / monaural signal can be obtained in the decoding device, and speech with good sound quality can be obtained.

  Also, by using the ICP process with the sample grouping method of this embodiment, the order of ICP parameters required to achieve a predetermined ICP prediction gain can be reduced. In other words, ICP with a sample grouping technique can reduce the bit rate.

  In conventional ICP, the adaptive filter coefficients are determined by minimizing the mean square error between the filter output signal and the reference signal. As an evaluation of the effectiveness by adding the sample grouping method, a simulation result for comparing the conventional ICP and the ICP of this embodiment is shown below. In this example, 6th order ICP parameters are used for each channel. FIG. 5 shows the waveform of one frame of the audio signal of the left and right channels. The corresponding monaural signal is calculated according to equation (2), and the left and right channels are predicted using that signal. When conventional ICP is used, the predicted gains of the left and right channels are 19.46 dB and 9.72 dB, respectively. On the other hand, when the ICP of the present embodiment is used, the prediction gains of the left channel and the right channel are 20.48 dB and 10.74 dB, respectively. As described above, by using the ICP process with the sample grouping method of the present embodiment, the prediction gain of each channel can be increased by 1 dB, and a better prediction gain can be obtained. .

(Embodiment 2)
In the second embodiment, a case where two types of ICPs, the ICP shown in the first embodiment and the conventional ICP, are used will be described. Specifically, the encoding apparatus sends the ICP parameter having the higher prediction gain of these two ICPs and the sub information indicating the type of ICP used to the decoding apparatus, and the decoding apparatus converts the sub information into the sub information. Based on the ICP method, an ICP synthesis filter is constructed.

  FIG. 6 is a block diagram showing a configuration of the encoding apparatus according to the present embodiment. In the encoding apparatus 600 shown in FIG. 6, the same reference numerals are given to the same parts as those of the encoding apparatus 100 shown in FIG. 1, and detailed description thereof is omitted. 6 has a configuration in which the ICP analysis units 104 and 105 are deleted and ICP parameter selection units 604 and 605 are added, compared to the encoding device 100 shown in FIG.

  The monaural signal synthesis unit 101 outputs the generated monaural signal M to the monaural encoding unit 102 and the ICP parameter selection units 604 and 605. The local decoding unit 103 outputs the monaural regeneration signal M ′ to the ICP parameter selection units 604 and 605.

  The ICP parameter selection unit 604 performs the analysis by the ICP shown in the first embodiment and the analysis by the conventional ICP, calculates the prediction gain in each ICP, selects the left channel ICP parameter having the higher prediction gain, The selected left channel ICP parameter is output to ICP parameter quantization section 106, and sub information indicating the type of ICP used is output to multiplexing section 108.

  The ICP parameter selection unit 605 performs the analysis by the ICP shown in the first embodiment and the analysis by the conventional ICP, calculates the prediction gain in each ICP, selects the right channel ICP parameter having the higher prediction gain, The selected right channel ICP parameter is output to ICP parameter quantization section 107, and sub information indicating the type of ICP used is output to multiplexing section 108.

  The multiplexing unit 108 uses the encoded data of the monaural signal M, the encoded data of the left channel ICP parameter, the encoded data of the right channel ICP parameter, the sub information indicating the type of ICP used in the left channel, and the right channel. The sub information indicating the type of the ICP is multiplexed, and the obtained bit stream is output.

  FIG. 7 is a block diagram showing an internal configuration of the ICP parameter selection unit 604. Signals input to the first ICP analysis unit 701 and the second ICP analysis unit 702 are the left channel signal L and the monaural signal M.

  The first ICP analysis unit 701 calculates the first left channel ICP parameter using the conventional FIR filter shown in FIG. 10, and outputs the obtained first left channel ICP parameter to the local decoding unit 703 and the determination unit 708. To do.

  The second ICP analysis unit 702 calculates the second left channel ICP parameter by the ICP with the sample grouping method using the FIR filter as shown in FIGS. 2 and 3, and the obtained second left channel ICP parameter Is output to the local decoding unit 704 and the determination unit 708.

  The local decoding unit 703 reconstructs the first left channel signal L ′ based on the first left channel ICP parameter and the monaural regeneration signal M ′, and outputs the first left channel signal L ′ to the gain calculation unit 705. The gain calculation unit 705 uses the reconstructed first left channel signal L ′ and the original left channel signal L, and uses Equation (3) to calculate the first prediction that is the prediction gain Gain in the conventional ICP. The gain Gtr is calculated.

  The local decoding unit 704 reconstructs the second left channel signal L ′ based on the second left channel ICP parameter and the monaural regeneration signal M ′, and outputs the second left channel signal L ′ to the gain calculation unit 706. The gain calculation unit 706 uses the reconstructed second left channel signal L ′ and the original left channel signal L, and uses Equation (3) to predict the predicted gain Gain in the ICP shown in the first embodiment. The second prediction gain Gsg is calculated.

  The comparison unit 707 compares the first prediction gain Gtr and the second prediction gain Gsg, generates sub-information indicating the type of ICP having the larger prediction gain, and outputs the sub-information to the determination unit 708 and the multiplexing unit 108 .

  The determination unit 708 selects the left channel ICP parameter having the higher prediction gain based on the sub information, and outputs the left channel ICP parameter to the ICP parameter quantization unit 106.

  Note that the internal configuration of the ICP parameter selection unit 605 is the same as that of FIG. 7, which is a block diagram of the ICP parameter selection unit 604, except that the input left channel signal is replaced with the right channel signal.

  FIG. 8 is a block diagram showing a configuration of the decoding apparatus according to the present embodiment. In the decoding apparatus 800 shown in FIG. 8, the same reference numerals are given to the parts common to the decoding apparatus 400 shown in FIG. 4, and the detailed description thereof is omitted. The decoding apparatus 800 shown in FIG. 8 differs from the ICP synthesis filters 405 and 406 of the decoding apparatus 400 shown in FIG. 4 in the operation of the ICP synthesis filters 805 and 806.

  Separating section 401 separates the bit stream transmitted from encoding apparatus 600 and received by decoding apparatus 800, encodes the encoded data of monaural signal M to monaural decoding section 402, and encodes the encoded data of the left channel ICP parameter to ICP. In the parameter decoding unit 403, the encoded data of the right channel ICP parameter is stored in the ICP parameter decoding unit 404, the sub information indicating the type of ICP used in the left channel is input in the ICP synthesis filter 805, and the ICP used in the right channel. The sub information indicating the type is output to the ICP synthesis filter 806, respectively.

  The monaural decoding unit 402 decodes the encoded data of the monaural signal M, and outputs the obtained monaural regeneration signal M ′ to the ICP synthesis filters 805 and 806. ICP parameter decoding section 403 decodes the encoded data of the left channel ICP parameter and outputs the decoded left channel ICP parameter to ICP synthesis filter 805. ICP parameter decoding section 404 decodes the encoded data of the right channel ICP parameter and outputs the decoded right channel ICP parameter to ICP synthesis filter 806.

  The ICP synthesis filters 805 and 806 each have a filter having the same structure as the adaptive filter of the first ICP analysis unit 701 and a filter having the same structure as the adaptive filter of the second ICP analysis unit 702.

  The ICP synthesis filter 805 determines a filter to be used based on the sub-information, configures an ICP synthesis filter with the left channel ICP parameters, and filters the monaural signal M ′ by this filter. As a result, a composite signal L ′ for the left channel is obtained. Similarly, the ICP synthesis filter 806 determines a filter to be used based on the sub-information, configures an ICP synthesis filter with the right channel ICP parameters, and filters the monaural signal M ′ by this filter. As a result, a composite signal R ′ for the right channel is obtained.

  When ICP with a sample grouping method is selected in coding apparatus 600, x ′ (n) in the above formulas (5) and (6) is replaced with x ′ (n) by M ′ (n). Calculated by equation (4) above. When the conventional ICP is selected in the encoding apparatus 600, x ′ (n) in the above formulas (5) and (6) is equal to M ′ (n).

  As described above, in the present embodiment, the ICP parameter having the higher predicted gain is selected from the two types of ICPs of the ICP shown in the first embodiment and the conventional ICP. This results in a low correlation between the current sample of interest and the older past sample in time, but if the correlation between the current sample of interest and the newer past sample in time is high, the conventional ICP In this case, the ICP performance can be improved by using the ICP parameters by the conventional ICP.

  In the present embodiment, in FIG. 7, the monaural signal M input to the first ICP analysis unit 701 and the second ICP analysis unit 702 can be replaced with the reconstructed monaural signal M ′. Since the reconstructed monaural signal M ′ is used for the ICP synthesis filters 805 and 806 of the decoding device 800 of FIG. 8, in this case, the first ICP analysis unit 701 and the second ICP analysis unit in the encoding device 600 are used. The input to 702 matches the input to the ICP synthesis filters 805 and 806 of the decoding device 800, and the prediction performance can be further improved. In this case, the monaural signal M is not necessary as an input signal to the ICP parameter selection units 604 and 605 in FIG.

  In this embodiment, another selection method can be applied to ICP. This alternative selection method is based on calculating the correlation between the ICP input sequence and the reference sequence. When the correlation between the input signal and the reference signal is high, for example, greater than a predetermined threshold value Thr (Thr = 0.7 as an example; the threshold value for the normalized correlation value), the reference signal is input from the input signal. That is, the first ICP analysis unit 701 and the second ICP analysis unit 702 select a conventional ICP.

  On the other hand, when the correlation between the input signal and the reference signal is low, a higher order ICP filter is required to obtain a predetermined prediction performance. In such a case, the first ICP analysis unit 701 and the second ICP analysis unit 702 select an ICP with a sample grouping technique. This adaptive selection method requires a smaller amount of calculation than the method shown in the second embodiment, and can reduce calculation complexity (required calculation amount) while maintaining ICP prediction performance.

(Other embodiments)
The encoding method described in each of the above embodiments is a type (referred to as M-LR type) that predicts left and right signals using a monaural signal. In the present invention, apart from this method, an encoding method called MS type can also be adopted. In this alternative scheme, a side signal can be predicted using a monaural signal. The side signal S (n) can be calculated by the following equation (7). This process is almost the same as the M-LR type process shown in the first and second embodiments except that the left channel signal is replaced with a side signal. In the decoding apparatus, as shown in the following equations (8) and (9), the synthesized speech signal of the left and right channels can be calculated by using the reconstructed side signal and monaural signal.

  The embodiment of the present invention has been described above.

  In each of the above embodiments, decoding apparatuses 400 and 800 receive bitstreams transmitted from encoding apparatuses 100 and 600, respectively. However, different codes that can generate bitstreams having the same configuration are used. The bit stream transmitted by the encoding device may be received.

  In each of the above embodiments, the ICP analysis unit 104 and the ICP analysis unit 105 have adaptive filters having the same structure. However, in the present invention, adaptive filters having different configurations may be used. That is, a different ICP analysis may be performed. In each of the above embodiments, the method of grouping a plurality of samples of the monaural regeneration signal M ′ to be input in the ICP analysis unit 104 and the ICP analysis unit 105 is the same method. Different grouping methods may be used. The same applies to the ICP parameter selection unit 604 and the ICP parameter selection unit 605.

  In each of the above embodiments, the ICP parameter is obtained by minimizing the mean square error (MSE) between the prediction signal and the reference signal, but the present invention is not limited to this.

  In each of the above embodiments, the stereo / monaural signal has been described as an audio signal, but the present invention can also be applied to an audio signal or the like.

  The present invention can also be used in the LP residual signal region. In this case, the present invention can be used by performing the ICP process using the LP residual signal as the input signal and the reference signal.

  The above description is an illustration of a preferred embodiment of the present invention, and the scope of the present invention is not limited to this. The present invention can be applied to any system as long as the system includes an encoding device and a decoding device.

  Also, the encoding device and the decoding device according to the present invention can be mounted on a communication terminal device and a base station device in a mobile communication system, whereby a communication terminal device and a base having the same operational effects as described above. A station apparatus and a mobile communication system can be provided.

  Further, here, the case where the present invention is configured by hardware has been described as an example, but the present invention can also be realized by software. For example, an algorithm according to the present invention is described in a programming language, and this program is stored in a memory and executed by information processing means, thereby realizing the same function as the encoding device / decoding device according to the present invention. be able to.

  Each functional block used in the description of the above embodiment is typically realized as an LSI which is an integrated circuit. These may be individually made into one chip, or may be made into one chip so as to include some or all of them.

  Although referred to as LSI here, it may be called IC, system LSI, super LSI, ultra LSI, or the like depending on the degree of integration.

  Further, the method of circuit integration is not limited to LSI's, and implementation using dedicated circuitry or general purpose processors is also possible. An FPGA (Field Programmable Gate Array) that can be programmed after manufacturing the LSI or a reconfigurable processor that can reconfigure the connection or setting of circuit cells inside the LSI may be used.

  Furthermore, if integrated circuit technology that replaces LSI emerges as a result of progress in semiconductor technology or other derived technology, it is naturally also possible to integrate functional blocks using this technology. Biotechnology can be applied as a possibility.

  The encoding device and decoding device according to the present invention are suitable for use in mobile phones, IP phones, video conferences, and the like.

FIG. 1 is a block diagram showing a configuration of an encoding apparatus according to Embodiment 1 of the present invention. The figure which shows the structure of the adaptive filter H (z) of the ICP analysis part of the encoding apparatus which concerns on Embodiment 1 of this invention. The figure which shows the other method of grouping the input sample in the ICP analysis part of the encoding apparatus which concerns on Embodiment 1 of this invention. The block diagram which shows the structure of the decoding apparatus which concerns on Embodiment 1 of this invention. The figure which shows the waveform of 1 frame of a stereo audio | voice signal Block diagram showing a configuration of an encoding apparatus according to Embodiment 2 of the present invention. The block diagram which shows the internal structure of the ICP parameter selection part which concerns on Embodiment 2 of this invention. The block diagram which shows the structure of the decoding apparatus which concerns on Embodiment 2 of this invention. General block diagram of ICP The figure which shows the filter structure in the conventional ICP

Explanation of symbols

DESCRIPTION OF SYMBOLS 100,600 Encoding apparatus 101 Monaural signal synthetic | combination part 102 Monaural encoding part 103 Local decoding part 104,105 ICP analysis part 106,107 ICP parameter quantization part 108 Multiplexing part 400,800 Decoding apparatus 401 Separation part 402 Mono decoding part 403, 404 ICP parameter decoding unit 405, 406, 805, 806 ICP synthesis filter 604, 605 ICP parameter selection unit 701 First ICP analysis unit 702 Second ICP analysis unit 703, 704 Local decoding unit 705, 706 Gain calculation unit 707 Comparison unit 708 Decision part

Claims (8)

Monaural reproduction signal generating means for generating a monaural reproduction signal using the first channel signal and the second channel signal of a stereo signal;
A first filter coefficient generated by using a signal obtained by a first filter process using a first sample sequence obtained by grouping a plurality of samples of the monaural reproduction signal input in the past and the first channel signal is obtained. A first inter-channel prediction analysis means for performing a first inter-channel prediction analysis as a first inter-channel prediction parameter;
A second filter coefficient generated by using the signal obtained by the second filter processing using the second sample sequence obtained by grouping a plurality of samples of the monaural reproduction signal inputted in the past and the second channel signal is obtained. Second inter-channel prediction analysis means for performing second inter-channel prediction analysis, which is a second inter-channel prediction parameter;
An encoding device comprising: The monaural reproduction signal generating means includes
Monaural signal generating means for generating a monaural signal by combining the first channel signal and the second channel signal;
Monaural encoding means for encoding the monaural signal to generate encoded data;
Monaural decoding means for decoding the encoded data and generating the monaural reproduction signal;
The encoding device according to claim 1, further comprising: The first inter-channel prediction analysis means includes:
Generating the first filter coefficient that minimizes the mean square error between the signal obtained by the first filter processing and the first channel signal;
The second inter-channel prediction analysis means includes:
Generating the second filter coefficient that minimizes the mean square error between the signal obtained by the second filter processing and the second channel signal;
The encoding device according to claim 1. The first inter-channel prediction analysis means includes:
A third filter coefficient generated by using the signal obtained by the third filter processing using the third sample sequence that is a sample sequence of the monaural reproduction signal input in the past and the first channel signal is expressed as a third filter coefficient. A third inter-channel prediction analysis is performed as an inter-channel prediction parameter, and the inter-channel prediction parameter having a higher prediction gain of the first inter-channel prediction analysis and the third inter-channel prediction analysis is first encoded. Select as the inter-channel prediction parameter for
The second inter-channel prediction analysis means includes:
Fourth inter-channel prediction analysis using a fourth filter coefficient generated by using the signal obtained by the fourth filter processing using the third sample sequence and the second channel signal as a fourth inter-channel prediction parameter. And the inter-channel prediction parameter having the higher prediction gain of the second inter-channel prediction analysis and the fourth inter-channel prediction analysis is selected as the second inter-channel prediction parameter for encoding.
The encoding device according to claim 1. Monaural decoding means for decoding the encoded data of the monaural signal generated by the encoding device to generate a monaural reproduction signal;
The first inter-channel prediction analysis using the encoding-side first sample sequence obtained by grouping a plurality of samples of the encoding-side monaural reproduction signal and the first-channel signal of the encoding-side stereo signal, performed in the encoding device. The first inter-channel prediction parameter generated by the first synthesis filter that can be used as a first filter coefficient, and a first sample sequence in which a plurality of samples of the monaural reproduction signal are grouped is input. First inter-channel prediction synthesis means for generating a first channel reproduction signal of a stereo reproduction signal by filtering with the first synthesis filter using the first inter-channel prediction parameter;
Between the second channels using the encoding-side second sample sequence obtained by grouping a plurality of samples of the encoding-side monaural reproduction signal and the second-channel signal of the encoding-side stereo signal made in the encoding device The second inter-channel prediction parameter generated by the predictive analysis has a second synthesis filter that can be used as a second filter coefficient, and a second sample sequence in which a plurality of samples of the monaural reproduction signal are grouped is input. Second inter-channel prediction synthesis means for generating a second channel reproduction signal of the stereo reproduction signal by filtering with the second synthesis filter using the second inter-channel prediction parameter,
A decoding device comprising: The first inter-channel prediction synthesis means includes:
Third inter-channel prediction analysis using the encoding-side third sample sequence, which is a sample sequence of the encoding-side monaural reproduction signal, and the first-channel signal of the encoding-side stereo signal, performed in the encoding device. A third synthesis filter capable of using the third inter-channel prediction parameter generated by the above as a third filter coefficient;
Instead of filtering the first sample sequence with the first synthesis filter,
Of the first synthesis filter and the third synthesis filter, the first sample sequence or the third sample sequence is filtered by a synthesis filter instructed by the encoding device, so that the stereo reproduction signal A first channel reproduction signal is generated;
The second inter-channel prediction synthesis means includes:
A fourth inter-channel prediction parameter generated by a fourth inter-channel prediction analysis using the third sample sequence on the encoding side and the second channel signal of the stereo signal on the encoding side made in the encoding device. And a fourth synthesis filter that can be used as a fourth filter coefficient,
Instead of filtering the second sample sequence with the second synthesis filter,
Of the second synthesis filter and the fourth synthesis filter, the second sample sequence or the third sample sequence is filtered by a synthesis filter instructed by the encoding device, so that the stereo reproduction signal Generating a second channel reproduction signal;
The decoding device according to claim 5. A monaural reproduction signal generating step of generating a monaural reproduction signal using the first channel signal and the second channel signal of the stereo signal;
A first filter coefficient generated by using a signal obtained by a first filter process using a first sample sequence obtained by grouping a plurality of samples of the monaural reproduction signal input in the past and the first channel signal is obtained. A first inter-channel prediction analysis step for performing a first inter-channel prediction analysis as a first inter-channel prediction parameter;
A second filter coefficient generated by using the signal obtained by the second filter processing using the second sample sequence obtained by grouping a plurality of samples of the monaural reproduction signal inputted in the past and the second channel signal is obtained. A second inter-channel prediction analysis step for performing a second inter-channel prediction analysis as a second inter-channel prediction parameter;
An encoding method comprising: A monaural decoding step of generating a monaural reproduction signal by decoding encoded data of the monaural signal generated by the encoding device;
The first inter-channel prediction analysis using the encoding-side first sample sequence obtained by grouping a plurality of samples of the encoding-side monaural reproduction signal and the first-channel signal of the encoding-side stereo signal, performed in the encoding device. Preparing a first synthesis filter that can use the first inter-channel prediction parameter generated as a first filter coefficient, and inputting a first sample sequence in which a plurality of samples of the monaural reproduction signal are grouped A first inter-channel prediction synthesis step of generating a first channel reproduction signal of a stereo reproduction signal by filtering with the first synthesis filter using a first inter-channel prediction parameter;
Between the second channels using the encoding-side second sample sequence obtained by grouping a plurality of samples of the encoding-side monaural reproduction signal and the second-channel signal of the encoding-side stereo signal made in the encoding device A second synthesis filter capable of using the second inter-channel prediction parameter generated by the prediction analysis as a second filter coefficient is prepared, and a second sample sequence obtained by grouping a plurality of samples of the monaural reproduction signal is input. A second inter-channel prediction synthesis step of generating a second channel reproduction signal of the stereo reproduction signal by filtering with the second synthesis filter using the second inter-channel prediction parameter;
A decryption method.

Images