JP4308229B2 - Encoding device and decoding device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an encoding device and a decoding device capable of performing encoding with high compressibility and decoding wide-band frequency spectra. <P>SOLUTION: The encoding device (200) includes an MDCT unit (202) that transforms an input signal in a time domain into a frequency spectrum, a BWE encoding unit (204) that generates extension data which prescribes a higher frequency spectrum at a higher frequency than a lower frequency spectrum included in the converted frequency spectrum by reference to the low frequency spectrum, and an encoded data stream generating unit (205) that encodes to output the lower frequency spectrum and the extension information. The BWE encoding unit (204) generates as the extension information a first parameter which prescribes a lower subband which is to be copied as the higher frequency spectrum from among a plurality of the lower subbands which form the lower frequency spectrum obtained by the MDCT unit (202) and a second parameter which prescribes a gain of the lower subband after being copied. <P>COPYRIGHT: (C)2007,JPO&amp;INPIT

Description

  The present invention compresses an audio signal such as an audio signal or a music signal by encoding a signal, which is converted from the time domain to the frequency domain, using a method such as orthogonal transformation with a smaller number of encoded sequences. The present invention relates to an encoding device that performs decoding, and a decoding device that decompresses information using an encoded sequence as an input.

  To date, a great number of audio signal encoding and decoding methods have been developed. Particularly in recent years, among them, IS13818-7 internationally standardized by ISO / IEC has been recognized and is evaluated as a high-quality and high-efficiency encoding method. This encoding method is called AAC. In recent years, the AAC has also been adopted for standardization called MPEG4, and a method called MPEG4-AAC having several extended functions has been formulated for the IS13818-7. An example of the encoding process is described in INFORMATIVE PART.

  Here, an audio encoding apparatus using a conventional encoding method will be described with reference to FIG. FIG. 13 is a block diagram showing a configuration of a conventional encoding apparatus 100. As shown in FIG. The encoding apparatus 100 includes a spectrum amplification unit 101, a spectrum quantization unit 102, a Huffman encoding unit 103, and an encoded sequence transfer unit 104. An audio discrete signal sequence on the time axis obtained by sampling an analog audio signal at a predetermined frequency is cut out by a predetermined number of samples at a constant time interval, and after passing through a time frequency converter (not shown), After being converted to data, it is given to the spectrum amplifying unit 101 as an input signal of the encoding apparatus 100. The spectrum amplifying unit 101 amplifies the spectrum included in the band with one gain for each predetermined band. The spectrum quantization unit 102 quantizes the above amplified spectrum with a determined conversion formula. In the case of the AAC method, quantization is performed by rounding frequency spectrum information expressed by a floating-point number to an integer value. The Huffman coding unit 103 collects several pieces of the quantized spectrum information and performs Huffman coding, and then specifies information for specifying the gain and quantization conversion formula for each predetermined band in the spectrum amplification unit 101. Huffman encoding is performed, and the code is sent to the encoding transfer unit 104. The Huffman-encoded encoded sequence is transferred from the encoded sequence transfer unit 104 to a decoding device via a transmission path or a recording medium, and is reproduced as an audio signal on the time axis by the decoding device. The conventional encoding device operates in this way.

  However, in the conventional encoding device 100, the compression capability of the information amount is left to the performance of the Huffman encoding unit 103 and the like, and when encoding with a high compression rate, that is, with a small amount of information, The spectrum amplifying unit 101 needs to reduce the gain sufficiently, and the Huffman encoding unit 103 needs to perform encoding so that the quantized spectrum sequence obtained by the spectrum quantization unit 102 has a small amount of information. If encoding is performed so that the amount of information is small according to such a method, the frequency band of the reproduced speech and music is narrowed. For this reason, there is a problem that a feeling of hearing is unavoidable and sufficient sound quality cannot be secured.

  In view of the above problems, an object of the present invention is to provide an encoding device and a decoding device capable of encoding an audio signal with a high compression rate in the encoding device and decoding wideband frequency spectrum information in the decoding device. To do.

  In order to solve the above-described problems, an encoding apparatus according to the present invention is an apparatus that encodes an input signal, a time-frequency conversion unit that converts an input signal on a time axis into a frequency spectrum, and the converted frequency. By referring to the first frequency spectrum included in the spectrum, the band extension means for generating extension information for specifying the second frequency spectrum at a frequency higher than the first frequency spectrum, and the time frequency conversion means obtained Encoding means for encoding and outputting the first frequency spectrum and the extension information obtained by the band extension means, wherein the band extension means constitutes the first frequency spectrum obtained by the time frequency conversion means A first parameter for specifying a partial spectrum to be copied as the second frequency spectrum from among the plurality of partial spectra And a second parameter specifying the gain of the partial spectrum after replication, and generates as the extension information.

  The decoding device of the present invention is a device for decoding an encoded signal, and the encoded signal includes a first frequency spectrum and a second frequency spectrum at a frequency higher than the first frequency spectrum. Decoding information for generating the first frequency spectrum and the extension information by decoding the encoded signal; and decoding means for decoding the encoded signal; , A frequency obtained by synthesizing the generated second frequency spectrum and the first frequency spectrum, band expanding means for generating the second frequency spectrum from the first frequency spectrum and the first and second parameters. Frequency time conversion means for converting a spectrum into a signal on a time axis, and the band extending means includes a plurality of components constituting the first frequency spectrum. Replicating a partial spectrum specified by the first parameter of the partial spectrum, determining a gain of the partial spectrum after the replication by the second parameter, and generating the obtained partial spectrum as the second frequency spectrum; Features.

  In order to solve the above-described problems, an encoding apparatus according to the present invention is an apparatus that encodes an input signal, a time-frequency conversion unit that converts an input signal on a time axis into a frequency spectrum, and the converted frequency. By referring to the first frequency spectrum included in the spectrum, the band extension means for generating extension information for specifying the second frequency spectrum at a frequency higher than the first frequency spectrum, and the time frequency conversion means obtained Encoding means for encoding and outputting the first frequency spectrum and the extension information obtained by the band extension means, wherein the band extension means constitutes the first frequency spectrum obtained by the time frequency conversion means A first parameter for specifying a partial spectrum to be copied as the second frequency spectrum from among the plurality of partial spectra And a second parameter specifying the gain of the partial spectrum after replication, and generates as the extension information.

  As described above, according to the encoding apparatus of the present invention, it is possible to provide a wide-band audio encoded sequence at a low bit rate. In the encoding device of the present invention, the low frequency component encodes the fine structure of the frequency using a compression technique such as Huffman coding, but the high frequency component does not encode the fine structure, Since only the information that substitutes and replicates the low-frequency spectrum as the high-frequency spectrum is mainly encoded, there is an effect that the amount of information consumed by the encoded sequence representing the high-frequency component can be minimized.

  The decoding device of the present invention is a device for decoding an encoded signal, and the encoded signal includes a first frequency spectrum and a second frequency spectrum at a frequency higher than the first frequency spectrum. Decoding information for generating the first frequency spectrum and the extension information by decoding the encoded signal; and decoding means for decoding the encoded signal; , A frequency obtained by synthesizing the generated second frequency spectrum and the first frequency spectrum, band expanding means for generating the second frequency spectrum from the first frequency spectrum and the first and second parameters. Frequency time conversion means for converting a spectrum into a signal on a time axis, and the band extending means includes a plurality of components constituting the first frequency spectrum. Replicating a partial spectrum specified by the first parameter of the partial spectrum, determining a gain of the partial spectrum after the replication by the second parameter, and generating the obtained partial spectrum as the second frequency spectrum; Features.

  Therefore, according to the decoding apparatus of the present invention, in the decoding process, the high frequency component is generated by applying a process such as gain adjustment to the duplication of the low frequency component, so that an encoded sequence with a small amount of data is generated. Therefore, there is an effect that a wide reproduction sound can be obtained.

  The band extending unit adds a noise spectrum to the generated second frequency spectrum, and the frequency time converting unit synthesizes the second frequency spectrum to which the noise spectrum is added and the first frequency spectrum. The frequency spectrum obtained in this way may be converted into a signal on the time axis.

  Therefore, according to the decoding apparatus of the present invention, the noise spectrum is added to the second frequency spectrum, and gain adjustment is performed on the copied low frequency component, so that the tonality of the second frequency spectrum is extremely reduced. There is an effect that it is possible to increase the bandwidth without increasing the frequency.

  Hereinafter, an encoding device and a decoding device according to an embodiment of the present invention will be described with reference to the drawings (FIGS. 1 to 12).

(Embodiment 1)
First, the encoding device will be described. FIG. 1 is a block diagram showing a configuration of coding apparatus 200 according to Embodiment 1 of the present invention. Encoding apparatus 200 divides a low-frequency part spectrum into subbands having a constant frequency width, and includes information for specifying a subband to be copied to a high-frequency part in an acoustic encoded bitstream for output. The apparatus includes a preprocessing unit 201, an MDCT unit 202, a quantization unit 203, a BWE encoding unit 204, and a coded sequence generation unit 205.

  In consideration of the fact that the audio quality of the input audio signal sequence changes due to quantization distortion due to quantization accompanying encoding and decoding, the preprocessing unit 201 gives priority to temporal resolution and is finer than 2048 samples. It is determined whether it is better to perform quantization on a frame basis (SHORT window) or to perform quantization with a 2048 sample size (LONG window).

  The MDCT unit 202 performs a modified discrete cosine transform (MDCT transform) on the time axis audio discrete signal sequence that is the output of the preprocessing unit 201, and outputs a frequency spectrum on the frequency axis. The quantization unit 203 quantizes the low frequency part of the frequency spectrum output from the MDCT unit 202, encodes it to Huffman, and outputs the result.

  The BWE encoding unit 204 receives the MDCT coefficient obtained by the MDCT unit 202 as an input, divides the input low-frequency part spectrum into subbands having a certain frequency width, and outputs the high frequency spectrum output from the MDCT part 202 Based on the part, the low band sub-band to be copied to the high band instead of the high band spectrum is specified.

  The BWE encoding unit 204 generates extended frequency spectrum information indicating the specified low frequency band subband for each high frequency band subband, quantizes the generated extended frequency spectrum information, if necessary, and performs Huffman coding To output an extended audio encoded sequence. The encoded sequence generation unit 205 determines the low band audio encoded sequence that is output from the quantizing unit 203 and the extended audio encoded sequence that is output from the BWE encoding unit 204, respectively, according to the AAC standard. The audio encoded stream portion and the extended audio encoded sequence portion of the obtained acoustic encoded stream are recorded and output to the outside.

  Hereinafter, the operation of the encoding apparatus 200 configured as described above will be described. First, for example, an audio discrete signal sequence sampled at a sampling frequency of 44.1 kHz is input to the preprocessing unit 201 by 2048 samples per frame. The audio signal sequence of one frame is not limited to 2048 samples, but in order to facilitate the description of the encoding apparatus described later, the case of 2048 samples will be mentioned. Based on the input audio signal sequence, the preprocessing unit 201 determines whether to encode the input audio signal sequence using the LONG window or the SHORT window. Hereinafter, a case will be described in which the preprocessing unit 201 determines to perform quantization in the LONG window.

  The audio discrete signal sequence output from the pre-processing unit 201 is converted from a discrete signal on the time axis into frequency spectrum information and output at certain time intervals by time frequency conversion of the MDCT unit 202. MDCT conversion is generally used as time frequency conversion. Generally, any one of 128, 256, 512, 1024, and 2048 samples is used as the time interval. In the case of MDCT conversion, the number of discrete signals on the time axis and the number of samples of frequency spectrum information after conversion can be handled in the same number. MDCT conversion is a technique apparent to those skilled in the art. Here, it is assumed that an audio signal of 2048 samples output from the preprocessing unit 201 is input to the MDCT unit 202 and subjected to MDCT conversion. Also, the MDCT unit 202 performs MDCT conversion using the past frame (2048 samples) and the newly input frame (2048 samples), and outputs 2048 sample MDCT coefficients. The MDCT conversion is given by a general formula (Formula 1) or the like.

  In general, in the process of encoding, the frequency spectrum information obtained as described above is completely reversible or expressed by an irreversible code such as a Huffman code corresponding to information compression to generate a coded sequence. Here, the quantizing unit 203 has low-frequency MDCT coefficients from 0th to 1023th in the low-frequency half of 2048 samples of MDCT coefficients arranged in order of frequency from low-frequency components to high-frequency components. Entered. The quantization unit 203 quantizes the input MDCT coefficient using a quantization method such as an AAC method, and generates a low-frequency audio encoded sequence. In general, in the quantization method such as the AAC method, the number of MDCT coefficients to be quantized is not defined. Therefore, the quantization unit 203 may quantize all of the input low frequency part MDCT coefficients (1024 coefficients) or only a part of them.

  Here, the quantization unit 203 quantizes and encodes a total of (maxline) coefficients from the 0th to the (maxline-1) th among the MDCT coefficients. Here, maxline is the upper limit frequency of the MDCT coefficient quantized and encoded by the conventional encoding device. On the other hand, all MDCT coefficients (2048 coefficients) output from the MDCT section 202 are input to the BWE encoding section 204.

  Hereinafter, the extended audio coded sequence generation processing in the BWE encoding unit 204 shown in FIG. 1 will be described in more detail with reference to FIGS. 2 (a) to 2 (c). FIG. 2A is a diagram illustrating an MDCT coefficient sequence output by the MDCT unit 202. FIG. 2B is a diagram illustrating the 0th to (maxline-1) -th MDCT coefficients encoded by the quantization unit 203 among the MDCT coefficients illustrated in FIG. FIG. 2C is a diagram illustrating an example of a method for generating an extended audio encoded sequence in the BWE encoding unit 204 illustrated in FIG.

  2A to 2C, the horizontal axis indicates the frequency, and the numbers of the MDCT coefficients are assigned from the 0th to the 2047th in order from the low frequency to the high frequency. The vertical axis represents the value of the MDCT coefficient. In addition, in the figure, the frequency spectrum is shown as a waveform that is continuous in the frequency direction, but it is not actually a continuous waveform. As shown in FIG. 2A, the 2048 MDCT coefficients output from the MDCT unit 202 can represent an original sound sampled for a certain period of time in a frequency band that is half the sampling frequency in the maximum bandwidth.

  In general, in the conventional coding apparatus, only the low-frequency MDCT coefficients up to, for example, maxline, among the MDCT coefficients shown in FIG. Often transmitted to For this reason, in the BWE encoding unit 204, the extended frequency spectrum information that represents the high frequency part above the maxline is not the MDCT coefficient itself shown in FIG. 2A, but represents the high frequency part MDCT coefficient instead of the high frequency part MDCT coefficient. Is generated. That is, in the BWE encoding unit 204, the 0th to (maxline-1) th of the MDCT coefficients are encoded in advance by the quantization unit 203. Therefore, as shown in FIG. The purpose is to encode MDCT coefficients from the () th to (targetline-1) th.

  First, the BWE encoding unit 204 assumes a high-frequency range (specifically, a frequency range from maxline to targetline) to be reproduced as an audio signal in the decoding apparatus, and the assumed range is sub-frequency constant. Divide into bands. Further, the BWE encoding unit 204 has a part or all of the low-frequency part composed of the MDCT coefficients from the 0th to the (maxline-1) -th among the inputted MDCT coefficients having the same frequency width as that of the high-frequency subband. In the high frequency band consisting of the (maxline) th to 2047th MDCT coefficients, the low frequency band subband that can be substituted for each subband is specified. For example, a low-frequency subband that minimizes an energy difference between the high-frequency subband and the low-frequency subband is identified as a low-frequency subband that can be substituted for each high-frequency subband. Alternatively, the low frequency subband having the closest position on the frequency axis of the MDCT coefficient having the maximum absolute value may be specified in each subband of the high frequency region and the low frequency region.

  In the case of the BWE encoding unit 204 in FIG. 2C, it is assumed that there is a relationship of (Equation 2) among startline, targetline, endline, and sbw representing the numbers of MDCT coefficients.

  Here, the shiftlen may be a preset value, or the shiftlen may be calculated according to a change in the input MDCT coefficient, and information indicating the value may be encoded by the BWE encoding unit 204.

  In FIG. 2 (c), when the high-frequency part is divided into eight subband MDCT coefficient sequences h0 to h7 with a frequency width composed of sbw MDCT coefficients, in the low-frequency part, from startline to endline, In the example, four subband MDCT coefficient subbands each consisting of sbw samples can be configured, and each of them is A, B, C, and D. Here, for convenience, the startline to the endline are divided into four subbands, and the maxline to the targetline are divided into eight subbands. However, these numbers and the number of samples per subband are not necessarily limited to these values. . The BWE encoding unit 204 specifies and specifies the low frequency subbands A, B, C, and D having the same frequency width sbw that replace the MDCT coefficient sequences of the high frequency subbands h0 to h7 having the frequency width sbw. Extended frequency spectrum information indicating the low frequency alternative subband is generated and encoded. Here, “replacement” refers to copying a part of the obtained MDCT coefficients, in this case, MDCT coefficients of the low-frequency subbands A to D as MDCT coefficients of the high-frequency subbands h0 to h7. The alternative meaning further includes controlling the gain with respect to the alternative MDCT coefficient.

  In the case of the BWE encoding 204, the amount of information necessary to express which low-frequency subband is substituted for the high-frequency subband is at most 2 bits for each of the high-frequency subbands h0 to h7. It is. This is because it is only necessary to identify one of the four types of low-frequency subbands A to D per high-frequency subband. In this way, the BWE encoding unit 204 encodes extended frequency spectrum information indicating which of the high frequency subbands h0 to h7 is replaced by the low frequency subbands A to D, and uses the code string as an extended audio code. Generate a sequence.

  Further, the BWE encoding unit 204 adjusts the amplitude of the generated extended audio encoded sequence. FIG. 3A is a waveform diagram showing an MDCT coefficient sequence of the original sound. FIG. 3B is a waveform diagram showing an MDCT coefficient sequence generated by substitution by the BWE encoding unit 204. FIG. 3C is a waveform diagram showing an MDCT coefficient sequence when gain control is performed on the MDCT coefficient sequence shown in FIG. As shown in FIG. 3A, the BWE encoding unit 204 divides the high-frequency MDCT coefficient from maxline to targetline into a plurality of bands, and encodes gain information for each band. The band division method from maxline to targetline for encoding gain information may be the same division method as that for the high frequency subbands h0 to h7 shown in FIG. 2, or may be another division method. Here, the case of the division method similar to FIG. 2 will be described with reference to FIG.

  As shown in FIG. 3A, the MDCT coefficients of the original sound included in the high frequency subband h0 are x (0), x (1),..., X (sbw-1), Assume that the MDCT coefficients by substitution in the high frequency subband h0 of 3 (b) are r (0), r (1), ..., r (sbw-1). Also, the MDCT coefficients of the subband h0 in FIG. 3C are y (0), y (1),..., Y (sbw-1), and between the arrays x, r, and y, (Equation 3) A gain g0 is obtained and encoded.

  Similarly, gain information is calculated and encoded for the high frequency subbands h1 to h7. These gain information g0 to g7 are also encoded into the extended audio encoded sequence with a predetermined number of bits.

  The extended audio encoded sequence encoded in this way is described in an acoustic encoded bit stream that is an output of the encoding apparatus 200, as schematically shown in FIG. FIG. 4A is a diagram illustrating an example of a normal audio encoded bitstream. FIG. 4B is a diagram illustrating an example of an acoustic coded bit stream output by the coding apparatus 200 according to the present embodiment. FIG. 4C is a diagram illustrating an example of the extended audio encoded sequence described in the extended audio encoded sequence unit illustrated in FIG. As shown in FIG. 4A, when the audio encoded bit stream is formed for each frame like the stream 1, the encoding apparatus 200 uses the stream 2 shown in FIG. A part of each frame (for example, the hatched portion in the figure) is used as the extended audio encoded sequence portion.

  This extended audio coding sequence portion is, for example, a data_stream_element area described in MPEG-2 AAC and MPEG-4 AAC. This data_stream_element is a preliminary area for describing data for expansion when the function of the conventional coding method is expanded, and no matter what data is recorded in this area, This is a region that is not recognized as an audio encoded sequence by the encoding device. Further, for example, an area in which meaningless data such as “0” is filled in order to make the data length of the audio encoded sequence uniform, for example, an area such as a fill element in MPEG-2 AAC and MPEG-4 AAC. If the extended audio coded sequence is described in such an area in the acoustic coded bitstream, even if the acoustic coded bitstream of the present invention is decoded using a conventional decoding device, the extended audio code An audio signal having the same band as that of the prior art can be reproduced without causing noise due to reproduction of the sequence as an audio signal.

  Also, as shown in FIG. 4C, an item indicating whether or not the extended audio coded sequence uses the low frequency subbands A to D divided by the same method as the extended audio coded sequence of the immediately preceding frame. And items representing the MDCT coefficients of the high frequency sub-bands h0 to h7. In the items representing the MDCT coefficients of the high frequency subbands h0 to h7, data indicating the specified low frequency subbands A to D and gain information thereof are described, respectively. The item indicating whether or not to use the same low frequency subbands A to D as the extended audio coded sequence of the immediately preceding frame includes, for example, one of the low frequency subbands A to D divided in the same manner as the immediately preceding frame. Is used to replace the MDCT coefficients of the high frequency subbands h0 to h7, otherwise, that is, that is, the low frequency subbands A to D newly divided by a different division method from the previous frame In the case of using one of these, a 1-bit value indicated by “0” is described.

  Among the low frequency subbands A to D, 2-bit data specifying one of the four low frequency subbands A to D is described in the item indicating the specified low frequency subband. The gain information is described in 4 bits, for example. In this way, when substituting the MDCT coefficients of the high frequency subbands h0 to h7 using the low frequency subbands A to D divided in the same manner as the previous frame, the high frequency MDCT coefficient of one frame is changed. 1 + 8 * (2 + 4) = 49-bit extended audio coded sequence. Also, in a frame that uses the same low frequency subbands A to D as the previous frame, the extended audio coded sequence can be represented by only one bit of a value “1” indicating that, for example.

  In this way, when the audio signal encoding method by the encoding apparatus 200 of the present invention is applied to the conventional encoding method, the high frequency region is expressed using an extended audio encoded sequence with a small amount of data, Rich wideband audio playback sound can be obtained.

<Decryption device>
On the other hand, in the decoding process, the input audio coded sequence is decoded, frequency spectrum information is obtained, and the frequency spectrum is subjected to frequency time conversion to reproduce an audio signal on the time axis.

  FIG. 5 is a block diagram illustrating a configuration of a decoding apparatus 600 that decodes the acoustic encoded bitstream output from the encoding apparatus 200 of FIG. The decoding apparatus 600 is a decoding apparatus that decodes an acoustic encoded bitstream including an extended audio encoded sequence and outputs wideband frequency spectrum information, and includes an encoded sequence separation unit 601 and an inverse quantization unit 602. , An IMDCT (Inversed Modified Discrete Cosine Transform) unit 603, a noise generation unit 604, a BWE decoding unit 605, and an extended IMDCT unit 606.

  The coded sequence separation unit 601 separates an audio coded sequence representing a low frequency portion and an extended audio coded sequence representing a high frequency portion from the input acoustic coded bitstream, and separates the encoded audio coding. The sequence is output to the inverse quantization unit 602, and the separated extended audio encoded sequence is output to the BWE decoding unit 605. The inverse quantization unit 602 inversely quantizes the audio coded sequence separated from the acoustic coded bitstream, and outputs a low frequency part MDCT coefficient.

  Note that the inverse quantization unit 602 may receive both the audio encoded sequence and the extended audio encoded sequence as inputs. In addition, if the AAC method is used as the quantization method in the quantization unit 203, the inverse quantization unit 602 restores the MDCT coefficient using the AAC method inverse quantization. As a result, the inverse quantization unit 602 restores and outputs the low-frequency MDCT coefficients from the 0th to the (maxline-1) th.

  The IMDCT unit 603 performs frequency-time conversion on the low-frequency part MDCT coefficient output from the inverse quantization unit 602 using IMDCT, and outputs a low-frequency part audio signal on the time axis. That is, when the output of the inverse quantization unit 602 is used as the input of the IMDCT unit 603, an audio output of 1024 samples per frame is obtained. Here, the IMDCT unit 603 performs IMDCT calculation of 1024 samples. A general expression of the IMDCT calculation is given by (Equation 4) and the like.

  On the other hand, the extended audio coded sequence separated from the acoustic coded bitstream by the coded sequence separation unit 601 is output to the BWE decoding unit 605. In addition, the 0th to (maxline-1) th low frequency MDCT coefficients and the output of the noise generation unit 604, which are the outputs of the inverse quantization unit 602, are input to the BWE decoding unit 605. The details of the operation of the BWE decoding unit 605 will be described later. Based on the extended frequency spectrum information obtained by decoding the separated extended audio encoded sequence, high frequencies corresponding to the (maxline) th to 2047th The partial MDCT coefficients are decoded and dequantized, and added to the 0th to (maxline-1) th low band MDCT coefficients obtained by the dequantization unit 602, corresponding to the 0th to 2047th Output a wideband MDCT coefficient. The extended IMDCT unit 606 obtains a wideband output audio signal of 2048 samples per frame by performing an IMDCT operation with twice as many samples as the IMDCT unit 603.

  Hereinafter, a more detailed operation of the BWE decoding unit 605 will be described. The BWE decoding unit 605 restores the MDCT coefficients from the maxline to the targetline using the 0th to (maxline-1) MDCT coefficients obtained by the inverse quantization unit 602 and the extended audio coded sequence. Startline, endline, maxline, targetline, sbw, and shiftlen are all the same values as those used in the BWE encoding unit 204 on the encoding apparatus 200 side. As shown in FIG. 4C, the extended audio coded sequence indicates which subband of the low frequency subbands A to D is substituted for the MDCT coefficients of the high frequency subbands h0 to h7. Since the information is encoded, the MDCT coefficients of the high frequency subbands h0 to h7 are respectively replaced with the MDCT coefficients of the designated low frequency subbands A to D based on the information.

  As a result, the BWE decoding unit 605 obtains the 0th to (targetline) MDCT coefficients. Further, the BWE decoding unit 605 performs gain control based on the gain information in the extended audio encoded sequence. As shown in FIG. 3B, the BWE decoding unit 605 generates an MDCT coefficient sequence that is substituted by the low-frequency subbands A to D in the high-frequency subbands h0 to h7 from maxline to targetline. Further, the BWE decoding unit 605 has gain information obtained from the extended audio coded sequence when the alternative MDCT coefficients in the high frequency subband h0 are r (0), r (1),..., R (sbw-1). Is g0 for the high frequency subband h0, the MDCT coefficient sequence subjected to gain control shown in FIG. 3C can be obtained from the relational expression given by (Equation 5). That is, if the MDCT coefficients in the high frequency sub-band h0 are y (0), y (1), ..., y (sbw-1), the value of the i-th MDCT coefficient y (i) subjected to gain control Is expressed by Equation 5 below.

  Similarly, the high-frequency subbands h1 to h7 can also obtain a gain-controlled MDCT coefficient sequence by multiplying the MDCT coefficient sequence by the gain information g1 to g7 for each high-frequency subband. Furthermore, the noise generation unit 604 generates white noise, pink noise, or noise that is a combination of all or part of the low-frequency MDCT coefficient sequence, and the generated noise is converted into a gain-controlled MDCT coefficient sequence. Append. At that time, it is also possible to correct the energy of the spectrum synthesized by the noise to be added and the spectrum replicated from the low frequency to the energy of the spectrum represented by (Equation 5).

  In the first embodiment, encoding of gain information to be multiplied by the replaced MDCT coefficient as described in (Equation 5) has been described, but the gain information is not a relative gain, but energy of the MDCT coefficient, You may encode and decode using absolute values, such as an average amplitude.

  By using the BWE decoding unit 605 configured in this way, even when an extended audio coded sequence represented by a small amount of data as shown in FIG. A wide-band audio playback sound can be obtained.

  Although the encoding apparatus 200 and the decoding apparatus 600 according to the AAC scheme have been described above, the encoding apparatus and the decoding apparatus according to the present invention are not limited to this, and use other encoding schemes. May be.

  In the encoding apparatus 200, the MDCT coefficient sequence from the 0th to the 2047th is output from the MDCT unit 202 to the BWE encoding unit 204. However, in the BWE encoding unit 204, the quantization unit 203 further performs quantization. An MDCT coefficient including quantization distortion obtained by inverse quantization of the converted MDCT coefficient may also be input. Further, the BWE encoding unit 204 receives, as input, the MDCT coefficient sequence obtained by dequantizing the output of the quantization unit 203 for the 0th to (maxline-1) th low-frequency parts. The output of the MDCT unit 202 may be input to the (targetline-1) th high frequency part.

  In the first embodiment, it has been described that the extended frequency spectrum information is quantized and encoded according to the case. However, information to be encoded (extended frequency spectrum information) is encoded using variable length coding such as a Huffman code. It goes without saying that what is expressed may be used as the extended audio coded sequence. Correspondingly, the decoding apparatus may decode a variable length code such as a Huffman code without dequantizing the extended audio coded sequence.

  In the first embodiment, the encoding method and the decoding method of the present invention are applied to MPEG-2 AAC and MPEG-4 AAC. However, the present invention is not limited to this, and MPEG-1 Audio and MPEG- 2 You may apply to other encoding systems, such as Audio. When used for MPEG-1 Audio or MPEG-2 Audio, the extended audio coded sequence is applied to ancillary_data described in the standard.

  In the first embodiment, the high frequency subband is replaced with the frequency spectrum of the low frequency subband within the range of the frequency spectrum (MDCT coefficient) obtained by time-frequency conversion of the input audio signal. However, the present invention is not limited to this, and it may be replaced by a region exceeding the upper limit of the frequency spectrum output by time-frequency conversion. However, in this case, it is not possible to specify a low-frequency subband used for substitution based on a high-frequency spectrum (MDCT coefficient) representing the original sound.

(Embodiment 2)
In the second embodiment of the present invention, the difference from the first embodiment is that the BWE encoding unit 204 of the first embodiment uses four low frequency band MDCT coefficient sequences from startline to endline. Although it is divided into subbands, in the BWE encoding unit of the second embodiment, the same band from startline to endline is divided into seven subbands from A to G allowing duplication.

  Note that the encoding apparatus and decoding apparatus in Embodiment 2 have basically the same configuration as encoding apparatus 200 and decoding apparatus 600 in Embodiment 1, and in the encoding apparatus, BWE encoding section 701 is used. However, only the processing of the BWE decoding unit 702 is different in this processing and the decoding device. Therefore, in the second embodiment, only the BWE encoding unit 701 and the BWE decoding unit 702 are described by changing the reference codes, and each of the components already described with respect to the encoding device 200 and the decoding device 600 in the first embodiment is described. Elements are denoted by the same reference numerals and description thereof is omitted. In the following embodiments, only points different from those already described will be described, and the portions already described will be omitted.

  Below, the BWE encoding part 701 of Embodiment 2 is demonstrated using FIG. FIG. 6 is a diagram illustrating a method of generating extended frequency spectrum information by the BWE encoding unit 701 according to the second embodiment. In the figure, the low-frequency subbands E, F, and G are the low-frequency subbands A, B, C, and D of the low-frequency subbands A, B, C, and D divided as in the first embodiment. This is a subband obtained by shifting sbw / 2 on the band side.

  Here, the low-frequency subbands A, B, and C are shifted by sbw / 2 in the high-frequency direction, but the band division method that allows overlapping, the frequency width of the shift, the number of divisions, etc. are not necessarily limited to this. It is not limited. The BWE encoding unit 701 replaces the MDCT coefficient sequence of the high frequency subbands h0 to h7 with information specifying one of the seven low frequency subbands A to G, for each of the high frequency subbands h0 to h7. Is generated and encoded, and output as an extended audio encoded sequence.

  On the other hand, in the decoding apparatus according to the second embodiment, the extended audio encoded by the encoding apparatus according to the second embodiment (including the BWE encoding unit 701 instead of the BWE encoding unit 204 in the encoding apparatus 200). Using the coded sequence as input, information specifying which subband MDCT coefficients of the low frequency subbands A to G are replaced with the high frequency subbands h0 to h7 is decoded, and the high frequency subbands h0 to h7 are decoded. The MDCT coefficients of the low frequency subbands A to G are replaced with the MDCT coefficients of the low frequency band.

  In addition, when information specifying any one of the low frequency sub-bands A to G is expressed using, for example, 3-bit code information, integer values from “0” to “6” are low as the code information. Assuming that the subbands A to G are represented, when code information having a code information value of “7” is created, the decoding apparatus may perform control that does not substitute any of A to G. Good. Although the case where 3 bits of information is used as the code information and the value of the code information is “7” is described here, the number of bits of the code information and the value of the code information may be other values.

  The gain control and / or the noise superposition used in the first embodiment are similarly used in the second embodiment of the present invention. By using the encoding device and the decoding device created in this way, it is possible to obtain a wide-band reproduced sound using an extended audio encoded sequence that is not large as an information amount.

(Embodiment 3)
The third embodiment is different from the second embodiment in that the BWE encoding unit 701 of the second embodiment allows the low-frequency MDCT coefficients from the startline to the endline to be overlapped in the frequency axis direction from A to G. However, in the BWE encoding unit of the third embodiment, the band from startline to endline is divided into seven subbands A to G, and within the low band subband. Are defined by reversing the order of the MDCT coefficients and by reversing the signs of the MDCT coefficients in the low-frequency subband.

  Also in the third embodiment, as in the second embodiment, the configuration differs from the encoding device 200 and the decoding device 600 of the first embodiment in that the BWE encoding unit 801 in the encoding device and the BWE in the decoding device are different. Only the decoding unit 802 is provided. Below, the BWE encoding part of this Embodiment 3 is demonstrated using FIG.

  FIG. 7 is a diagram illustrating an extended frequency spectrum information generation method by the BWE encoding unit 801 according to the third embodiment. FIG. 7A is a diagram showing subbands of the low frequency band and the high frequency band divided in the same manner as in the second embodiment. FIG. 7B is a diagram illustrating an example of the MDCT coefficient sequence of the low frequency subband A. FIG. 7C is a diagram illustrating an example of the MDCT coefficient sequence of the subband As obtained by reversing the order of the MDCT coefficient sequence of the low frequency subband A.

  FIG. 7D shows a subband Ar obtained by inverting the sign of the MDCT coefficient sequence of the low frequency subband A. For example, the MDCT coefficient sequence of the low frequency subband A is represented by (p0, p1,..., PN). In this case, for example, p0 represents the value of the 0th MDCT coefficient of subband A. The MDCT coefficient sequence of the subband As obtained by inverting the order of the MDCT coefficient sequence of the subband A in the frequency direction is (pN, p (N−1),..., P0). The MDCT coefficients of the subband Ar obtained by inverting the sign of the MDCT coefficient sequence of the low frequency subband A are represented by (−p0, −p1,..., −pN). Similarly, not only the subband A but also the subbands B to G define subbands Bs to Gs whose order is reversed and subbands Br to Gr obtained by reversing the negative sign.

  In this way, in the BWE encoding unit 801 in the third embodiment, for each of the high frequency subbands h0 to h7, one of the seven low frequency subbands A to G or the seven low frequency subbands. MDCT of high-frequency subbands h0 to h7, each of seven low-frequency subbands As to Gs and low-frequency subbands Ar to Gr obtained by inverting the order and signs of the MDCT coefficient sequences of A to G Identify one that replaces the coefficient.

  The BWE encoding unit 801 encodes information for representing a high-frequency part MDCT coefficient sequence using the specified low-frequency subband, and generates an extended audio encoded sequence as illustrated in FIG. In this case, as the extended frequency spectrum information, for each high frequency subband, the order of the information specifying the low frequency subband substituting the MDCT coefficient of the high frequency subband and the MDCT coefficient of the specified low frequency subband are reversed. The information indicating whether or not to perform and the information indicating whether or not to reverse the sign of the identified low frequency subband MDCT coefficient are encoded.

  On the other hand, in the decoding apparatus according to the third embodiment, the high frequency sub-bands h0 to h7 are set to be low by using the extended audio coded sequence encoded by the coding apparatus according to the third embodiment as described above. Decodes extended frequency spectrum information indicating which MDCT coefficients in the subbands A to G are replaced, whether to reverse the order of the MDCT coefficients, and whether to reverse the sign of the MDCT coefficients Turn into. Next, according to the decoded extended frequency spectrum information, the MDCT coefficients of the identified low-frequency subbands A to G are reversed by changing the order of the MDCT coefficients or by inverting the positive / negative sign. Generate MDCT coefficients of h0 to h7.

  Furthermore, the third embodiment includes not only an extension of the order and positive and negative signs of the low-frequency subband MDCT coefficients but also an alternative by filtering the low-frequency subband MDCT coefficients. Note that the filter processing is, for example, an IIR filter, an FIR filter, and the like, and is a technique that is obvious to those skilled in the art. When performing such filter processing, the encoding device side encodes the filter coefficient in the extended audio encoded sequence, so that the decoding device side can specify the MDCT coefficient of the specified low-frequency subband. Further, the IIR filter or FIR filter indicated by the decoded filter coefficient is applied, and the high-frequency subband can be replaced using the filtered MDCT coefficient.

  The gain control used in the first embodiment can be similarly used in the third embodiment. By using the encoding apparatus and decoding apparatus configured as described above, it is possible to obtain a wide-band reproduced sound using an extended audio encoded sequence that is not large in amount of information.

(Embodiment 4)
The fourth embodiment differs from the third embodiment in that the decoding apparatus of the fourth embodiment uses the MDCT coefficients of the high frequency subbands h0 to h7 as the specified low frequency subbands A to G. Instead of using only the MDCT coefficients of the low frequency subbands A to G, the MDCT coefficients generated by the noise generation unit are used for replacement. Therefore, the decoding apparatus according to the fourth embodiment is different from the decoding apparatus 600 according to the first embodiment only in the configuration of the noise generation unit 901 and the BWE decoding unit 902.

  Hereinafter, with regard to the decoding process of the extended audio coded sequence in the decoding apparatus according to the fourth embodiment, for example, a case where the high frequency subband h0 to be BWE decoded is replaced by using the low frequency subband A is illustrated. 8 will be used for explanation. FIG. 8A is a diagram illustrating an example of the MDCT coefficients of the low frequency subband A specified for the high frequency subband h0. FIG. 8B is a diagram illustrating an example of the same number of MDCT coefficients as the low frequency subband A generated by the noise generation unit 901. FIG. 8C illustrates a high frequency band generated using the MDCT coefficient of the low frequency sub-band A illustrated in FIG. 8A and the MDCT coefficient generated by the noise generation unit 901 illustrated in FIG. It is a figure which shows an example of the MDCT coefficient which substitutes subband h0. Here, the MDCT coefficient sequence of the low-frequency subband A is A = (p0, p1,..., PN).

  In addition, it is assumed that the noise generation unit 901 obtains the same number N of noise signal MDCT coefficient sequences as M = (n0, n1,. The BWE decoding unit 902 substitutes the MDCT coefficient of the high frequency subband h0 by adjusting the MDCT coefficient sequence A of the low frequency subband A and the noise signal MDCT coefficient sequence M using the weighting coefficients α and β. An MDCT coefficient sequence A ′ is generated. The substitution coefficient sequence A ′ is expressed by the following mathematical formula (Formula 6).

  Note that the weighting factors α and β may be preset values in the decoding apparatus according to the fourth embodiment, or the control information indicating the values of the weighting factors α and β is extended audio coding on the encoding device side. It is also possible to use what is encoded in the sequence and restored in the decoding apparatus.

  Although the subband h0 output from the BWE decoding unit 902 has been described as an example here, the same processing is performed for the other high frequency subbands h1 to h7. Moreover, although the low-frequency subband A has been described as an example of the low-frequency subband to be replaced, another low-frequency subband obtained from the inverse quantization unit may be used, and the processing in that case is the same. The weighting coefficients α and β may take values such that when one is “0”, the other is “1”, or “α + β” may be “1”.

  In this case, for example, when α = 0, the energy ratio between the MDCT coefficient sequence of the high frequency subband and the MDCT coefficient sequence of the noise information is obtained, and the obtained energy ratio is expanded as gain information for the MDCT coefficient sequence of the noise information. It shall be encoded into an audio encoding sequence. Furthermore, a value representing the ratio between the weighting factor α and the weighting factor β may be encoded. Further, when all MDCT coefficients of one low-frequency subband copied by the BWE decoding unit 902 are “0”, control is performed such as setting the value of β to “1” regardless of the value of α. It may be done.

  The noise generation unit 901 may have a configuration in which a table prepared in advance is stored therein, and values in the table are output as a noise signal MDCT coefficient sequence, or may be obtained by MDCT conversion of a noise signal in the time domain. The configuration may be such that the generated noise signal MDCT coefficient sequence is generated every frame. Further, gain control is performed for a noise signal in the time domain in the time domain, and a noise signal MDCT coefficient sequence is output using all or part of the MDCT coefficient sequence obtained by MDCT conversion of the gain-controlled signal sequence. But you can.

  In particular, when an MDCT coefficient sequence obtained by controlling the gain of a noise signal in the time domain and performing MDCT conversion is used, an effect of suppressing the pre-echo of the reproduced sound can be expected. At that time, the gain control information for gain control in the time domain for the noise signal is encoded in advance on the encoding device side in the fourth embodiment, and the decoding device side decodes and uses it. Also good. When the decoding apparatus configured in this way is used, the MDCT coefficient of the noise signal can be obtained even when the MDCT coefficient of the high frequency subband to be BWE decoded cannot be sufficiently expressed by the MDCT coefficient of the low frequency subband. By using it, it is expected that the bandwidth can be increased without extremely increasing tonality.

(Embodiment 5)
The fifth embodiment is different from the fourth embodiment in that the function is expanded so that a plurality of time frames can be controlled together. The operations of BWE encoding section 1001 and BWE decoding section 1002 in the encoding apparatus and decoding apparatus according to Embodiment 5 of the present invention will be described with reference to FIGS.

  FIG. 9A shows the MDCT coefficient of one frame at time t0. FIG. 9B shows the MDCT coefficient of the next frame at time t1. FIG. 9C shows the MDCT coefficient of the next frame at time t2. Times t0, t1, and t2 are continuous times, and are time synchronized with the frame. In the first to fourth embodiments, the extended audio coded sequence is generated at each of the times t0, t1, and t2. In the coding apparatus according to the fifth embodiment, the extended audio of a plurality of consecutive frames is used. An encoded sequence is generated in common. Although the figure shows a case where the number of consecutive frames is three, any number of consecutive frames may be used.

  In FIG. 4C of the first embodiment, whether or not the low frequency subbands A to D divided by the same method as the extended audio encoded sequence of the immediately preceding frame are used at the head of the extended audio encoded sequence. In the same manner as this, the BWE encoding unit 1001 of the fifth embodiment uses the same extended audio encoded sequence as that of the immediately preceding frame at the beginning of the extended audio encoded sequence of each frame. An item indicating whether or not to perform is provided. Hereinafter, for example, a case will be described in which, in each frame at time t0, t1, and t2, the high frequency subband of each frame is decoded using the extended audio encoded sequence of time t0 frame.

  The decoding apparatus according to the fifth embodiment performs BWE decoding of each frame with an extended audio coded sequence generated in common for a plurality of consecutive frames as an input. For example, when the high frequency subband h0 in the time t0 frame is replaced with the low frequency subband C in the same time t0 frame, the BWE decoding unit 1002 also converts the high frequency subband h0 in the time t1 frame into the time t1 frame. Decoding is performed using the low frequency subband C, and similarly, the high frequency subband h0 in the time t2 frame is also decoded using the low frequency subband C in the time t2 frame.

  The BWE decoding unit 1002 performs the same processing for the other high frequency subbands h1 to h7. If the encoding device and the decoding device configured in this way are used, the extended audio encoded sequence is entirely included in the audio encoded bitstream for a plurality of frames using the same extended audio encoded sequence. The occupied area can be kept small, and more efficient encoding and decoding can be realized.

  Hereinafter, another example of the encoding device and the decoding device according to Embodiment 5 will be described with reference to FIG. In this example, the difference from the above example is that the BWE encoding unit 1101 converts the high frequency part MDCT coefficient sequence decoded using the same extended audio encoded sequence in a plurality of consecutive frames with a gain different for each frame. The gain information for gain control is encoded into an extended audio encoded sequence.

  FIG. 10 is also a diagram showing an MDCT coefficient sequence in a plurality of consecutive frames at times t0, t1, and t2, as in FIG. In another encoding apparatus according to Embodiment 5, the relative value of the gain of the high frequency MDCT coefficient that is BWE decoded in a plurality of frames is generated in the extended audio encoded sequence. For example, the average amplitude of the MDCT coefficient in the BWE-decoded band (high band from maxline to targetline) is set to G0, G1, and G2 for the time t0, t1, and t2 frames, respectively.

  First, a frame serving as a reference is determined among the frames at time t0, t1, and t2. The reference frame may be determined in advance, for example, at the first time t0 frame. For example, a frame that gives the maximum average amplitude is determined as a reference, and the maximum average amplitude is given. Information indicating the position of the frame may be separately encoded in the extended audio encoded sequence.

  Here, for example, it is assumed that the average amplitude G0 in the time t0 frame is the maximum average amplitude in consecutive frames in which the high frequency MDCT coefficient sequence is decoded using the same extended audio encoded sequence. In this case, the high frequency average amplitude in the time t1 frame is expressed as G1 / G0 with respect to the reference time t0 frame, and the high frequency average amplitude in the time t2 frame is G2 / Represented by G0. The BWE encoding unit 1101 quantizes these high band average amplitude relative values G1 / G0, G2 / G0, etc., and encodes them into an extended audio encoded sequence.

  On the other hand, in another decoding apparatus according to the fifth embodiment, BWE decoding section 1102 receives an extended audio coded sequence as an input, specifies a frame serving as a reference from the extended audio coded sequence, decodes it, or is determined in advance. And the average amplitude value of the reference frame is decoded. Further, the relative average amplitude value of the high frequency MDCT coefficient sequence to be BWE decoded with respect to the reference frame is decoded, and the high frequency MDCT coefficient sequence of each frame decoded according to the common extended audio encoded sequence is gain controlled. To do.

  As described above, according to the BWE decoding unit 1102 illustrated in FIG. 10, the average amplitude of the MDCT coefficient can be easily corrected for a plurality of frames decoded using a common extended audio coded sequence. Can do. As a result, it is possible to encode and decode an audio encoded sequence that can reproduce a wideband audio signal that is more faithful to the original sound with a small amount of data.

(Embodiment 6)
The sixth embodiment is different from the fifth embodiment in that the encoding apparatus and decoding apparatus of the fifth embodiment use a polyphase QMF (Quadrature Mirror Filter) filter for the audio signal on the time axis. Thus, it is a point of conversion and inverse conversion into a time-frequency signal representing a time change of the frequency spectrum.

  For example, out of one frame and 1024 samples of an audio signal sampled at a sampling frequency of 44.1 kHz, 32 consecutive samples are frequency-converted about every 0.73 msec to obtain a frequency spectrum consisting of 32 samples. In one frame and 1024 samples, a total of 32 frequency spectra having a time difference of about 0.73 msec are obtained.

  This frequency spectrum represents the reproduction band from 0 kHz to a maximum of 22.05 kHz with 32 samples each. Of this frequency spectrum, a waveform obtained by connecting spectral data values of the same frequency in the time direction is a time-frequency signal that is the output of the QMF filter. The encoding apparatus according to the present embodiment quantizes, for example, the low-frequency part 0th to 15th time frequency signals out of the time frequency signals output from the QMF filter in the same manner as the conventional encoding apparatus. And variable length encoding.

  On the other hand, for the 16th to 31st time frequency signals of the high frequency part, one of the low frequency part 0th to 15th time frequency signals substituting each other is specified, and the specified low frequency part 0th to 15th is specified. An extended time frequency signal including information indicating the th time frequency signal and gain information for adjusting the amplitude of the identified low frequency frequency signal is generated.

  Here, for example, when filters having different processes or characteristics are used according to the parameters, parameters for specifying the processing contents or characteristics of the filters are described in the extended time frequency signal. Next, the encoding device includes a low-frequency audio encoded sequence obtained by quantizing and variable-length encoding the low-frequency time frequency signal, and a high-frequency code obtained by variable-length encoding the extended time frequency signal. The encoded sequence is described in an audio encoded bitstream and output.

  FIG. 11 is a block diagram illustrating a configuration of a decoding apparatus 1200 that decodes a wideband time-frequency signal from an audio encoded bitstream encoded using a QMF filter. The decoding apparatus 1200 includes an encoded sequence obtained by variable-length encoding an extended time frequency signal representing a high frequency part time frequency signal, and an encoding obtained by quantizing and encoding the low frequency signal A decoding device that decodes a wideband time-frequency signal from an input audio encoded bitstream including a sequence, and includes a nuclear decoding unit 1201, an extended decoding unit 1202, and a spectrum addition unit 1203.

  The nuclear decoding unit 1201 decodes the input audio encoded bitstream, and separates the quantized low frequency signal and the extended time frequency signal representing the high frequency signal. Further, the nuclear decoding unit 1201 performs inverse quantization on the low frequency frequency signal separated from the audio encoded bit stream and outputs the result to the spectrum adding unit 1203. The spectrum adding unit 1203 adds the low frequency time frequency signal decoded and inverse quantized by the nuclear decoding unit 1201 and the high frequency time frequency signal generated by the extended decoding unit 1202 to obtain a total reproduction band For example, a time frequency signal having a reproduction band of 0 kHz to 22.05 kHz is output. This output time frequency signal is converted into an audio signal on the time axis by, for example, a QMF inverse conversion filter (not shown) in the subsequent stage, and further converted into audible sounds such as voice and music by a speaker in the subsequent stage.

  The extended decoding unit 1202 receives the low frequency time frequency signal decoded by the nuclear decoding unit 1201 and the extended time frequency signal, and converts the high frequency time frequency signal based on the separated extended time frequency signal. A processing unit that identifies a low frequency time frequency signal to be substituted and copies it to a high frequency part, further adjusts the amplitude thereof to generate a high frequency time frequency signal, and further includes a substitution control unit 1204 and a gain adjustment unit 1205. Is provided.

  The substitution control unit 1204 identifies, for example, one of the 0th to 15th low-frequency time frequency signals that substitutes for the 16th high-frequency time frequency signal according to the decoded extended time-frequency signal. The low frequency time frequency signal is copied as the 16th high frequency time frequency signal. The gain adjusting unit 1205 amplifies the low-frequency time frequency signal copied as the 16th high-frequency time frequency signal in the high-frequency part according to the gain information described in the extended time frequency signal, and adjusts the amplitude. The extended decoding unit 1202 further performs the above processing by the substitution control unit 1204 and the gain adjustment unit 1205 for each of the 17th to 31st high frequency signals. If 4 bits are used to specify one of the 0th to 15th low frequency time frequency signals, and 4 bits are used for gain information for adjusting the amplitude of the copied low frequency time frequency signal, The 16th to 31st high frequency signals can be represented by (4 + 4) * 32 = 256 bits at most.

  FIG. 12 is a diagram illustrating an example of a time-frequency signal decoded by the decoding apparatus 1200 according to the sixth embodiment. For example, the spectrum sequence of the kth (k is an integer of 0 ≦ k ≦ 15) -th low-frequency frequency signal is represented by Bk = (pk (t0), pk (t1),..., Pk (t31)). As shown in the figure, in the audio encoded bitstream generated by the encoding apparatus (not shown) of the sixth embodiment, for example, the 0th to 15th low frequency frequency signals B0 to B15 are quantized. And coded and described.

  On the other hand, for the 16th to 31st high frequency signals B16 to B31, information for specifying one of the 0th to 15th low frequency signals B0 to B15 that substitute each other, and the high frequency And gain information for adjusting the amplitude of each low-frequency time-frequency signal copied to. For example, in order to represent the 16th high frequency time frequency signal B16, the extended time frequency signal includes information indicating the 10th low frequency time frequency signal B10 replacing the 16th high frequency time frequency signal B16, and 16 The gain information G0 for adjusting the amplitude of the low frequency time frequency signal B10 copied to the high frequency part as the first high frequency time frequency signal B16 is described.

  In accordance with this, the tenth low frequency time frequency signal B10 decoded and inversely quantized by the nuclear decoding unit 1201 is copied to the high frequency part as the 16th high frequency time frequency signal B16, and gain information G0 is obtained. And the 16th high frequency frequency signal B16 is generated. The same applies to the 17th high frequency time frequency signal B17. The 11th low frequency time frequency signal B11 described in the extended time frequency signal is copied by the alternative control unit 1204 as the 17th high frequency time frequency signal B17. Amplified with the gain indicated by the gain information G1, the 17th high frequency frequency signal B17 is generated. By repeating the same processing for the 18th to 31st high frequency time frequency signals B18 to 31, all the high frequency time frequency signals can be obtained.

  As described above, according to the sixth embodiment, the encoding apparatus applies the substitution of the high frequency time frequency signal by the low frequency time frequency signal of the present invention to the time frequency signal that is the output of the QMF filter. Thus, it is possible to encode a wideband audio time-frequency signal with only a relatively small increase in the amount of data, and the decoding apparatus can decode a high-frequency rich audio signal.

  In the sixth embodiment, it has been described that each of the high frequency signals is replaced by each of the low frequency signals. However, the present invention is not limited to this, for example, the low frequency portion and the high frequency portion. May be divided into a plurality of (for example, eight) groups of the same number (for example, four) of time frequency signals, and each of the high frequency groups may be replaced with one time frequency signal of the low frequency group. .

  Alternatively, noise composed of 32 spectral values may be generated and superimposed to adjust the amplitude of the low frequency frequency signal copied to the high frequency. In the sixth embodiment, the sampling frequency is 44.1 kHz, one frame is 1024 samples, the number of samples constituting one time frequency signal is 22 and the time frequency signal is one 32 frames. However, the present invention is not limited to this. However, the present invention is not limited thereto, and the sampling frequency and the number of samples constituting one frame may be other numerical values.

  The encoding apparatus according to the present invention is an acoustic encoding apparatus provided in a satellite broadcasting station including BS and CS, and an acoustic encoding of a content distribution server that distributes content via a communication network such as the Internet. The apparatus is further useful as a program for encoding an acoustic signal executed by a general-purpose computer.

  Moreover, the decoding apparatus according to the present invention is provided not only as an acoustic decoding apparatus provided in a home STB but also as a program for decoding an acoustic signal executed by a general-purpose computer, and provided in an STB or a general-purpose computer. It is useful as a dedicated circuit board for decoding acoustic signals, LSI, etc., and further as an IC card inserted into an STB or general-purpose computer.

It is a block diagram which shows the structure of the encoding apparatus in Embodiment 1 of this invention. FIG. 2A is a diagram illustrating an MDCT coefficient sequence output by the MDCT unit. FIG. 2B is a diagram illustrating the 0th to (maxline-1) -th MDCT coefficients encoded by the quantization unit among the MDCT coefficients illustrated in FIG. FIG. 2C is a diagram illustrating an example of a method for generating an extended audio encoded sequence in the BWE encoding unit illustrated in FIG. FIG. 3A is a waveform diagram showing an MDCT coefficient sequence of the original sound. FIG. 3B is a waveform diagram showing an MDCT coefficient sequence generated by substitution by the BWE encoding unit. FIG. 3C is a waveform diagram showing an MDCT coefficient sequence when gain control is performed on the MDCT coefficient sequence shown in FIG. FIG. 4A is a diagram illustrating an example of a normal audio encoded bitstream. FIG. 4B is a diagram illustrating an example of an acoustic coded bit stream output by the coding apparatus according to the present embodiment. FIG. 4C is a diagram illustrating an example of the extended audio encoded sequence described in the extended audio encoded sequence unit illustrated in FIG. It is a block diagram which shows the structure of the decoding apparatus which decodes the audio | voice coding bit stream output from the encoding apparatus of FIG. It is a figure which shows the extended frequency spectrum information generation method by the BWE encoding part of Embodiment 2. FIG. FIG. 7A is a diagram showing subbands of the low frequency band and the high frequency band divided in the same manner as in the second embodiment. FIG. 7B is a diagram illustrating an example of the MDCT coefficient sequence of the low frequency subband A. FIG. 7C is a diagram illustrating an example of the MDCT coefficient sequence of the subband As obtained by reversing the order of the MDCT coefficient sequence of the low frequency subband A. FIG. 7D is a diagram showing a subband Ar obtained by inverting the sign of the MDCT coefficient sequence of the low frequency subband A. FIG. 8A is a diagram illustrating an example of the MDCT coefficients of the low frequency subband A specified for the high frequency subband h0. FIG. 8B is a diagram illustrating an example of the same number of MDCT coefficients as the low frequency subband A generated by the noise generation unit. FIG. 8C illustrates a high-frequency subband generated using the MDCT coefficient of the low-frequency subband A illustrated in FIG. 8A and the MDCT coefficient generated by the noise generation unit illustrated in FIG. It is a figure which shows an example of the MDCT coefficient which substitutes the band h0. FIG. 9A shows the MDCT coefficient of one frame at time t0. FIG. 9B shows the MDCT coefficient of the next frame at time t1. FIG. 9C shows the MDCT coefficient of the next frame at time t2. FIG. 10A shows the MDCT coefficient of one frame at time t0. FIG. 10B shows the MDCT coefficient of the next frame at time t1. FIG. 10C shows the MDCT coefficient of the next frame at time t2. It is a block diagram which shows the structure of the decoding apparatus which decodes a wideband time frequency signal from the audio encoding bit stream encoded using the QMF filter. FIG. 38 is a diagram illustrating an example of a time-frequency signal decoded by the decoding device according to the sixth embodiment. It is a block diagram which shows the structure of the conventional encoding apparatus.

Explanation of symbols

DESCRIPTION OF SYMBOLS 200 Encoding apparatus 201 Preprocessing part 202 MDCT part 203 Quantization part 204 BWE encoding part 205 Encoded sequence generation part 1200 Decoding apparatus 1201 Nuclear decoding part 1202 Extended decoding part 1203 Spectrum addition part

Claims (20)

An encoding device for encoding an input signal,
Time frequency conversion means for converting an input signal on the time axis into a frequency spectrum including a low frequency spectrum;
Band extension means for generating extension information used to identify a high frequency spectrum at a frequency higher than the low frequency spectrum;
Encoding means for encoding the low frequency spectrum and the extension information, and outputting the encoded low frequency spectrum and extension information,
The band spreading means, optimally from said a plurality of partial spectrum that constitutes a part or the whole of the lower frequency spectrum, to be replicated to each partial spectrum of the plurality of partial spectrum constituting the entire high-frequency spectrum Generating as the extension information a first parameter used to identify a partial spectrum and a second parameter used to determine the gain of the partial spectrum after replication;
The band extension means further generates, as the extension information , a third parameter used for determining whether or not to invert the partial spectrum to be duplicated on the frequency axis. Encoding device.   The band extending means further generates, as the extension information, a fourth parameter indicating whether or not to invert the phase of the partial spectrum to be copied.
  The encoding apparatus according to claim 1. The encoding apparatus according to claim 1, wherein the time-frequency conversion means performs MDCT (modified discrete cosine transform) on an input signal on a time axis into a frequency spectrum including a low frequency spectrum. The band extension means further generates, as the extension information, a noise parameter that specifies energy of a noise spectrum to be added to a high frequency spectrum specified by the first parameter and the second parameter,
The encoding device according to any one of claims 1 to 3, wherein the noise parameter is an energy ratio of the noise spectrum to the high frequency spectrum. The encoding device according to any one of claims 1 to 3, wherein the first parameter includes information on whether to use the same extension information as the extension information of the preceding frame. The encoding device according to claim 4, wherein the first parameter includes information indicating whether or not the same extension information as that of the immediately preceding frame is used. An encoding method for encoding an input signal, comprising:
A time-frequency conversion step of converting an input signal on the time axis into a frequency spectrum including a low-frequency spectrum;
A band extension step of generating extension information used to identify a high frequency spectrum at a frequency higher than the low frequency spectrum;
An encoding step of encoding the low frequency spectrum and the extension information, and outputting the encoded low frequency spectrum and extension information;
The bandwidth extension step includes:
From among the plurality of partial spectrum that constitutes a part or the whole of the low frequency spectrum to identify the optimal partial spectrum to be replicated to each partial spectrum of the plurality of partial spectrum constituting the entire high-frequency spectrum A first parameter used for
A second parameter used to determine the gain of the partial spectrum after replication is generated as the extension information;
The band extension step further generates, as the extension information , a third parameter used to determine whether or not to invert the partial spectrum to be duplicated on the frequency axis. Encoding method.   The band extension step further generates, as the extension information, a fourth parameter indicating whether or not to invert the phase of the partial spectrum to be copied.
  The encoding method according to claim 7. A program for encoding an input signal,
A program for causing a computer to execute the encoding method according to claim 7 or 8 . A computer-readable recording medium for recording the encoding program according to claim 9 . A decoding apparatus for decoding a No. code Cassin,
Thus the low frequency spectrum to decode No. said code Cassin, a decoding means for generating the extension information the is used to identify the high frequency spectrum at higher frequency than the lower frequency spectrum,
High frequency spectrum generating means for generating the high frequency spectrum based on the low frequency spectrum and the extended information;
Wherein the generated the high-frequency spectrum and a frequency-time converting means for converting the frequency spectrum obtained by combining the low frequency spectrum to a signal on the time axis,
The extension signal is
An optimum partial spectrum to be copied to each partial spectrum of the plurality of partial spectra constituting the entire high frequency spectrum is identified from among the plurality of partial spectra constituting a part or the whole of the low frequency spectrum A first parameter used for
A second parameter used to determine the gain of the partial spectrum after replication;
And a third parameter indicating whether or not the partial spectrum to be copied is inverted on the frequency axis .   The extended signal further includes a fourth parameter indicating whether or not to invert the phase of the partial spectrum to be copied.
  The decoding apparatus according to claim 11, wherein: The frequency-time conversion means, characterized in that the generated the high band frequency spectrum the signal to the I MDCT on the frequency spectrum time axis obtained by synthesizing the low-frequency spectrum (inverse modified discrete cosine transform) The decoding device according to claim 11 or 12 . The extended signal further includes a noise parameter that specifies an energy of a noise spectrum added to the high frequency spectrum specified by the first parameter and the second parameter;
Before Kino size parameter, Ri energy ratio der of the noise spectrum for the high band frequency spectrum,
The said high frequency spectrum generation means adds the noise spectrum which has the energy specified by the said noise parameter to the produced | generated said high frequency spectrum, The any one of Claims 11-13 characterized by the above-mentioned. Decryption device. The first parameter includes information on whether to use the same extension information as the extension information of the preceding frame,
The high-frequency spectrum generating means, decoding apparatus according to any one of claims 1 1 to 1 4, characterized in that to generate the high band frequency spectrum by using the information. The decoding device according to claim 15, wherein the first parameter includes information as to whether or not the same extension information as the extension information of the immediately preceding frame is used. A decoding method for decoding a No. code Cassin,
Thus the low frequency spectrum to decode No. said code Cassin, a decoding step of generating the extension information the is used to identify the high frequency spectrum at higher frequency than the lower frequency spectrum,
A high-frequency spectrum generation step for generating the high-frequency spectrum based on the low-frequency spectrum and the extended information;
A frequency time conversion step of converting a frequency spectrum obtained by combining the generated high frequency spectrum and the low frequency spectrum into a signal on a time axis ,
The extension signal is
From among a plurality of partial spectra constituting part or all of the low frequency spectrum, an optimum partial spectrum to be copied to each partial spectrum of the plurality of partial spectra constituting the whole high frequency spectrum is specified. A first parameter used for
A second parameter used to determine the gain of the partial spectrum after replication;
And a third parameter indicating whether or not the partial spectrum to be copied is inverted on the frequency axis .   The extended signal further includes a fourth parameter indicating whether or not to invert the phase of the partial spectrum to be copied.
  The decoding method according to claim 17, wherein: A program for decoding an encoded signal,
A program for causing a computer to execute the decoding method according to claim 17 or 18 . A computer-readable recording medium for recording the decryption program according to claim 19 .

Images