JP2011154383A - Voice encoding device, voice decoding device and methods thereof - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To accurately specify a band having a large error among all the bands by using a small calculation amount. <P>SOLUTION: A first position identification unit 201 uses a first layer error conversion coefficient including an error of a decoding signal for an input signal to search for a band having a large error in a relatively wide bandwidth in all the bands of the input signal and generates first position information indicating the identified band. A second position identification unit 202 searches for a target frequency band having a large error in a relatively narrow bandwidth in the band identified by the first position identification unit 201, and generates second position information indicating the identified target frequency band. The encoding unit 203 encodes a first layer decoding error conversion coefficient contained in the target frequency band, and generates encoding information. The first position information, the second position information, and the encoding information are transmitted to a communication partner. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a speech coding apparatus, a speech decoding apparatus, and a method thereof used in a scalable coding communication system.

  In a mobile communication system, it is required to compress and transmit an audio signal at a low bit rate in order to effectively use radio resources and the like. On the other hand, it is also desired to improve the quality of call voice and to realize a call service with a high sense of reality. For this purpose, not only the quality of the audio signal but also the audio signal with a wider bandwidth, etc. It is desirable to encode these signals with high quality.

  For such two conflicting requirements, a technique for hierarchically integrating a plurality of encoding techniques is considered promising. This technology is a model suitable for audio signals and a first layer that encodes an input signal at a low bit rate, and a differential signal between the input signal and the decoded signal of the first layer is also a model suitable for signals other than audio. The second layer to be encoded is combined hierarchically. The technique of performing hierarchical encoding in this way is general because the bitstream obtained from the encoding device has scalability, that is, a decoded signal can be obtained even from partial information of the bitstream. This is called scalable coding (hierarchical coding).

  The scalable coding scheme can be flexibly adapted to communication between networks having different bit rates because of its nature, and can be said to be suitable for a future network environment in which various networks are integrated by the IP protocol.

  As an example of realizing scalable coding using a technique standardized by MPEG-4 (Moving Picture Experts Group phase-4), there is a technique disclosed in Non-Patent Document 1, for example. This technique uses CELP (Code Excited Linear Prediction) coding suitable for a speech signal in the first layer, and subtracts the first layer decoded signal from the original signal in the second layer. On the other hand, transform coding such as AAC (Advanced Audio Coder) and TwinVQ (Transform Domain Weighted Interleave Vector Quantization) is used.

  On the other hand, Non-Patent Document 2 discloses a technique of hierarchically encoding MDCT coefficients in a desired frequency band using TwinVQ modularized as a basic structural unit. By using the module in common and using it a plurality of times, a simple and highly flexible scalable encoding can be realized. In this method, the subbands to be encoded in each layer (layer) are basically determined in advance, but the position of the subband to be encoded in each layer according to the nature of the input signal. A configuration is also disclosed in which the frequency fluctuates within a predetermined band.

Edited by Junichi Miki, “All about MPEG-4”, first edition, Industrial Research Co., Ltd., September 30, 1998, p. 126-127 Jinakio et al., “Scalable Music Coding Constructed by Hierarchical Transform Coding Basic Module”, IEICE Transactions A, Vol. J83-A, No.3, pp.241-252, March 2000 “AMR Wideband Speech Codec; Transcoding functions”, 3GPP TS 26.190, March 2001. “Source-Controlled-Variable-Rate Multimode Wideband Speech Codec (VMR-WB), Service options 62 and 63 for Spread Spectrum Systems”, 3GPP2 C.S0052-A, April 2005. "7/10/15 kHz band scalable speech coding using band expansion technology by pitch filtering", pp.327-328, March 2004.

  However, in order to improve the voice quality of the output signal, it is important how to set the subband (target frequency band) of the second layer encoding unit. According to the technique disclosed in Non-Patent Document 2, the subbands to be encoded in the second layer are determined in advance (FIG. 21A). In this case, since the quality of a predetermined subband is always improved, there is a problem that a sufficient voice quality improvement effect cannot be obtained when error components are concentrated in a band other than the subband.

  Further, it is described that the position of the subband to be encoded in each layer (layer) is changed in a predetermined band in accordance with the nature of the input signal (FIG. 21B). Since the position where the subband can be taken is limited to a predetermined band, the above-described problem cannot be solved. In addition, if the bandwidth that can be taken by the subband extends over the entire bandwidth of the input signal (FIG. 21C), there is a problem that the amount of calculation for specifying the position of the subband increases. Further, when the number of layers is increased, this problem becomes significant because it is necessary to specify the position of the subband for each layer.

  The present invention has been made in view of the above points, and in a scalable coding system, a speech coding apparatus and speech decoding apparatus that can accurately identify a band having a large error from all bands with a small amount of computation. And an object thereof.

  An encoding apparatus according to a first aspect of the present invention includes a conversion unit that converts an input signal into a conversion coefficient, a specifying unit that specifies a target frequency band to be encoded, and a target frequency among the conversion coefficients. Encoding means for encoding a transform coefficient included in a band, wherein the specifying means defines a first band having the largest transform coefficient in a bandwidth wider than the target frequency band in a predetermined first step. A first position specifying means for searching for a width and generating first position information indicating the specified first band; and the target frequency band with a second step width smaller than the first step width across the first band. And a second position specifying means for generating second position information indicating the specified target frequency band, and before being included in the target frequency band specified by the first position information and the second position information. A configuration that includes encoding means for generating encoded information transform coefficients is encoded, the.

  The decoding device according to the second aspect of the present invention provides encoded data obtained by performing an encoding process on a transform coefficient included in a target frequency band to be encoded, a band wider than the target frequency. Receiving means for receiving first position information indicating the first band having the largest transform coefficient in width and second position information indicating the target frequency band in the first position band; and decoding the encoded data Decoding means for generating a decoded transform coefficient, and an arrangement means for specifying the target frequency band based on the first position information and the second position information, and arranging the decoded transform coefficient in the target frequency band; The structure which comprises is taken.

  The encoding method according to the third aspect of the present invention includes a conversion step for converting an input signal into a conversion coefficient, a specifying step for specifying a target frequency band to be encoded, and a target frequency among the conversion coefficients. An encoding step for encoding a transform coefficient included in a band, wherein the specifying step includes a first band in which the transform coefficient is the largest in a bandwidth wider than the target frequency band in a predetermined first step. A first position specifying step of searching for a width and generating first position information indicating the specified first band; and the target frequency band in a second step width smaller than the first step width across the first band. A second position specifying step for generating second position information indicating the specified target frequency band, and the target specified by the first position information and the second position information. Adopt a method comprising an encoding step of generating encoded information the transform coefficients included in the frequency band is encoded, a.

  The decoding method according to the fourth aspect of the present invention provides encoded data obtained by performing an encoding process on a transform coefficient included in a target frequency band to be encoded, a band wider than the target frequency. A receiving step of receiving first position information indicating a first band having the largest transform coefficient in width and second position information indicating the target frequency band in the first position band; and decoding the encoded data A decoding step of generating a decoded transform coefficient, an arrangement step of identifying the target frequency band based on the first position information and the second position information, and arranging the decoded transform coefficient in the target frequency band; A method comprising:

  According to the present invention, it is possible to accurately specify a band having a large error from all bands with a small amount of calculation, and to improve sound quality.

FIG. 1 is a block diagram showing the main configuration of an encoding apparatus according to Embodiment 1 of the present invention. The block diagram which shows the structure of the 2nd layer encoding part shown in FIG. The figure which shows the position of the zone | band which the 1st position specific part shown in FIG. 2 specifies. The figure which shows the other position of the zone | band which the 1st position specific part shown in FIG. 2 specifies. The figure which shows the position of the target frequency band which the 2nd position specific part shown in FIG. 2 specifies. The block diagram which shows the structure of the encoding part shown in FIG. The block diagram which shows the main structures of the decoding apparatus which concerns on Embodiment 1 of this invention. The figure which shows the structure of the 2nd layer decoding part shown in FIG. The figure which shows the mode of the 1st layer decoding error conversion coefficient output from the arrangement | positioning part shown in FIG. The figure which shows the position of the target frequency which the 2nd position specific part shown in FIG. 2 specifies. The block diagram which shows the structure of another aspect of the encoding part shown in FIG. The block diagram which shows the structure of another aspect of the 2nd layer decoding part shown in FIG. The block diagram which shows the structure of the 2nd layer encoding part of the encoding apparatus which concerns on Embodiment 3 of this invention. The figure which shows the position of the target frequency which the some sub position specific | specification part of the encoding apparatus which concerns on Embodiment 3 specifies. The block diagram which shows the structure of the 2nd layer encoding part of the encoding apparatus which concerns on Embodiment 4 of this invention. The block diagram which shows the structure of the encoding part shown in FIG. The figure which shows an encoding part in case each 2nd position information candidate memorize | stored in the 2nd position information codebook of FIG. 16 has three target frequencies. The block diagram which shows another structure of the encoding part shown in FIG. Block diagram showing the configuration of the second layer encoding section according to Embodiment 5 of the present invention The figure which shows the position of the zone | band which the 1st position specific part shown in FIG. 19 specifies. The figure which shows the encoding band of the 2nd layer encoding part of the conventional audio | voice coding apparatus. FIG. 9 is a block diagram showing the main configuration of an encoding apparatus according to Embodiment 6 FIG. 22 is a block diagram showing the configuration of the first layer encoding unit of the encoding apparatus shown in FIG. FIG. 22 is a block diagram showing the configuration of the first layer decoding unit of the encoding apparatus shown in FIG. FIG. 22 is a block diagram showing the main configuration of a decoding apparatus corresponding to the encoding apparatus shown in FIG. FIG. 9 is a block diagram showing the main configuration of an encoding apparatus according to Embodiment 7 FIG. 26 is a block diagram showing the main configuration of a decoding apparatus corresponding to the encoding apparatus shown in FIG. FIG. 11 is a block diagram showing the main configuration of an encoding apparatus according to another aspect according to Embodiment 7. The figure which shows the position of the band in the 2nd layer encoding part shown in FIG. The figure which shows the position of the band in the 3rd layer encoding part shown in FIG. The figure which shows the position of the band in the 4th layer encoding part shown in FIG. FIG. 28 is a block diagram showing the main configuration of a decoding apparatus corresponding to the encoding apparatus shown in FIG. The figure which shows the other position of the band in the 2nd layer encoding part shown in FIG. The figure which shows the other position of the band in the 3rd layer encoding part shown in FIG. The figure which shows the other position of the band in the 4th layer encoding part shown in FIG. The figure for demonstrating operation | movement of the 1st position specific | specification part which concerns on Embodiment 8. FIG. FIG. 9 is a block diagram showing a configuration of a first position specifying unit according to the eighth embodiment. The figure which illustrates a mode that 1st position information is comprised in the 1st position information structure part which concerns on Embodiment 8. FIG. The figure for demonstrating the decoding process which concerns on Embodiment 8. FIG. The figure for demonstrating the variation which concerns on Embodiment 8. FIG. The figure for demonstrating the variation which concerns on Embodiment 8. FIG.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

(Embodiment 1)
FIG. 1 is a block diagram showing the main configuration of the coding apparatus according to Embodiment 1 of the present invention. 1 includes a frequency domain transform unit 101, a first layer coding unit 102, a first layer decoding unit 103, a subtraction unit 104, a second layer coding unit 105, and a multiplexing unit. Unit 106.

  Frequency domain transform section 101 transforms a time domain input signal into a frequency domain signal (input transform coefficient), and outputs the input transform coefficient to first layer encoding section 102.

  First layer encoding section 102 performs encoding processing on the input transform coefficients, generates first layer encoded data, and uses this first layer encoded data as first layer decoding section 103 and multiplexing section 106. Output to.

  First layer decoding section 103 performs decoding processing using the first layer encoded data, generates a first layer decoded transform coefficient, and outputs the first layer decoded transform coefficient to subtracting section 104.

  The subtracting unit 104 subtracts the first layer decoded transform coefficient generated by the first layer decoding unit 103 from the input transform coefficient to generate a first layer error transform coefficient, and the first layer error transform coefficient is converted into a second layer code. To the conversion unit 105.

  Second layer encoding section 105 performs encoding processing of the first layer error transform coefficient output from subtracting section 104, generates second layer encoded data, and multiplexes the second layer encoded data. The data is output to 106.

  The multiplexing unit 106 multiplexes the first layer encoded data obtained by the first layer encoding unit 102 and the second layer encoded data obtained by the second layer encoding unit 105 to form a bit stream. Is output to the communication path as final encoded data.

  FIG. 2 is a block diagram showing a configuration of second layer encoding section 105 shown in FIG. The second layer encoding unit 105 illustrated in FIG. 2 includes a first position specifying unit 201, a second position specifying unit 202, an encoding unit 203, and a multiplexing unit 204.

  The first position specifying unit 201 uses the first layer error transform coefficient input from the subtracting unit 104 to determine a band that can be taken by the target frequency band to be encoded with a predetermined bandwidth and a predetermined step size. The information indicating the searched and specified band is output as first position information to the second position specifying unit 202, the encoding unit 203, and the multiplexing unit 204. Details of the first position specifying unit 201 will be described later. The specified band can also be referred to as “range” or “area”.

  The second position specifying unit 202 has a bandwidth narrower than the bandwidth of the first position specifying unit 201 and the step size of the first position specifying unit 201 out of the bands specified by the first position specifying unit 201. The target frequency band is searched with a fine step size, and information indicating the identified target frequency band is output to the encoding unit 203 and the multiplexing unit 204 as second position information. Details of the second position specifying unit 202 will be described later.

  The encoding unit 203 encodes the first layer error transform coefficient included in the target frequency band specified by the first position information and the second position information, generates encoded information, and outputs the encoded information to the multiplexing unit 204. Details of the encoding unit 203 will be described later.

  The multiplexing unit 204 multiplexes the first position information, the second position information, and the encoded information, generates second layer encoded data, and outputs it. Note that the multiplexing unit 204 is not essential, and such information may be output directly to the multiplexing unit 106 shown in FIG.

  FIG. 3 is a diagram showing a band specified by the first position specifying unit 201 shown in FIG.

In FIG. 3, the first position specifying unit 201 specifies one of three bands set in advance with a predetermined bandwidth, and uses the position information of this band as the first position information, and the second position specifying unit 201 202, and output to the encoding unit 203 and the multiplexing unit 204. Each band shown in FIG. 3 is set to have a bandwidth equal to or larger than the target frequency band (band 1 is F ₁ or more and less than F ₃ , band 2 is F ₂ or more and less than F ₄ , and band 3 is F ₃ or more and F _5. Less than). In this embodiment, each band is set to have the same bandwidth, but each band may be set to have a different bandwidth. For example, it may be set so that the bandwidth of the band located in the low band is narrow and the bandwidth of the band located in the high band is wide like the critical bandwidth of human hearing.

ここで、ｉは帯域を特定する識別子、ＦＲＬ（ｉ）は帯域ｉの最低域周波数、ＦＲＨ（ｉ)は帯域ｉの最高域周波数を表す。 Next, a band specifying method by the first position specifying unit 201 will be described. Here, the 1st position specific | specification part 201 specifies a zone | band on the basis of the magnitude | size of the energy of a 1st layer error conversion coefficient. The first layer error conversion coefficient is denoted as e ₁ (k), and the energy E _R (i) of the first layer error conversion coefficient included in each band is calculated by the following equation (1).

Here, i is an identifier for identifying a band, FRL (i) is the lowest frequency of the band i, and FRH (i) is the highest frequency of the band i.

  In this way, by identifying the band where the energy of the first layer error transform coefficient is large and encoding the first layer error transform coefficient included in the band where the error is large, the error of the decoded signal with respect to the input signal is reduced, Voice quality can be improved.

Instead of the energy of the first layer error conversion coefficient, normalized energy NE _R (i) normalized by the bandwidth may be calculated as in the following equation (2).

In addition, as a reference for specifying the band, instead of the energy of the first layer error conversion coefficient, the energy of the first layer error conversion coefficient WE _R (i), WNE _R ( i) (Normalized energy normalized by the bandwidth) may be calculated by the equations (3) and (4). Here, w (k) represents a weight related to human auditory characteristics.

  In this case, the first position specifying unit 201 increases the weight for a frequency with high importance in order to facilitate selection of a band including the frequency, while weighting for a frequency with low importance. To make it difficult to select a band including the frequency. As a result, a band that is audibly important is preferentially selected, so that the same sound quality improvement effect as described above can be obtained. As this weight, for example, an auditory masking threshold calculated based on an input signal or a decoded signal of the first layer or a human auditory loudness characteristic may be used.

  Further, in the band selection method, a band may be selected from bands arranged in a low frequency part whose frequency is lower than a preset reference frequency (Fx). In the example of FIG. 4, a band is selected from band 1 to band 8. The reason for setting the limit (reference frequency) for the selection of the band is as follows. A harmonic structure or a harmonic structure (a structure in which a spectrum appears in a peak shape at a certain frequency interval), which is one of the characteristics of an audio signal, has a larger peak in the low band than in the high band, and is caused by the encoding process. Similarly, in the quantization error (error spectrum or error conversion coefficient), the peak property in the low frequency region is stronger than that in the high frequency region. Therefore, even if the energy of the low-frequency part error spectrum (error conversion coefficient) is smaller than that of the high-frequency part, the peak of the low-frequency part error spectrum (error conversion coefficient) is stronger than that of the high-frequency part. Due to this tendency, the low-frequency part error spectrum (error conversion coefficient) tends to exceed the auditory masking threshold (threshold by which a human can perceive sound), resulting in auditory sound quality degradation.

  According to this method, by setting the reference frequency in advance, the error conversion coefficient (or error vector) has a higher peak characteristic of the error conversion coefficient (or error vector) than the high frequency part where the frequency is higher than the reference frequency (Fx). Since the frequency is determined, the peak of the error conversion coefficient can be suppressed and the sound quality can be improved.

  Further, in the band selection method, the band may be selected from the bands arranged in the low and mid-range parts. In the example of FIG. 3, band 3 is excluded from selection candidates, and a band is selected from band 1 and band 2. As a result, the target frequency band is determined from the low mid-range portion.

  Hereinafter, the first position specifying unit 201 sets the first position information to “1” when the band 1 is specified, “2” when the band 2 is specified, and “3” when the band 3 is specified. Output as.

  FIG. 5 is a diagram showing the position of the target frequency band specified by the second position specifying unit 202 shown in FIG.

  The second position specifying unit 202 specifies a target frequency band with a finer step size in the band specified by the first position specifying unit 201, and the position information of the target frequency band is set as second position information in the encoding unit 203. And output to the multiplexing unit 204.

Next, a method for specifying a target frequency band by the second position specifying unit 202 will be described. Here, the case where the first position information output from the first position specifying unit 201 shown in FIG. 2 is “2” is taken as an example, and the width of the target frequency band is BW. Further, the lowest frequency F ₂ in the band 2 is set as a starting point, and this lowest frequency F ₂ is set as G ₁ for convenience. Then, the lowest frequency of the target frequency band can be a second position specifying section 202 specifies a G ₂ ~G _N. The step size of the target frequency band specified by the second position specifying unit 202 is G _n −G _n−1 , while the step size of the band specified by the first position specifying unit 201 is F _n −F _n−. it is _{1 (G n -G n-1} <F n -F n-1).

Second position specifying section 202, G ₁ lowest frequency, _{respectively,} ..., from the target frequency band candidates G _N, the energy or reference similar to that of the first layer error transform coefficients, to identify the target frequency band. For example, for all G _n target frequency band candidates, the energy of the first layer error conversion coefficient is calculated by Equation (5), and the target frequency band where the calculated energy E _R (n) is maximum is specified. The position information of the target frequency band is output as the second position information.

Incidentally, as described above, if energy WE R of first layer error transform coefficients weighting that reflects the characteristics of human perception has been performed _(n) is a reference, the following equation (6) WE R _(n) Is calculated. Here, w (k) represents a weight related to human auditory characteristics. As this weight, for example, an auditory masking threshold calculated based on an input signal or a decoded signal of the first layer or a human auditory loudness characteristic may be used.

  In this case, the second position specifying unit 202 increases the weight for a frequency with high importance in order to facilitate selection of a target frequency band including the frequency, while reducing the frequency to a frequency with low importance. Reduces the weight so that the target frequency band including the frequency is not easily selected. Thereby, since the target frequency band important perceptually is preferentially selected, the sound quality can be further improved.

  FIG. 6 is a block diagram showing a configuration of encoding section 203 shown in FIG. The encoding unit 203 illustrated in FIG. 6 includes a target signal configuration unit 301, an error calculation unit 302, a search unit 303, a shape code book 304, and a gain code book 305.

The target signal constituting unit 301 specifies the target frequency band using the first position information input from the first position specifying unit 201 and the second position information input from the second position specifying unit 202, and the subtracting unit A portion included in the target frequency band is extracted from the first layer error conversion coefficient input from 104, and the extracted first layer error conversion coefficient is output to the error calculation unit 302 as a target signal. This first error conversion coefficient is represented as e ₁ (k).

ここで、ｓｈ（ｉ，ｋ）は第ｉ番目の形状候補、ｇａ（ｍ）は第ｍ番目のゲイン候補を表す。 The error calculation unit 302 stores the i-th shape candidate input from the shape codebook 304 that stores candidates (shape candidates) representing the shape of the error conversion coefficient, and candidates (gain candidates) representing the gain of the error conversion coefficient. The error E is calculated by the following equation (7) based on the mth gain candidate input from the gain codebook 305 and the target signal input from the target signal configuration unit 301, and the calculated error E is Output to the search unit 303.

Here, sh (i, k) represents the i-th shape candidate, and ga (m) represents the m-th gain candidate.

  The search unit 303 searches for a combination of a shape candidate and a gain candidate with the smallest error E based on the error E calculated by the error calculation unit 302, and encodes shape information and gain information as a search result. Information is output to the multiplexing unit 204 shown in FIG. Here, the shape information indicates a parameter m when the error E is minimized, and the gain information indicates a parameter i when the error E is minimized.

Note that the error calculation unit 302 may increase the influence of the audibly important spectrum by applying a large weight to the audibly important spectrum, and obtain the error E by the following equation (8). Here, w (k) represents a weight related to human auditory characteristics.

  In this way, weights are increased for frequencies that are more important for auditory characteristics, and the influence of quantization distortion for frequencies that are more important for auditory characteristics is increased, while weights for frequencies that are less important are weighted. And the subjective quality can be improved by reducing the influence of quantization distortion at a low importance frequency.

  FIG. 7 is a block diagram showing the main configuration of the decoding apparatus according to the present embodiment. A decoding apparatus 600 illustrated in FIG. 7 includes a separation unit 601, a first layer decoding unit 602, a second layer decoding unit 603, an addition unit 604, a switching unit 605, a time domain conversion unit 606, and a post filter 607. With.

  Separating section 601 separates the bit stream input via the communication path into first layer encoded data and second layer encoded data, and converts the first layer encoded data to first layer decoding section 602, respectively. The second layer encoded data is output to second layer decoding section 603. In addition, when both the first layer encoded data and the second layer encoded data are included in the input bitstream, the separation unit 601 outputs “2” as layer information to the switching unit 605. On the other hand, when only the first layer encoded data is included in the bitstream, the separation unit 601 outputs “1” to the switching unit 605 as layer information. Note that all encoded data may be discarded. In this case, the decoding unit of each layer performs predetermined error compensation processing, and the post filter performs processing with layer information “1”. . In the present embodiment, description will be made on the premise that all the encoded data or the encoded data in which the second layer encoded data is discarded is obtained in the decoding apparatus.

  First layer decoding section 602 performs a decoding process on the first layer encoded data, generates a first layer decoded transform coefficient, and outputs the first layer decoded transform coefficient to addition section 604 and switching section 605.

  Second layer decoding section 603 performs decoding processing on the second layer encoded data, generates a first layer decoding error transform coefficient, and outputs the first layer decoding error transform coefficient to adding section 604.

  Adder 604 adds the first layer decoded transform coefficient and the first layer decoded error transform coefficient to generate a second layer decoded transform coefficient, and outputs the second layer decoded transform coefficient to switching section 605.

  Based on the layer information input from the separation unit 601, the switching unit 605 performs the first layer decoding transform coefficient when the layer information is “1” and the second layer decoding transform when the layer information is “2”. The coefficients are output to the time domain transform unit 606 as decoded transform coefficients.

  The time domain transform unit 606 converts the decoded transform coefficient into a time domain signal, generates a decoded signal, and outputs the decoded signal to the post filter 607.

  The post filter 607 performs post filter processing on the decoded signal output from the time domain conversion unit 606 to generate an output signal.

  FIG. 8 is a diagram showing a configuration of second layer decoding section 603 shown in FIG. Second layer decoding section 603 shown in FIG. 8 includes shape codebook 701, gain codebook 702, multiplication section 703, and arrangement section 704.

  The shape codebook 701 selects a shape candidate sh (i, k) based on the shape information included in the second layer encoded data output from the separation unit 601 and outputs the shape candidate sh (i, k) to the multiplication unit 703.

  Gain codebook 702 selects gain candidate ga (m) based on the gain information included in the second layer encoded data output from demultiplexing section 601, and outputs it to multiplication section 703.

  The multiplication unit 703 multiplies the shape candidate sh (i, k) by the gain candidate ga (m) and outputs the result to the arrangement unit 704.

  Arrangement section 704 performs gain candidate multiplication input from multiplication section 703 on the target frequency band specified by the first position information and the second position information included in the second layer encoded data output from separation section 601. Subsequent shape candidates are arranged and output to the adding unit 604 as first layer decoding error transform coefficients.

FIG. 9 is a diagram illustrating a state of the first layer decoding error transform coefficients output from the arrangement unit 704 illustrated in FIG. Here, F _m represents a frequency specified by the first position information, and G _n represents a frequency specified by the second position information.

  As described above, according to the present embodiment, the first position specifying unit 201 searches and specifies a band having a large error with a predetermined bandwidth and a predetermined step width over the entire band of the input signal, The second position specifying unit 202 searches and specifies a target frequency band in a band specified by the first position specifying unit 201 with a bandwidth narrower than the predetermined bandwidth and a step size finer than the predetermined step width. By doing so, it is possible to accurately specify a band having a large error from all bands with a small amount of calculation, and to improve sound quality.

(Embodiment 2)
In the second embodiment, another method for specifying the target frequency band by the second position specifying unit 202 will be described. FIG. 10 is a diagram showing the position of the target frequency specified by the second position specifying unit 202 shown in FIG. The second position specifying unit of the coding apparatus according to the present embodiment is different from the second position specifying unit of the coding apparatus described in the first embodiment, and specifies a single target frequency. The shape candidate of the error conversion coefficient corresponding to a single target frequency is represented by a pulse (or line spectrum). In the present embodiment, the configuration of the encoding apparatus is the same as that of the encoding apparatus shown in FIG. 1 except for the internal configuration of encoding section 203, and the configuration of the decoding apparatus is second layer decoding section 603. 7 is the same as the decoding apparatus shown in FIG. 7, the description thereof will be omitted, and only the encoding unit 203 related to the second position identification and the second layer decoding unit 603 of the decoding apparatus will be described. explain.

  In the present embodiment, the second position specifying unit 202 specifies a single target frequency in the band specified by the first position specifying unit 201. Therefore, in the present embodiment, a single first layer error transform coefficient is selected as an encoding target. Here, a case where the first position specifying unit 201 specifies the band 2 will be described as an example. When the bandwidth of the target frequency band is BW, in this embodiment, BW = 1.

Specifically, second position specifying section 202, as shown in FIG. 10, for a plurality of target frequency candidates G _N contained in the band 2, by the above equation (5), the first layer error transform each The energy of the coefficient is calculated, or the energy of the first layer error conversion coefficient that is weighted to reflect the human auditory sensation characteristic is calculated by the above equation (6). Further, the second position specifying unit 202 specifies the target frequency G _n (1 ≦ n ≦ N) that maximizes the calculated energy, and encodes the position information of the specified target frequency G _n as the second position information. The data is output to the unit 203.

  FIG. 11 is a block diagram showing a configuration of another aspect of encoding section 203 shown in FIG. 11 employs a configuration in which the shape codebook 305 is deleted from FIG. This configuration corresponds to the case where the signal output from the shape codebook 304 is always “1”.

The encoding unit 203 encodes the first layer error transform coefficient included in the target frequency _Gn specified by the second position specifying unit 202, generates encoding information, and outputs the encoded information to the multiplexing unit 204. Here, since the target frequency input from the second position specifying unit 202 is single and the first layer error transform coefficient to be encoded is also single, the encoding unit 203 is based on the shape codebook 304. The shape information is not required, and only the gain codebook 305 is searched, and the gain information of the search result is output to the multiplexing unit 204 as encoded information.

  FIG. 12 is a block diagram showing a configuration of another aspect of second layer decoding section 603 shown in FIG. Second layer decoding section 603 shown in FIG. 12 adopts a configuration in which shape codebook 701 and multiplication section 703 are deleted from FIG. This configuration corresponds to the case where the signal output from the shape codebook 701 is always “1”.

  Arrangement unit 704 selects a single target frequency specified by the first position information and the second position information included in the second layer encoded data output from separation unit 601 from the gain codebook using gain information. The gain candidates are arranged and output to the adding unit 604 as first layer decoding error transform coefficients.

  As described above, according to the present embodiment, the second position specifying unit 202 specifies the single target frequency from the band specified by the first position specifying unit 201, thereby accurately determining the line spectrum. Therefore, it is possible to improve the sound quality of a signal having a strong tonality such as a vowel (a signal having a spectrum characteristic in which many peaks are observed).

(Embodiment 3)
In the third embodiment, another method for specifying the target frequency band by the second position specifying unit will be described. In the present embodiment, the configuration of the encoding device is the same as that of the encoding device shown in FIG. 1 except for the internal configuration of second layer encoding section 105, and thus the description thereof is omitted.

  FIG. 13 is a block diagram showing a configuration of second layer encoding section 105 of the encoding apparatus according to the present embodiment. Second layer encoding section 105 shown in FIG. 13 employs a configuration including second position specifying section 301 in place of second position specifying section 202 with respect to FIG. The same components as those of second layer encoding section 105 shown in FIG.

  13 includes a first sub-position specifying unit 311-1, a second sub-position specifying unit 311-2,..., A J-th sub-position specifying unit 311-J, and a multiplexing unit. 312.

  The plurality of sub position specifying units (311-1, ..., 311-J) specify different target frequencies in the band specified by the first position specifying unit 201. Specifically, the n-th sub-position specifying unit 311-n includes the first to (n-1) -th sub-position specifying units (311-1,..., From the band specified by the first position specifying unit 201. In the band excluding the target frequency specified by 311-n-1), the nth target frequency is specified.

  FIG. 14 is a diagram illustrating the positions of target frequencies specified by a plurality of sub-position specifying units (311-1,..., 311-J) of the encoding apparatus according to the present embodiment. Here, a case where the first position specifying unit 201 specifies the band 2 and the second position specifying unit 301 specifies the positions of the J target frequencies will be described as an example.

As shown in FIG. 14A, the first sub-position specifying unit 311-1 specifies one target frequency from the target frequency candidates in the band 2 (here, G ₃ ), and the position information of the target frequency Is output to the multiplexing unit 312 and output to the second sub-position specifying unit 311-2.

As shown in FIG. 14 (B), second sub-position specifying section 311-2, from the band 2 of the candidates of the target frequency by the first sub-position specifying section 311-1 except target frequency _{G 3} identified One target frequency is specified (G _{N-1 in this case} ), and the position information of the target frequency is output to the multiplexing unit 312 and output to the third sub-position specifying unit 311-3.

Similarly, as shown in FIG. 14C, the J-th sub-position specifying unit 311 -J starts with the first to J-1 sub-position specifying units (311-1,..., 311-J-1 from the band 2. ) Selects one target frequency from the target frequency candidates excluding the J−1 target frequencies specified (here, G ₅ ), and outputs position information specifying the target frequency to the multiplexing unit 312.

  The multiplexing unit 312 generates the second position information by multiplexing the J pieces of position information input from the sub-position specifying units (311-1,..., 311-J), and the encoding unit 203 and the multiplexing unit 204 Output to. Note that this multiplexing unit 312 is not essential, and J pieces of position information may be directly output to the encoding unit 203 and the multiplexing unit 204.

  As described above, the second position specifying unit 301 can specify J target frequencies and express a plurality of peaks in the band specified by the first position specifying unit 201. The sound quality of signals with strong tonality can be further improved. In addition, since J target frequencies may be determined from the band specified by the first position specifying unit 201, a plurality of targets may be used as compared with the case where J target frequencies are determined from the entire band. The number of frequency combinations can be greatly reduced. Thereby, a low bit rate and a low calculation amount can be realized.

(Embodiment 4)
In Embodiment 4, another encoding method in second layer encoding section 105 will be described. In the present embodiment, the configuration of the encoding device is the same as that of the encoding device shown in FIG. 1 except for the internal configuration of second layer encoding section 105, and thus the description thereof is omitted.

  FIG. 15 is a block diagram showing a configuration of second layer encoding section 105 of another aspect of the encoding apparatus according to the present embodiment. The second layer encoding unit 105 illustrated in FIG. 15 does not include the second position specifying unit 202 illustrated in FIG. 2, and further includes an encoding unit 221 instead of the encoding unit 203 illustrated in FIG. 2. Take.

  The encoding unit 221 determines the second position information so that the quantization distortion that occurs when encoding the error transform coefficient included in the target frequency is minimized. This second position information is stored in the second position information codebook 321.

  FIG. 16 is a block diagram showing a configuration of encoding section 221 shown in FIG. 16 employs a configuration in which a second location information codebook 321 is added to the encoding unit 203 illustrated in FIG. 6 and a search unit 322 is provided instead of the search unit 303. In addition, the same number is attached | subjected to the structure same as the encoding part 203 shown in FIG. 6, and the description is abbreviate | omitted.

  The second position information codebook 321 selects one second position information from the stored second position information candidates according to a control signal from the search unit 322 described later, and outputs the second position information to the target signal configuration unit 301. In the second position information codebook 321 of FIG. 16, the black dots represent the positions of the target frequencies of the respective second position information candidates.

  The target signal constituting unit 301 specifies the target frequency using the first position information input from the first position specifying unit 201 and the second position information selected in the second position information codebook 321, A portion included in the specified target frequency is extracted from the input first layer error conversion coefficient, and the extracted first layer error conversion coefficient is output to the error calculation unit 302 as a target signal.

  Based on the error E input from the error calculation unit 302, the search unit 322 searches for a combination of a shape candidate, a gain candidate, and a second position information candidate that minimizes the error E, and the search result shape information, gain The information and the second position information are output as encoded information to multiplexing section 204 shown in FIG. In addition, the search unit 322 outputs a control signal for selecting and outputting the second position information candidate to the target signal configuration unit 301 to the second position information codebook 321.

  Thus, according to the present embodiment, since the second position information is determined so as to minimize the quantization distortion generated when the error transform coefficient included in the target frequency is encoded, the final quantization is performed. Since the distortion is reduced, the voice quality can be improved.

  In the present embodiment, the second position information codebook 321 shown in FIG. 16 has been described as an example of storing the second position information candidate having a single target frequency as an element. Not limited to this, as shown in FIG. 17, the second position information codebook 321 may store second position information candidates having a plurality of target frequencies as elements. FIG. 17 is a diagram illustrating the encoding unit 221 when the second position information candidates stored in the second position information codebook 321 each have three target frequencies.

  Further, in the present embodiment, the example in which the error calculation unit 302 illustrated in FIG. 16 calculates the error E based on the shape codebook 304 and the gain codebook 305 has been described, but the present invention is not limited thereto, As shown in FIG. 18, the shape codebook 304 may be deleted, and the error E may be calculated based only on the gain codebook 305. 18 is a block diagram showing another configuration of the encoding unit 221 shown in FIG. This configuration corresponds to a case where the signal output from the shape codebook 304 is always “1”. In this case, since the shape is composed of a plurality of pulses and the shape codebook 304 is not required, the search unit 322 searches only the gain codebook 305 and the second position information codebook 321, and gain information of the search result and The second position information is output as encoded information to the multiplexing unit 204 shown in FIG.

  In the present embodiment, the second position information codebook 321 has been described on the premise that the second position information codebook 321 actually secures a storage area and stores the second position information candidates. However, the present invention is not limited to this. Instead, the second position information codebook 321 may generate the second position information candidate according to a predetermined processing procedure. In this case, the second location information codebook 321 does not require a storage area.

(Embodiment 5)
In the fifth embodiment, another band specifying method by the first position specifying unit will be described. In the present embodiment, the configuration of the encoding device is the same as that of the encoding device shown in FIG. 1 except for the internal configuration of second layer encoding section 105, and thus the description thereof is omitted.

  FIG. 19 is a block diagram showing a configuration of second layer encoding section 105 of the encoding apparatus according to the present embodiment. The second layer encoding unit 105 shown in FIG. 19 employs a configuration including a first position specifying unit 231 instead of the first position specifying unit 201 shown in FIG.

  A calculation unit (not shown) performs pitch analysis on the input signal to obtain a pitch period, and calculates a pitch frequency from the reciprocal of the obtained pitch period. Note that the calculation unit may calculate the pitch frequency from the first layer encoded data generated by the encoding process of the first layer encoding unit 102. In this case, since the first layer encoded data is transmitted, it is not necessary to separately transmit information for specifying the pitch frequency. In addition, the calculation unit outputs pitch cycle information for specifying the pitch cycle to the multiplexing unit 106.

  The first position specifying unit 231 specifies a band with a predetermined relatively wide bandwidth based on a pitch frequency input from a calculation unit (not shown), and uses position information of the specified band as first position information. The data is output to the second position specifying unit 202, the encoding unit 203, and the multiplexing unit 204.

FIG. 20 is a diagram showing the position of the band specified by the first position specifying unit 231 shown in FIG. The three bands shown in FIG. 20 are bands in the vicinity of integer multiples of the reference frequencies F _{1 to} F ₃ determined based on the input pitch frequency PF. The reference frequency is a frequency obtained by adding a predetermined value to the pitch frequency PF. As a specific example, here, -1, 0, and 1 are added to PF, and the reference frequencies are F ₁ = PF-1, F ₂ = PF, and F ₃ = PF + 1.

  The reason for setting a band based on an integer multiple of the pitch frequency is that the audio signal has a spectrum peak in the vicinity of an integral multiple of the reciprocal of the pitch period (pitch frequency), particularly in the vowel part having a strong pitch periodicity. This is because there is a characteristic (harmonic structure or harmonics), and a large error is likely to occur near the integral multiple of the pitch frequency in the first layer error conversion coefficient.

  As described above, according to the present embodiment, since the first position specifying unit 231 specifies a band in the vicinity of an integer multiple of the pitch frequency, the target frequency finally specified by the second position specifying unit 202 is the pitch. Since the frequency is close, the voice quality can be improved with a small amount of calculation.

(Embodiment 6)
In the sixth embodiment, the encoding method according to the present invention is applied to an encoding apparatus having a first layer encoding unit that uses a method of substituting an approximate signal due to noise or the like in an encoding process. explain. FIG. 22 is a block diagram showing the main configuration of encoding apparatus 220 according to the present embodiment. 22 includes a first layer encoding unit 2201, a first layer decoding unit 2202, a delay unit 2203, a subtraction unit 104, a frequency domain transform unit 101, and a second layer encoding unit. 105 and a multiplexing unit 106. In the encoding device 220 of FIG. 22, the same components as those of the encoding device 100 shown in FIG.

  First layer encoding section 2201 according to the present embodiment employs a scheme that substitutes the high frequency section with an approximate signal such as noise. Specifically, the high-frequency part that is less perceptually important is represented by an approximate signal, and instead, the bit distribution of the low-frequency part (or low-middle part) that is perceptually important is increased and the original signal of this band is increased. Improve fidelity to. As a result, the overall sound quality is improved. For example, an AMR-WB system (Non-patent Document 3) and a VMR-WB system (Non-Patent Document 4) can be mentioned.

  First layer encoding section 2201 encodes the input signal to generate first layer encoded data, and outputs the first layer encoded data to multiplexing section 106 and first layer decoding section 2202. Details of first layer encoding section 2201 will be described later.

  First layer decoding section 2202 performs a decoding process using the first layer encoded data input from first layer encoding section 2201, generates a first layer decoded signal, and outputs the first layer decoded signal to subtraction section 104. Details of first layer decoding section 2202 will be described later.

  Next, details of first layer encoding section 2201 will be described using FIG. FIG. 23 is a block diagram showing a configuration of first layer encoding section 2201 of encoding apparatus 220. As shown in FIG. 23, first layer encoding section 2201 includes a downsampling section 2210 and a core encoding section 2220.

  The down-sampling unit 2210 down-samples the time-domain input signal, converts it to a desired sampling rate, and outputs the down-sampled time-domain signal to the core encoding unit 2220.

  Core encoding section 2220 performs an encoding process on the output signal of downsampling section 2210, generates first layer encoded data, and outputs the first layer encoded data to first layer decoding section 2202 and multiplexing section 106.

  Next, details of first layer decoding section 2202 will be described using FIG. FIG. 24 is a block diagram showing a configuration of first layer decoding section 2202 of encoding apparatus 220. As shown in FIG. 24, first layer decoding section 2202 includes core decoding section 2230, upsampling section 2240, and high frequency component adding section 2250.

  Core decoding section 2230 performs decoding processing using the first layer encoded data input from core encoding section 2220, generates a decoded signal, outputs the decoded signal to upsampling section 2240, and is obtained by the decoding processing. The decoded LPC coefficient is output to high frequency component adding section 2250.

  The upsampling unit 2240 upsamples the decoded signal output from the core decoding unit 2230, converts the decoded signal into the same sampling rate as the input signal, and outputs the upsampled signal to the high frequency component adding unit 2250.

  The high frequency component adding unit 2250 generates an approximate signal of the high frequency component for the signal up-sampled by the down-sampling unit 2240 by the method described in Non-Patent Document 3 and Non-Patent Document 4, for example, To compensate for high frequencies.

  FIG. 25 is a block diagram showing the main configuration of a decoding apparatus corresponding to the encoding apparatus according to the present embodiment. 25 has the same basic configuration as decoding apparatus 600 shown in FIG. 7, and includes first layer decoding section 2501 instead of first layer decoding section 602. Similarly to the first layer decoding unit 2202 of the encoding device, the first layer decoding unit 2501 includes a core decoding unit, an upsampling unit, and a high frequency component adding unit (not shown). Here, detailed description thereof is omitted.

  A signal that can be generated without additional information by the encoding unit and decoding unit such as a noise signal is passed through a synthesis filter composed of decoded LPC coefficients given by the core decoding unit, and the output signal of the synthesis filter is approximated to a high frequency component Used for signals. At this time, since the high frequency component of the input signal and the high frequency component of the first layer decoded signal have completely different waveforms, the energy of the high frequency component of the error signal obtained by the subtracting unit is higher than the energy of the high frequency component of the input signal. But it will get bigger. As a result, the second layer encoding unit has a problem that it is easy to select a band arranged in a high frequency part having low auditory importance.

  According to the present embodiment, as described above, in encoding process of first layer encoding section 2201, encoding apparatus 220 that uses a method of substituting an approximate signal due to noise or the like in the high-frequency section is set in advance. Even if the energy of the high frequency part of the error signal (or error conversion coefficient) is increased by selecting the band from the low frequency part whose frequency is lower than the reference frequency, the second low frequency part having high auditory sensitivity is selected. Since it can be selected as an encoding target of the layer encoding unit, sound quality can be improved.

  In the present embodiment, the configuration in which the information related to the high frequency band is not sent to the decoding unit has been described as an example. However, the present invention is not limited to this, and for example, as in Non-Patent Document 5, May be encoded at a lower bit rate than the low-frequency part and sent to the decoding part.

  In the encoding device 220 shown in FIG. 22, the subtractor 104 is configured to take a difference between signals in the time domain, but the subtractor may be configured to take a difference between transform coefficients in the frequency domain. In this case, the frequency domain transform unit 101 is disposed between the delay unit 2203 and the subtraction unit 104 to obtain an input transform coefficient, and the frequency domain transform unit 101 is newly added between the first layer decoding unit 2202 and the subtraction unit 104. Thus, the first layer decoding transform coefficient is obtained. The subtracting unit 104 is configured to take the difference between the input transform coefficient and the first layer decoded transform coefficient and directly give the error transform coefficient to the second layer encoding unit. According to this configuration, it is possible to perform subtraction processing suitable for each band, such as making a difference in one band and not taking a difference in another band, thereby further improving sound quality.

(Embodiment 7)
In Embodiment 7, a case will be described in which the encoding method according to the present invention is applied to an encoding device and a decoding device having other configurations. FIG. 26 is a block diagram showing the main configuration of encoding apparatus 260 according to the present embodiment.

  26 has a configuration in which a weight filter unit 2601 is added to the encoding device 220 illustrated in FIG. In addition, in the encoding apparatus 260 of FIG. 26, the same number is attached | subjected about the component similar to FIG. 22, and the description is abbreviate | omitted.

ここで、α（ｉ）は復号ＬＰＣ係数、ＮＰはＬＰＣ係数の次数、そしてγはスペクトル平坦化（白色化）の程度を制御するパラメータであり、０≦γ≦１の範囲の値をとる。γが大きいほど平坦化の程度が大きくなり、例えばγには０.９２を用いる。 The weighting filter unit 2601 performs filter processing for perceptually weighting the error signal input from the subtraction unit 104, and outputs the filtered signal to the frequency domain conversion unit 101. The weighting filter unit 2601 has a spectral characteristic opposite to the spectral envelope of the input signal, and changes the spectrum of the input signal to flattening (whitening) or a spectral characteristic close thereto. For example, the weight filter W (z) is configured as in the following equation (9) using the decoded LPC coefficient obtained by the first layer decoding unit 2202.

Here, α (i) is a decoded LPC coefficient, NP is the order of the LPC coefficient, and γ is a parameter that controls the degree of spectrum flattening (whitening), and takes a value in the range of 0 ≦ γ ≦ 1. As γ increases, the degree of flattening increases. For example, 0.92 is used for γ.

  A decoding device 270 illustrated in FIG. 27 has a configuration in which a synthesis filter unit 2701 is added to the decoding device 250 illustrated in FIG. In the decoding device 270 of FIG. 27, the same components as those of FIG.

ここで、α（ｉ）は復号ＬＰＣ係数、ＮＰはＬＰＣ係数の次数、そしてγはスペクトル平坦化（白色化）の程度を制御するパラメータであり、０≦γ≦１の範囲の値をとる。γが大きいほど平坦化の程度が大きくなり、例えばγには０.９２を用いる。 The synthesis filter unit 2701 performs a filter process for restoring the flattened spectrum characteristic to the original characteristic on the signal input from the time domain conversion unit 606, and outputs the signal after the filter process to the addition unit 604. To do. The synthesis filter unit 2701 has a spectrum characteristic opposite to that of the weighting filter expressed by Expression (9), that is, a characteristic similar to the spectrum envelope of the input signal. The synthesis filter B (z) is expressed by the following equation (10) using the equation (9).

Here, α (i) is a decoded LPC coefficient, NP is the order of the LPC coefficient, and γ is a parameter that controls the degree of spectrum flattening (whitening), and takes a value in the range of 0 ≦ γ ≦ 1. As γ increases, the degree of flattening increases. For example, 0.92 is used for γ.

  In general, in the encoding device and the decoding device as described above, the spectral envelope of the audio signal appears such that the energy in the low frequency part is larger than the energy in the high frequency part, so the encoding distortion of the signal before passing through the synthesis filter Even when the low frequency region and the high frequency region are equivalent, the coding distortion of the low frequency region becomes large after passing through the synthesis filter. When audio signals are compressed and transferred at a low bit rate, the coding distortion cannot be reduced sufficiently. Therefore, the energy of the low frequency part of the coding distortion is affected by the influence of the synthesis filter part of the decoding part as described above. Is increased, and there is a problem that quality degradation in the low frequency region is likely to appear.

  According to the encoding method of the present embodiment, since the target frequency is determined from the low frequency part whose frequency is lower than the reference frequency, the low frequency part is easily selected as the encoding target of second layer encoding part 105. As a result, the coding distortion in the low frequency band can be reduced. That is, according to the present embodiment, even if the low frequency region is emphasized by the synthesis filter, the encoding distortion of the low frequency region is hardly perceived, so that an effect of improving the sound quality can be obtained.

  In the present embodiment, the subtracting unit 104 of the encoding device 260 is configured to take a difference between signals in the time domain, but the present invention is not limited to this, and is configured to take a difference between transform coefficients in the frequency domain. Also good. Specifically, the weight filter unit 2601 and the frequency domain transform unit 101 are arranged between the delay unit 2203 and the subtraction unit 104 to obtain an input transform coefficient, and the weight between the first layer decoding unit 2202 and the subtraction unit 104 is obtained. A filter unit 2601 and a frequency domain transform unit 101 are newly added to obtain first layer decoded transform coefficients. Then, the subtraction unit 104 is configured to take the difference between the input transform coefficient and the first layer decoding transform coefficient and directly give the error transform coefficient to the second layer coding unit 105. According to this configuration, it is possible to perform subtraction processing suitable for each band, such as making a difference in one band and not taking a difference in another band, thereby further improving sound quality.

  Further, in the present embodiment, the case where the number of layers of the encoding device 220 is 2 has been described as an example. However, the present invention is not limited to this, for example, as in the encoding device 280 illustrated in FIG. The encoding layer may have a configuration with two or more layers.

  FIG. 28 is a block diagram showing the main configuration of encoding apparatus 280. For the coding apparatus 100 shown in FIG. 1, a second layer decoding unit 2801, a third layer coding unit 2802, a third layer decoding unit 2803, a fourth layer coding unit 2804, and two additions A configuration is adopted in which a device 2805 is added and three subtracting units 104 are provided.

  The third layer encoding unit 2802 and the fourth layer encoding unit 2804 shown in FIG. 28 have the same configuration as that of the second layer encoding unit 105 shown in FIG. The decoding unit 2801 and the third layer decoding unit 2803 have the same configuration as the first layer decoding unit 103 shown in FIG. 1 and perform the same operation. Here, the position of the band in each layer encoding unit will be described with reference to FIG.

  As an example of band arrangement in each layer encoding unit, FIG. 29A shows a band position in the second layer encoding unit, FIG. 29B shows a band position in the third layer encoding unit, and FIG. The band positions in the fourth layer encoding unit are shown, and the number of bands is 4 respectively.

  More specifically, in the second layer encoding unit 105, four bands are arranged so as not to exceed the reference frequency Fx (L2) of layer 2, and in the third layer encoding unit 2802, the reference frequency Fx of layer 3 is arranged. Four bands are arranged so as not to exceed (L3), and the fourth layer encoding unit 2804 arranges bands so as not to exceed the reference frequency Fx (L4) of layer 4. There is a relationship of Fx (L2) <Fx (L3) <Fx (L4) between the reference frequencies of the layers. That is, in layer 2 where the bit rate is low, the band to be encoded is determined from the low frequency part with high perceptual sensitivity, and the band including the high frequency part is included in the higher layer where the bit rate is high. The band to be encoded is determined from the inside.

  By adopting such a configuration, it is possible to achieve higher sound quality of the audio signal in order to emphasize the low frequency band in the lower layer and cover a wider band in the higher layer.

  FIG. 30 is a block diagram showing a main configuration of decoding apparatus 300 corresponding to encoding apparatus 280 shown in FIG. 30 has a configuration in which a third layer decoding section 3001, a fourth layer decoding section 3002, and two adders 604 are added to decoding apparatus 600 shown in FIG. Note that the third layer decoding section 3001 and the fourth layer decoding section 3002 have the same configuration as the second layer decoding section 603 of the decoding apparatus 600 shown in FIG. Detailed description thereof is omitted.

  As another example of the band arrangement in each layer encoding unit, FIG. 31A shows the positions of four bands in the second layer encoding unit 105, and FIG. 31B shows the six bands in the third layer encoding unit 2802. FIG. 31C shows the positions of the eight bands in the fourth layer encoding unit 2804.

  In FIG. 31, in each layer encoding unit, each band is arranged at equal intervals, and in the lower layer as shown in FIG. 31A, only the band arranged in the low band part is the target of encoding, and FIG. 31B or FIG. As the higher layer becomes, the band to be encoded increases.

  According to such a configuration, when the bands are arranged at equal intervals in each layer and the band to be encoded is selected in the lower layer, the number of bands arranged in the low frequency part which is a selection candidate is small. Therefore, the calculation amount and the bit rate can be reduced.

(Embodiment 8)
The eighth embodiment of the present invention is different from the first embodiment only in the operation of the first position specifying unit, and in order to show this, the first position specifying unit according to the present embodiment has a number “801”. Is attached. The first position specifying unit 801 divides the entire band into a plurality of partial bands in advance when specifying the band that can be taken by the target frequency to be encoded, and uses a predetermined bandwidth and a predetermined step size in each partial band. Perform a search. Then, the first position specifying unit 801 combines the bands in the partial bands obtained by the search so that the target frequency to be encoded can be taken.

  The operation of the first position specifying unit 801 according to the present embodiment will be described with reference to FIG. FIG. 32 illustrates a case where the number of partial bands N = 2, partial band 1 is set so as to cover the low frequency part, and partial band 2 is set so as to cover the high frequency part. In the partial band 1, one band is selected from a plurality of bands set in advance to a predetermined bandwidth (position information of this band is referred to as first partial band position information). Similarly, in the partial band 2, one band is selected from a plurality of bands set in advance to a predetermined bandwidth (position information of this band is referred to as second partial band position information).

  Next, the first position specifying unit 801 combines the band selected in the partial band 1 and the band selected in the partial band 2 to form a combined band. This combined band becomes a band specified by the first position specifying unit 801, and then the second position specifying unit 202 specifies the second position information based on the combined band. For example, when the band selected in the partial band 1 is the band 2 and the band selected in the partial band 2 is the band 4, the first position specifying unit 801 displays the two bands as shown in the lower part of FIG. To be a band that can be taken by the frequency band to be encoded.

  FIG. 33 is a block diagram illustrating a configuration of the first position specifying unit 801 corresponding to the case where the number of partial bands is N. In FIG. 33, the first layer error conversion coefficient input from the subtracting unit 104 is provided to each of the partial band 1 specifying unit 811-1 to the partial band N specifying unit 811-N. Each partial band n specifying unit 811-n (n = 1 to N) selects one band from predetermined partial bands n, and indicates information on the position of the selected band (nth partial band position information). ) To the first position information configuration unit 812.

  The first location information configuration unit 812 configures the first location information using the nth partial band location information (n = 1 to N) input from each of the partial bandwidth n identification units 811-n, and The position information is output to the second position specifying unit 202, the encoding unit 203, and the multiplexing unit 204.

  FIG. 34 is a diagram illustrating a state in which the first position information configuring unit 812 configures the first position information. In this figure, the first position information configuration unit 812 configures the first position information by arranging the first partial band position information (A1 bit) to the Nth partial band position information (AN bit) in order. Here, the bit length An of each n-th partial band position information is determined by the number of candidate bands included in each partial band n, and may have different values.

  FIG. 35 is a diagram illustrating a state in which the first layer decoding error transform coefficient is obtained using the first position information and the second position information in the decoding process according to the present embodiment. Here, a case where the number of partial bands is 2 will be described as an example. In the following description, the names and numbers of the constituent elements constituting second layer decoding section 603 according to Embodiment 1 are used.

  The placement unit 704 rearranges the shape candidate after gain candidate multiplication input from the multiplication unit 703 using the second position information. Next, the placement unit 704 further rearranges the shape candidates after rearrangement using the second position information into the partial band 1 and the partial band 2 using the first position information. Arrangement section 704 outputs the signal obtained in this way as the first layer decoding error transform coefficient.

  According to the present embodiment, since the first position specifying unit selects one band from each partial band, it is possible to arrange at least one decoded spectrum in the partial band. As a result, a plurality of bands whose sound quality is to be improved can be set in advance as compared with the embodiment in which one band is determined from all the bands. For example, this embodiment is effective when it is desired to simultaneously improve the quality of both the low frequency region and the high frequency region.

  Further, according to the present embodiment, the subjective quality of the decoded signal can be improved even when encoding at a low bit rate in the lower layer (first layer in the present embodiment). The configuration using the CELP method for the lower layer is an example. Since the CELP method is an encoding method based on waveform matching, the encoding is performed so that the quantization distortion in the low frequency region where the energy is large is smaller than that in the high frequency region. As a result, the spectrum in the high frequency region is attenuated, and this is perceived as a feeling of being full (absence of a band feeling). On the other hand, CELP encoding is a low bit rate encoding method, and thus low-band quantization distortion cannot be sufficiently suppressed, and the quantization distortion is perceived as noise. In this embodiment, since the band to be encoded is selected from each of the low-frequency part and the high-frequency part, two different deterioration factors such as the noise feeling of the low-frequency part and the feeling of the high-frequency part are simultaneously eliminated. It becomes possible to improve subjective quality.

  Further, according to the present embodiment, a band selected from the low band and a band selected from the high band are combined to form a combined band, and the spectrum shape is determined in the combined band. Select a spectrum shape that emphasizes the low range for frames that require quality improvement in the lower range than the low range, and select a spectrum shape that emphasizes the high range for frames that require higher quality improvement than the low range. The adaptive process of selecting can be performed, and the subjective quality can be improved. For example, when the shape of the spectrum is represented by pulses, many pulses are placed in the low frequency range in frames that require quality improvement in the low frequency range rather than in the high frequency range, and high in frames that require quality improvement in the higher frequency range than the low frequency range. Many pulses can be arranged in a region, and subjective quality can be improved by such adaptive processing.

  As a variation of the present embodiment, as shown in FIG. 36, a fixed band may always be selected in a specific partial band. In the example shown in FIG. 36, the band 4 is always selected in the partial band 2, and this is a part of the combined band. As a result, similarly to the effect of the present embodiment, it is possible to set in advance the band for which the sound quality is to be improved, and for example, the partial band position information of the partial band 2 becomes unnecessary. The number of bits for representing the first position information as shown can be made smaller.

  FIG. 36 shows an example in which a fixed range is always selected in the high frequency band (partial band 2). However, the present invention is not limited to this, and the fixed range is always fixed in the low frequency band (partial band 1). A range may be selected, or a fixed range may always be selected in a partial band of the middle region not shown in FIG.

  As a variation of the present embodiment, as shown in FIG. 37, the bandwidths of candidate bands set in each partial band may be different. In FIG. 37, the case where the bandwidth of the partial band set in the partial band 2 is shorter than the candidate band set in the partial band 1 is illustrated.

  The embodiment of the present invention has been described above.

  The band arrangement in each layer encoding unit is not limited to the example described above in the present invention. For example, the bandwidth of each band is narrowed in the lower layer and the bandwidth of each band is widened in the higher layer. It may be configured.

  In the above embodiments, the band of the current frame may be selected in association with the band selected in the past frame. For example, the band of the current frame may be determined from bands positioned in the vicinity of the band selected in the previous frame. Alternatively, the current frame band candidate may be rearranged in the vicinity of the band selected in the previous frame, and the current frame band may be determined from the rearranged band candidates. Further, the range information may be transmitted at a rate of once every several frames, and the range represented by the range information transmitted in the past may be used in a frame where the range information is not transmitted (intermittent transmission of band information).

  In each of the above embodiments, the band of the current layer may be selected in association with the band selected in the lower layer. For example, the band of the current layer may be determined from bands positioned in the vicinity of the band selected in the lower layer. The current layer band candidate may be rearranged in the vicinity of the band selected in the lower layer, and the current layer band may be determined from the rearranged band candidates. Further, the range information may be transmitted at a rate of once every several frames, and the range represented by the range information transmitted in the past may be used in a frame where the range information is not transmitted (intermittent transmission of band information).

  In the present invention, the number of scalable encoding layers is not limited.

  Moreover, in the said embodiment, although the audio | voice signal is assumed as a decoded signal, this invention is not limited to this, For example, an audio signal etc. may be sufficient.

  Further, although cases have been described with the above embodiment as examples where the present invention is configured by hardware, the present invention can also be realized by software.

  Each functional block used in the description of each of the above embodiments 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 a part or all of them. The name used here is LSI, but it may also be called IC, system LSI, super LSI, or ultra LSI 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 and setting of circuit cells inside the LSI may be used.

  Furthermore, if integrated circuit technology comes out to replace LSI's as a result of the advancement of semiconductor technology or a derivative other technology, it is naturally also possible to carry out function block integration using this technology. Biotechnology can be applied.

  The present invention is suitable for use in an encoding device, a decoding device, or the like used in a scalable encoding communication system.

101 Frequency domain transform unit 102, 2201 First layer encoding unit 103, 2202 First layer decoding unit 104 Subtracting unit 105 Second layer encoding unit 106, 204 Multiplexing unit 201, 801 First position specifying unit 202 Second position Identification unit 203, 221 Encoding unit 301 Target signal configuration unit 302 Error calculation unit 303 Search unit 304 Shape codebook 305 Gain codebook 311-1, ..., 311-J Sub-position identification unit 321 Second position information codebook 601 Separation Unit 602, 2501 first layer decoding unit 603, 2502 second layer decoding unit 604 addition unit 605 switching unit 606 time domain conversion unit 607 post filter 701 shape codebook 702 gain codebook 703 multiplication unit 704 arrangement unit 2203 delay unit 2210 down Sampling unit 2220 Core encoding unit 22 0 core decoding unit 2240 upsampling unit 2250 high frequency component adding unit 2601 weight filter unit 2701 synthesis filter unit 2801 2nd layer decoding unit 2802 3rd layer encoding unit 2803 3rd layer decoding unit 2804 4th layer encoding unit 3001 1st 3 layer decoding section 3002 4th layer decoding section

Claims (8)

First layer encoding means for performing encoding processing on the input speech signal to generate first layer encoded data;
First layer decoding means for generating a first layer decoded signal by performing decoding processing using the first layer encoded data;
First layer error conversion coefficient calculating means for calculating a first layer error conversion coefficient by converting a first layer error signal, which is an error between the input audio signal and the first layer decoded signal, into a frequency domain;
Second layer encoding means for performing encoding processing on the first layer error transform coefficient to generate second layer encoded data,
The second layer encoding means includes
A first band is selected based on the energy level of the first layer error conversion coefficient in the band candidate from a plurality of band candidates arranged with a predetermined bandwidth and a step size narrower than the bandwidth. Band selection means for generating first position information indicating the position of the selected first band;
Second position information indicating the positions of the plurality of identified pulses by identifying the positions of the plurality of pulses from the pulse candidate positions set with a step size smaller than the step size in the selected first band. Pulse position specifying means for generating
Encoded data generation means for generating the second layer encoded data using the first position information and the second position information,
Speech encoding device. The pulse position specifying means specifies the position of the pulse based on the magnitude of the energy of the first layer error conversion coefficient;
The speech encoding apparatus according to claim 1. The second layer encoding means further comprises gain encoding means for generating gain information indicating the amplitude of the pulse at the pulse position based on the first layer error transform coefficient,
The encoded data generation means generates second layer encoded data by further using the gain information.
The speech encoding apparatus according to claim 1. The band selecting means selects the first band from a low frequency part lower than a preset reference frequency,
The speech encoding apparatus according to claim 1. First layer encoded data obtained by performing encoding processing on the input speech signal in the speech encoding device, and a first layer decoded signal obtained by decoding the first layer encoded data in the speech encoding device Obtained by converting the first layer error signal, which is an error with the input audio signal, into the frequency domain, calculating a first layer error conversion coefficient, and performing an encoding process on the first layer error conversion coefficient Receiving means for receiving second layer encoded data;
First layer decoding means for decoding the first layer encoded data and generating the first layer decoded signal;
Second layer decoding means for decoding the second layer encoded data to generate first layer decoded error transform coefficients;
Time domain transforming means for transforming the first layer decoding error transform coefficient into the time domain to generate a first layer decoded error signal;
Adding means for adding the first layer decoded signal and the first layer decoded error signal to generate a decoded signal;
The second layer decoding means includes
Decoding the second layer encoded data, first position information indicating a position of a first band having a predetermined bandwidth, and second position information indicating positions of a plurality of pulses in the first band Generate
Using the first position information and the second position information to identify the positions of the plurality of pulses to generate the first layer decoding error transform coefficient;
Speech decoding device. The second layer decoding means includes
Decoding the second layer encoded data to generate gain information indicating the amplitude of the pulse, and further generating the first layer decoding error transform coefficient using the gain information;
The speech decoding apparatus according to claim 5. A first layer encoding step of performing encoding processing on the input speech signal to generate first layer encoded data;
A first layer decoding step of generating a first layer decoded signal by performing a decoding process using the first layer encoded data;
A first layer error conversion coefficient calculation step of converting a first layer error signal, which is an error between the input audio signal and the first layer decoded signal, to a frequency domain and calculating a first layer error conversion coefficient;
A second layer encoding step of performing encoding processing on the first layer error transform coefficient to generate second layer encoded data; and
The second layer encoding step includes:
A first band is selected based on the energy level of the first layer error conversion coefficient in the band candidate from a plurality of band candidates arranged with a predetermined bandwidth and a step size narrower than the bandwidth. A band selection step for generating first position information indicating the position of the selected first band;
Second position information indicating the positions of the plurality of identified pulses by identifying the positions of the plurality of pulses from the pulse candidate positions set with a step size smaller than the step size in the selected first band. A pulse locating step for generating
An encoded data generation step for generating the second layer encoded data using the first position information and the second position information;
Speech encoding method. First layer encoded data obtained by performing encoding processing on an input speech signal in the speech encoding method, and a first layer decoded signal obtained by decoding the first layer encoded data in the speech encoding method Obtained by converting the first layer error signal, which is an error with the input audio signal, into the frequency domain, calculating a first layer error conversion coefficient, and performing an encoding process on the first layer error conversion coefficient Receiving a second layer encoded data; and
A first layer decoding step of decoding the first layer encoded data to generate a first layer decoded signal;
A second layer decoding step of decoding the second layer encoded data to generate a first layer decoding error transform coefficient;
A time domain transforming step of transforming the first layer decoding error transform coefficients into a time domain to generate a first layer decoded error signal;
An adding step of adding the first layer decoded signal and the first layer decoded error signal to generate a decoded signal;
The second layer decoding step includes
Decoding the second layer encoded data, first position information indicating a position of a first band having a predetermined bandwidth, and second position information indicating positions of a plurality of pulses in the first band Generate
Identifying the positions of the plurality of pulses using the first position information and the second position information, and generating the first layer decoding error transform coefficient;
Speech decoding method.

Images