JP4317355B2 - Encoding apparatus, encoding method, decoding apparatus, decoding method, and acoustic data distribution system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an encoding device which can efficiently encode a sound signal over a wide band. <P>SOLUTION: A sound data input part 310 of the encoding device 300 cuts 4096 pieces of successive sound data out of a sound data sequence and a conversion part 320 converts the cut sound data into spectrum data on the frequency axis. A data separation part 330 separates the spectrum data into a low-frequency part and a high-frequency part based upon 11.025 kHz (f1) as a border. The low-frequency part spectrum data are quantized and encoded by a 1st quantization part 340 and an encoding part 350 as usual. A 2nd encoding part 355 generates auxiliary information showing a feature of a high-frequency part frequency spectrum and a 2nd encoding part 355 encodes the auxiliary information. A stream output part 390 puts together and output the code obtained by the 1st encoding part 350 and the code obtained by the 2nd encoding part 355. Here, f1 is less than a half of the sampling frequency f2 when the sound data sequence was generated. <P>COPYRIGHT: (C)2003,JPO

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a high-quality compression encoding and decompression decoding technique for acoustic signals.
[0002]
[Prior art]
In recent years, various types of techniques for high-quality compression encoding and decompression decoding of audio signals such as voice and music have been developed, and are referred to as “MPEG-2 Advanced Audio Coding” (hereinafter referred to as “MPEG-2 AAC” or “AAC”). "Is abbreviated as") "is one of the methods (see Non-Patent Document 1).
[0003]
[Non-Patent Document 1]
M.M. Bosi et al., “IS 13818-7 (MPEG-2 Advanced Audio Coding, AAC)”, April 1997.
FIG. 24 is a block diagram illustrating a functional configuration of a conventional AAC encoding apparatus and decoding apparatus.
[0004]
The encoding apparatus 1000 is an apparatus that compresses and encodes an input acoustic signal based on an AAC encoding scheme, and includes an A / D converter 1050, an acoustic data input unit 1100, a conversion unit 1200, a quantization unit 1400, It comprises an encoding unit 1500 and a stream output unit 1900.
[0005]
The A / D converter 1050 samples the input acoustic signal at a sampling frequency of, for example, 22.05 kHz, and converts the analog acoustic signal into a digital acoustic data string. Each time the acoustic data input unit 1100 reads an acoustic data string as an input signal by 1024 samples (this 1024 samples are hereinafter referred to as “frames”), 50% of the frame and adjacent frames before and after that frame are read. A total of 2048 samples of acoustic data strings that overlap the sample (512) are cut out. The conversion unit 1200 converts the data of 2048 samples on the time axis extracted by the acoustic data input unit 1100 into spectrum data on the frequency axis by MDCT (Modified Discrete Case Transform). Note that 1024 points of spectrum data, which is half of the data obtained by the conversion, represent a reproduction band of half the sampling frequency of 11.025 kHz or less, and are classified into a plurality of groups. Each group is set so that each of the plurality of groups includes one or more points of spectral data. In addition, each group simulates a critical band in human hearing. Each group is called a “scale factor band”. The quantization unit 1400 quantizes the spectrum data in the scale factor band obtained from the conversion unit 1200 into a predetermined number of bits using one normalization coefficient for each scale factor band. This normalization coefficient is called “scale factor”. The result of quantizing each spectrum data with each scale factor is referred to as a “quantized value”. The encoding unit 1500 performs Huffman encoding on the data quantized by the quantization unit 1400, that is, each scale factor and spectrum data quantized using the scale factor. The stream output unit 1900 converts the encoded signal obtained from the encoding unit 1500 into the stream format of the AAC bit stream and outputs it. The bit stream output from the encoding apparatus 1000 is transmitted to the decoding apparatus 2000 via a transmission medium or a recording medium.
[0006]
The decoding device 2000 is a device that decodes the bitstream encoded by the encoding device 1000, and includes a stream input unit 2100, a decoding unit 2200, an inverse quantization unit 2300, an inverse transform unit 2800, and an acoustic data output. Part 2900 and a D / A converter 2950.
[0007]
The stream input unit 2100 inputs the bit stream encoded by the encoding apparatus 1000 via a transmission medium or a recording medium, and extracts an encoded signal from the input bit stream. The decoding unit 2200 decodes the Huffman-coded encoded signal into quantized data. The inverse quantization unit 2300 performs inverse quantization on the quantized data decoded by the decoding unit 2200 using a scale factor. The inverse transform unit 2800 transforms the spectrum data of 1024 points on the frequency axis obtained by the inverse quantization unit 2300 into acoustic data of 1024 samples on the time axis using an IMDCT (Inverse Modified Discrete Cosine Transform). . The acoustic data output unit 2900 sequentially combines the acoustic data of 1024 samples on the time axis obtained by the inverse transform unit 2800, and outputs the acoustic data of 1024 samples one by one in time order. The D / A converter 2950 converts digital sound data into an analog sound signal at a sampling frequency of 22.05 kHz.
[0008]
According to the encoding apparatus 1000 and the decoding apparatus 2000 according to such a conventional AAC standard, the compression rate can be increased to 1 bit or less for each point, and reproduction at half the sampling frequency of 11.025 kHz or less. Since the spectral data of 1024 points of the low-frequency portion that represent the band and have high auditory priority are encoded, it is possible to reproduce the acoustic signal with relatively high sound quality.
[0009]
[Problems to be solved by the invention]
However, according to the conventional AAC encoding apparatus 1000 and decoding apparatus 2000 (conventional example 1), since the sampling frequency is 22.05 kHz, there is absolutely no band exceeding 11.25 kHz in the encoded spectral data. Not included. For this reason, there exists a problem that it cannot respond to the request | requirement of the further quality improvement which wants to view and listen to the zone | band exceeding 11.025kHz.
[0010]
In order to solve such a problem, the sampling frequency applied to the A / D converter 1050 and the D / A converter 2950 of the encoding device 1000 and the decoding device 2000 in FIG. It is conceivable to increase the frequency to 1 kHz (hereinafter, this method is also referred to as “conventional example 2”).
[0011]
However, if the sampling frequency is 44.1 kHz, 512 points of spectral data can be encoded in the high frequency region exceeding 11.25 kHz while maintaining the compression ratio, but the auditory priority is low. The spectral data of the part is halved to 512 points. That is, there is a trade-off relationship between the sampling frequency and the number of spectra in the low frequency region, and it is impossible to increase both at the same time with the conventional AAC. This causes another problem that the sound quality as a whole deteriorates.
[0012]
Such a situation is the same in encoding apparatuses and decoding apparatuses of other systems (for example, MP3, AC3, etc.).
The present invention has been made to solve the above-described technical problem, and an encoding device and a decoding that can realize high-quality reproduction of an acoustic signal without significantly increasing the amount of information after encoding. It is to provide a device or the like.
[0013]
[Means for Solving the Problems]
In order to solve the above problems, an encoding apparatus according to the present invention is an encoding apparatus that encodes acoustic data, and includes an extraction unit that extracts a predetermined number of continuous acoustic data from an acoustic data sequence, and The conversion means for converting the acoustic data into spectral data on the frequency axis, and the spectral data obtained by the conversion are separated into low-frequency spectral data up to f1 Hz and high-frequency spectral data higher than f1 Hz. Separation means, low-band coding means for quantizing and coding the separated low-frequency spectrum data, and auxiliary information indicating the characteristics of the high-frequency spectrum from the separated high-frequency spectrum data Auxiliary information generating means for generating, a high frequency band encoding means for encoding the generated auxiliary information, a code obtained by the low frequency band encoding means, and the high frequency Output means for synthesizing and outputting the code obtained by the partial encoding means, wherein f1 is less than or equal to half the sampling frequency f2 when the acoustic data sequence is created, and the auxiliary information generating means Is Respectively Divided into groups High region Spectral data And the low-frequency spectrum data For each group in the high frequency region, it is closest to the spectrum in that group. With spectrum Low range Spectral data of group Is generated as the auxiliary information.
[0014]
Here, f1 is f2 / 4, the converting means converts the acoustic data into spectral data of 0 to 2 × f1 Hz, and the separating means includes low-frequency spectral data of 0 to f1 Hz and f1 ˜2 × f1 Hz high-frequency spectrum data, or the low-frequency spectrum data up to f1 Hz is composed of n spectral data, and the clipping means is 2 × n The number of pieces of acoustic data necessary to generate spectrum data is cut out, the converting unit converts the cut out acoustic data into 2 × n pieces of spectral data, and the separating unit uses n pieces of low-frequency spectrums. Data and n high-frequency spectrum data are separated, or the cut-out means includes n acoustic data corresponding to one frame which is a unit of encoding. And 2 × n pieces of spectrum data that are combined with n / 2 pieces of sound data belonging to each of two frames adjacent to the frame, and the converting means cuts out 2 × n pieces of sound data. Alternatively, the conversion may be performed by MDCT and converted to a spectrum of 0 to 2 × f1 Hz including 2 × n spectrum data.
[0015]
Furthermore, a decoding device according to the present invention is a decoding device that decodes encoded data input via a recording medium or a transmission medium, and includes a low-band encoded data included in the encoded data and a high-frequency encoded data. Extraction means for extracting each of the band encoded data, and the low band encoded data extracted by the extraction means are decoded and inverse quantized to output spectrum data of the low band below the frequency f1 Low-frequency dequantizing means, and auxiliary information decoding means for generating auxiliary information representing characteristics of high-frequency spectrum data by decoding high-frequency data extracted by the extracting means, A high frequency band inverse quantization means for outputting high band spectrum data based on the auxiliary information generated by the auxiliary information decoding means, and a low frequency band spectrum output by the low frequency band inverse quantization means. Synthesis means for synthesizing the data and the high-frequency spectrum data output by the high-frequency inverse quantization means, and inverse transform for inversely transforming the spectrum data synthesized by the synthesis means into acoustic data on the time axis Means and acoustic data output means for outputting the acoustic data inversely transformed by the inverse transform means in time order, the auxiliary information is Respectively Divided into several groups The high frequency part Spectral data And the low-frequency spectrum data For each group in the high frequency region, it is closest to the spectrum in that group. With spectrum Low range Spectral data of group The high frequency band inverse quantization means generates a predetermined noise in each group of the high frequency band based on the auxiliary information, and adds it to the spectrum data to add the high frequency band spectrum data. Is generated.
[0016]
Note that the present invention can be realized as a communication system including the encoding device and the decoding device, or an encoding method including steps as characteristic means constituting the encoding device, the decoding device, and the communication system, It can be realized as a decoding method or a communication method, or it can be realized as an encoding program or decoding program for causing a CPU to execute characteristic means and steps constituting the encoding device and the decoding device. Needless to say, the present invention can be realized as a computer-readable recording medium.
[0017]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, a case where an embodiment of the present invention is applied to a broadcasting system as an acoustic data distribution system will be described in detail with reference to the drawings.
[0018]
FIG. 1 is a block diagram showing a functional configuration of the broadcasting system according to the present embodiment.
The broadcasting system 1 according to the present embodiment shown in the figure is provided in a broadcasting station, and is provided in an encoding device 300 that encodes an input acoustic signal and a user terminal. And a decoding device 400 for decoding the converted bitstream audio signal.
[0019]
(Encoding device 300)
The encoding device 300 is an encoding device that encodes an input acoustic signal, and includes an A / D converter 305, an acoustic data input unit 310, a conversion unit 320, a data separation unit 330, and a first And second quantization units 340 and 345, first and second encoding units 350 and 355, and a stream output unit 390.
[0020]
The A / D converter 305 samples the input acoustic signal at, for example, a sampling frequency 44.1 kHz that is twice that of the conventional example 1, and converts the analog acoustic signal into digital acoustic data (for example, 16 bits). The acoustic data string on the time axis is generated.
[0021]
The acoustic data input unit 310 is time-expanded to a cycle (about 45.4 msec) every time 2048 samples (2 frames) are received by the acoustic data sequence generated by the A / D converter 305, that is, twice as long as the conventional example 2. In a slow cycle, 2048 samples of these two frames and 50% of adjacent frames before and after these frames, an acoustic data string that overlaps 1024 samples, that is, twice the number of conventional (4096 samples). An acoustic data string is extracted, and includes a counter 311 for detecting the extraction timing every time 2048 samples are received, and a FIFO buffer 312 that temporarily stores an acoustic data string of 4096 samples.
[0022]
The converting unit 320 converts the acoustic data of 4096 samples for two frames cut out by the acoustic data input unit 310 into spectral data on the frequency axis, and 4096 points of acoustic data of the 4096 samples on the time axis. MDCT 321 for converting the spectrum data into spectrum data on the frequency axis, and a grouping unit 322 for grouping the spectrum data by a scale factor band.
[0023]
Specifically, the MDCT 321 converts the acoustic data of 4096 samples on the time axis into 4096 points of spectral data (16 bits). However, since it becomes two groups of symmetrical spectral data, the spectrum of one 2048 samples Only the data is to be encoded, and the other is discarded.
[0024]
As described above, when the configurations of the A / D converter 305, the acoustic data input unit 310, and the conversion unit 320 of the encoding device 300 are compared with those of the encoding device 1000 of the conventional example 1 described above, the A / D converter 305 is compared. The sampling frequency of the conversion unit 320 is increased to twice the frequency (44.1 kHz), the cut-out length in the acoustic data input unit 310 is increased to twice (4096 samples), and the encoding unit in the MDCT 321 of the conversion unit 320 is doubled (4096 samples). It is greatly different in that it has been raised.
[0025]
Compared to the configuration of the encoding apparatus 1000 of the above-described conventional example 2, the sampling frequency in the A / D converter 305 is the same, but the cut-out length in the acoustic data input unit 310 is doubled (4096 samples). The encoding unit in the MDCT 321 of the conversion unit 320 is greatly different in that it is doubled (encoding unit 4096 samples).
[0026]
As a result, from the conversion unit 320, there are 1024 pieces of spectrum data belonging to the low frequency band of 11.025 kHz or lower (hereinafter referred to as “low band spectral data”) belonging to the high frequency band exceeding 11.025 kHz. 1024 pieces of spectrum data (hereinafter referred to as “spectral data in the high frequency region”) are output, and a total of 2048 pieces of spectrum data are output.
[0027]
The grouping unit 322 of the conversion unit 320 converts the spectral data of one 2048 sample that is the target of encoding into a plurality of scale factor bands each including spectral data of one sample or more (practically a multiple of 4). Classify.
[0028]
In this AAC standard, the number of samples (spectral data) included in each scale factor band is determined according to the frequency, and the scale factor band is finely divided for each small number of samples in the low frequency range. The section is largely divided to include many samples. In AAC, the number of scale factor bands corresponding to one frame of spectrum data is also determined according to the sampling frequency. For example, when the sampling frequency is 44.1 kHz, the number of scale factor bands included in one frame is 49, and spectrum data of 1024 samples is included in the 49 scale factor bands. On the other hand, among the scale factor bands determined as described above, which scale factor band is transmitted is not particularly defined by the AAC, and the most preferable scale factor band is selected according to the transfer rate of the transmission path. It may be transmitted. For example, when the transfer rate of the transmission path is 96 kbps, only the low band 40 scale factor band (640 samples) of one frame may be selected and transmitted.
[0029]
On the other hand, in the present embodiment, the spectrum data of 2 frames (1024 low-frequency spectrum data and 1024 high-frequency spectrum data) is twice as long as the conventional cycle (about 45.4 msec). ) Is output from the MDCT 321. For this reason, when the transfer rate of the transmission path is 96 kbps, even if all the scale factor bands (1024 samples) in the low frequency part of the two frames are to be transmitted, compared to the data transfer for two frames according to the conventional method, AAC 2 times (640 × 2 = 1280 samples), a sufficient margin is generated in the transfer rate. Therefore, in the present embodiment, a case will be described in which the grouping unit 322 classifies the converted spectrum data into uniquely defined division methods and numbers of scale factor bands.
[0030]
The data separation unit 330 separates the 2048 spectral data output from the conversion unit 320 into low-frequency spectral data (1024) and high-frequency spectral data (1024). Then, the data separation unit 330 sends the separated low-band spectrum data (1024 pieces) to the first quantization unit 340 and the high-band spectrum data (1024 pieces) to the second quantization unit 345, respectively. Output.
[0031]
The first quantization unit 340 determines the scale factor for each of the scale factor bands from the low-frequency spectrum data transferred from the data separation unit 330, and the spectrum in the scale factor band is determined by the determined scale factor. Quantization is performed, and the quantization value that is the quantization result, the head of the determined scale factor, and the difference between the scale factors are output to the first encoding unit 350, and the scale factor calculation unit 341 is provided. For example, according to the formula, the scale factor calculation unit 341 calculates one normalization coefficient (scale factor, 8 bits) so that the spectrum data in the band falls within a predetermined number of bits for each scale factor band. The scale factor is used to quantize each spectrum in the scale factor band and calculate the scale factor difference.
[0032]
The first encoding unit 350 encodes the data quantized by the first quantization unit 340, the scale factor for each scale factor band, and the like into a predetermined stream format. A Huffman coding table 351 for further compressing each piece of data and each scale factor is provided. Specifically, Huffman coding is performed so that each data quantized using the Huffman coding table 351, each scale factor, and the like are transmitted at a low bit rate.
[0033]
The second quantizing unit 345 calculates auxiliary information based on the spectral data of the band not quantized by the first quantizing unit 340, that is, the high-frequency part data of 11.0525 kHz or higher output from the data separating unit 330. And includes an auxiliary information generation unit 346 for generating auxiliary information. Here, the auxiliary information is simplified information that is calculated based on the spectrum data of the high frequency region and briefly represents the characteristics of the spectrum data of the high frequency region with a small amount of information. In other words, it is information representing the characteristics of the high frequency part of the spectrum data obtained by converting the input acoustic data for a certain period of time, and a specific example is absolute within the scale factor band of the high frequency part. This is an optimum scale factor for each scale factor band and its quantized value so that the quantized value of the maximum spectral data (spectrum data having the maximum absolute value) is “1”.
[0034]
The second encoding unit 355 encodes the auxiliary information output from the second quantization unit 345 into a predetermined stream format and outputs the encoded information as second encoded information. A Huffman coding table 356 is provided.
[0035]
The stream output unit 390 adds header information and other sub information as necessary to the first encoded signal output from the first encoding unit 350, and converts the first encoded signal into an AAC encoded bit stream as usual. The second encoded signal output from the second encoding unit 355 is stored in an area in the bit stream that is ignored by the conventional decoding device or whose operation is not defined. Specifically, the stream output unit 390 stores the encoded signal output from the second encoding unit 355 in a Fill Element, a Data Stream Element, or the like in an AAC encoded bit stream.
[0036]
However, the information indicating the sampling frequency of the bit stream stored in the header information stores a value that is half the sampling frequency of the acoustic data. That is, when the sampling frequency of the acoustic data is 44.1 kHz, the actual half 22.02 kHz information is stored in the header. The information indicating the actual sampling frequency 44.1 kHz may be stored in an area where the auxiliary information is stored.
[0037]
The bit stream output from the encoding device 300 is transmitted to the decoding device 400 via a transmission medium such as the Internet configured by radio waves, light, metal, or the like.
[0038]
Thus, when the encoding device 300 quantizes and encodes the spectrum data on the frequency axis obtained by the conversion unit 320, the data separation unit 330 performs low-frequency spectrum data (1024 points), and Separated into high-frequency spectrum data (1024 points), low-frequency spectrum data is quantized and encoded in the same manner as before, and high-frequency spectrum data is quantized in a different method And encoding (generating auxiliary information, encoding auxiliary information), and encoding the low-frequency encoded bitstream into the high-frequency encoded bitstream for output. Is significantly different from the conventional coding apparatus 1000 that performs the same method over the entire band.
[0039]
As a result, it is possible to encode a high-quality acoustic signal within a range where the total amount of information does not increase significantly compared to the conventional case.
In addition, since the information of the sampling frequency 22.05 kHz is stored in the header, there is an effect that the conventional decoding apparatus 2000 can also decode the bitstream generated by the encoding apparatus 300 in the present embodiment. .
[0040]
(Decryption device 400)
Decoding apparatus 400 according to the present embodiment performs a process substantially reverse to that of encoding apparatus 300 on the bit stream output from encoding apparatus 300 to generate an acoustic signal (reproduction upper limit frequency 22.05 kHz) on the time axis. A stream input unit 410, first and second decoding units 420 and 425, first and second inverse quantization units 430 and 435, and an inverse quantized data synthesis unit 440; , An inverse conversion unit 480, an acoustic data output unit 490, and a D / A converter 495.
[0041]
The stream input unit 410 inputs the bit stream encoded by the encoding device 300 via a transmission medium, and the first code stored in the area used by the conventional decoding device from the input bit stream. The first decoding unit 420 and the second decoding unit 425 are respectively extracted from the encoded signal and the second encoded signal stored in an area where the conventional decoding device is not ignored or specified for operation. And output respectively.
[0042]
The first decoding unit 420 receives the first encoded signal output from the stream input unit 410, decodes the first encoded signal from the stream format into quantized data, and performs Huffman decoding for decoding A conversion table 421 is provided.
[0043]
The first dequantization unit 430 dequantizes the quantized data decoded by the first decoding unit 420 and outputs spectrum data. The first dequantization unit 430 dequantizes the quantized data based on a formula. The processing part 431 for performing this is provided. Here, the number of samples of the spectrum data output from the first inverse quantization unit 430 is 1024, and these represent a reproduction band of 11.025 kHz or less.
[0044]
The second decoding unit 425 receives the second encoded signal output from the stream input unit 410 and decodes auxiliary information, and includes a Huffman decoding table 426 for decoding the auxiliary information. .
[0045]
The second inverse quantization unit 435 generates high-frequency spectrum data based on the auxiliary information, and includes a spectrum data generation unit 436. Here, the number of samples of the spectrum data output from the second inverse quantization unit 435 is 1024, and these represent a reproduction band exceeding 11.025 kHz.
[0046]
The spectrum data generation unit 436 generates noise in a predetermined procedure based on the spectrum data output from the first inverse quantization unit 430, and also includes auxiliary information output from the second decoding unit 425. Then, the above-mentioned noise is shaped and high-frequency spectrum data is output. This noise includes white noise, pink noise, and the like, as well as spectral data obtained by partially or entirely copying the low-frequency spectrum data.
[0047]
Specifically, the spectrum data generation unit 436, for example, copies the low-frequency part spectrum data output from the first inverse quantization unit 430 to the high-frequency part, and each scale factor band of the high-frequency part. The ratio between the absolute maximum value of the spectrum data copied in the band and the value obtained by dequantizing the quantized value “1” using the scale factor value corresponding to the band described in the auxiliary information By multiplying each spectrum data in the band as a coefficient, the spectrum in the high frequency part is restored.
[0048]
The inverse quantized data synthesizer 440 synthesizes the spectrum data output from the first inverse quantizer 430 and the spectrum data output from the second inverse quantizer 435. Here, the number of samples of spectrum data output from the inverse quantized data combining unit 440 is 2048, which represents a reproduction band of 0 to 22.05 kHz.
[0049]
As described above, the decoding apparatus 400 includes the first (low frequency band) encoded signal stored in the area used by the conventional decoding apparatus, from the bitstream encoded by the encoding apparatus 300. The conventional decoding apparatus separates each of the second (high frequency band) encoded signals stored in an area where the ignoring or the operation is not specified, and only the first (low frequency band) encoded signals. Is decoded and dequantized in the same manner as in the past, and the second (high band) encoded signal is decoded and dequantized in a scheme different from the conventional one, and the low band spectrum data and the high band The configuration is greatly different from the decoding apparatuses 2000 of the conventional examples 1 and 2 that perform the decoding and inverse quantization of the bitstream in the same manner over the entire band in that they are synthesized and output.
[0050]
As a result, it is possible to decode the amount of information that is significantly increased compared to the conventional amount from the same small amount of information as in the past, and to decode a high-quality sound signal.
The inverse transform unit 480 transforms the spectrum data on the frequency axis output from the inverse quantized data synthesis unit 440 into acoustic data on the time axis of 2048 samples (2 frames) using IMDCT.
[0051]
The acoustic data output unit 490 sequentially combines the acoustic data of 2048 samples on the time axis obtained by the inverse transform unit 480, and outputs the acoustic data of 2048 samples one by one in time order.
[0052]
The D / A converter 495 converts digital sound data into an analog sound signal at a sampling frequency of 44.1 kHz.
As described above, in the decoding apparatus 400, the inverse transform unit in the inverse transform unit 480 is doubled (2048 samples), the frame length in the acoustic data output unit 490 is doubled (2048 samples), and D The configuration of the decoding apparatus 2000 of the conventional example 1 is greatly different in that the sampling frequency in the / A converter 495 is increased to a double frequency (44.1 kHz).
[0053]
As a result, from the D / A converter 495, a high band (0 to 22.2) based on low-frequency spectrum data (1024) of 11.25 kHz or less and high-frequency spectrum data (1024). (05 kHz) and a high-quality acoustic signal is output.
[0054]
As described above, according to the functional configuration of the present embodiment, the low-frequency part performs conventional encoding while the high-frequency part has an extremely small amount of information while being based on the amount of information in a range that does not increase significantly compared to the conventional case. By performing the encoding with, it is possible to encode and decode a high-quality acoustic signal.
[0055]
In addition, in the configuration of encoding apparatus 300 and decoding apparatus 400 in the present embodiment, data separation section 330, second quantization section 345 and second encoding section 355 are added to conventional encoding apparatus 1000. In addition, since only the second decoding unit 425, the second inverse quantization unit 435, and the inverse quantized data synthesis unit 440 are added to the conventional decoding device 2000, the existing encoding device 1000 and decoding There is an effect that it can be realized without significantly changing the configuration of the apparatus 2000.
[0056]
In addition, the bit stream generated by the encoding apparatus 300 according to the present embodiment can be decoded by the conventional decoding apparatus 2000.
Next, the encoding process of each unit of the encoding device 300 in the broadcast system 1 will be specifically described.
[0057]
FIG. 2 is a diagram illustrating a state change of an acoustic signal processed in the acoustic data input unit 310 and the conversion unit 320 of the encoding device 300 illustrated in FIG. 2A is a waveform diagram showing 2048 sample data on the time axis cut out by the acoustic data input unit 310 shown in FIG. 1, and FIG. 2B is a waveform diagram showing sample data on the time axis. 3 is a waveform diagram showing spectrum data on the frequency axis after being converted by the MDCT 321 of the conversion unit 320 shown in FIG. In FIG. 2A and FIG. 2B, the sample data and spectrum data are shown as analog waveforms, but in actuality, both are digital signals. The same applies to the following waveform diagrams.
[0058]
The acoustic data input unit 310 receives acoustic data sampled at 44.1 kHz. The sound data input unit 310 overlaps and cuts out the 1024 samples before and after the sound data is input every 2048 samples, and outputs them to the conversion unit 320.
[0059]
The conversion unit 320 MDCTs the data of a total of 4096 samples, but since the spectrum obtained by MDCT has a symmetrical waveform, the spectrum data corresponding to half of the 2048 samples is output as shown in FIG. To do.
[0060]
In the spectrum data shown in FIG. 2 (b), the vertical axis indicates the frequency spectrum value, that is, the amount (size) of the frequency component of the acoustic data represented by the voltage value of 2048 samples in FIG. 2 (a). , 2048 points corresponding to the number of samples. Further, since the acoustic signal input to the encoding device 300 is A / D converted at the sampling frequency of 44.1 kHz, the reproduction band of the spectrum data is 22.05 kHz. Furthermore, since the spectrum obtained by MDCT 321 may take a negative value as shown in FIG. 2B, when the spectrum obtained by MDCT 321 is encoded, the sign of the spectrum is also added. It is necessary to make it. Hereinafter, in order to avoid confusion with the encoding code, information representing the positive and negative signs of the spectrum data is referred to as “sign information”.
[0061]
Spectral data and sign information output from the conversion unit 320 are separated into a low frequency range of 0 to 11.025 kHz and a high frequency range of 11.025 kHz in the data separation unit 330, and spectral data of the low frequency range Are output to the first quantization unit 340, and the spectral data of the high frequency region are output to the second quantization unit 345, respectively.
[0062]
FIG. 3 is a flowchart showing an operation in the scale factor determination process of the first quantizing unit 340 shown in FIG.
The first quantization unit 340 first determines a common scale factor for each scale factor band as an initial value of the scale factor (S91), and uses that scale factor as acoustic data for one frame (1024 samples). All the low-frequency spectrum data to be transmitted is quantized, the difference before and after the obtained scale factor is obtained, and the difference, the head scale factor, and each quantized value are Huffman-coded (S92). Note that quantization and encoding here are performed only for counting the number of bits, and therefore, in order to simplify the processing, only data is performed, and information such as a header is not added.
[0063]
Next, the first quantization unit 340 determines whether or not the number of bits of the Huffman-encoded data exceeds a predetermined number of bits (S93), and if it exceeds, decreases the initial value of the scale factor ( S101) Using the value of the scale factor, quantization and Huffman coding are performed again for the same lowband spectrum data (S92), and lowband coding data for one frame after Huffman coding is performed. It is determined whether or not the number of bits exceeds a predetermined number of bits (S93), and this process is repeated until the number of bits becomes equal to or less than the predetermined number of bits.
[0064]
If the number of bits of the low band encoded data does not exceed the predetermined number of bits, the first quantization unit 340 repeats the following processing for each scale factor band, and determines the scale factor of each scale factor band. (S94). First, each quantized value in the scale factor band is inversely quantized (S95), and the difference between each absolute value between each inverse quantized value and the corresponding original spectrum data is obtained and summed (S96). Further, it is determined whether or not the sum of the obtained differences is within the allowable range (S97), and if it is within the allowable range, the above processing is repeated for the next scale factor band (S94 to S98). .
[0065]
On the other hand, if the allowable range is exceeded, the scale factor value is increased and the spectrum data of the scale factor band is quantized (S100), and the quantized value is inversely quantized (S95). The difference in absolute value between the value and the corresponding spectrum data is summed (S96). Further, it is determined whether or not the sum of the differences is within the allowable range (S97), and if it exceeds the allowable range, the scale factor is sequentially increased until it is within the allowable range (S100), and the above processing (S95 to S97 and S100) is repeated.
[0066]
The first quantizing unit 340 is configured such that, for all scale factor bands, the sum of absolute value differences between the values obtained by dequantizing the quantized values in the scale factor bands and the original spectrum data falls within the allowable range. When the scale factor is determined (S98), the low-frequency spectrum data for one frame is quantized again using the determined scale factor, and the difference between each scale factor, the head scale factor, and each quantized value are obtained. Huffman coding is performed, and it is determined whether or not the number of bits of the low-frequency part encoded data exceeds a predetermined number of bits (S99). If the number of bits of the low-frequency encoded data exceeds the predetermined number of bits, the initial value of the scale factor is lowered until it becomes equal to or less than the predetermined number of bits (S101), and then the scale within each scale factor band The process for determining the factor (S94 to S98) is repeated. If the number of bits of the low frequency band encoded data does not exceed the predetermined number of bits (S99), the value of each scale factor at that time is determined as the scale factor of each scale factor band.
[0067]
The first quantizing unit 340 quantizes the low-frequency spectrum data using the scale factor determined in this way, and obtains the quantized value, the head of the determined scale factor, and the difference between the scale factors. The signature information received from the data separator 330 is output to the first encoder 350.
[0068]
It should be noted that whether the sum of the absolute values of the quantized values in the scale factor band and the inverse spectral values of the quantized values falls within the allowable range is determined based on data such as a psychoacoustic model. Is called.
[0069]
Also, here, the initial value of the scale factor is set to a relatively large value, and when the number of bits of the low band encoded data after Huffman coding exceeds a predetermined number of bits, the scale factor is sequentially changed. Although the scale factor is determined by a method of lowering the initial value, it is not always necessary to do so. For example, when the initial value of the scale factor is set to a low value in advance, the initial value is gradually increased, and the total number of bits of the low-frequency encoded data first exceeds the predetermined number of bits. Thus, the scale factor of each scale factor band may be determined using the initial value of the scale factor set immediately before.
[0070]
Furthermore, although the scale factor of each scale factor band is determined here so that the number of bits of the entire low-frequency encoded data for one frame does not exceed a predetermined number of bits, this need not necessarily be done. For example, in each scale factor band, the scale factor may be determined so that each quantized value in the scale factor band does not exceed a predetermined number of bits. Hereinafter, the operation of the first quantization unit 340 in this process will be described with reference to FIG.
[0071]
FIG. 4 is a flowchart showing an operation in another scale factor determination process of the first quantization unit 340 shown in FIG.
The first quantizing unit 340 calculates scale factors according to the following procedure for all the scale factor bands in the low frequency range to be encoded (S1). Further, the first quantization unit 340 calculates the scale factor according to the following procedure for all spectrum data in each scale factor band (S2).
[0072]
First, the first quantizing unit 340 quantizes the spectrum data based on a formula with a value of a predetermined scale factor (S3), and the quantized value represents a predetermined number of bits to represent the quantized value. For example, it is determined whether or not it exceeds 4 bits (S4).
[0073]
As a result of the determination, if the quantized value exceeds 4 bits, the scale factor value is adjusted (S8), and the same spectrum data is quantized with the adjusted scale factor value (S3). The first quantization unit 340 determines whether or not the obtained quantized value exceeds 4 bits (S4), and adjusts the scale factor until the quantized value of the spectrum data becomes a value of 4 bits or less. (S8) and quantization by the scale factor after adjustment (S3) are repeated.
[0074]
As a result of the determination, if the quantized value is 4 bits or less, the next spectrum data is quantized with a value of a predetermined scale factor (S3).
When the quantized values of all the spectrum data in one scale factor band are 4 bits or less (S5), the first quantizing unit 340 converts the scale factor value at that time into the scale factor of the scale factor band. (S6).
[0075]
Furthermore, when the first quantizing unit 340 determines scale factors for all scale factor bands (S7), the process is terminated.
With the above processing, one scale factor is determined for each of all the scale factor bands in the low frequency band to be encoded. The first quantizing unit 340 quantizes the spectrum data of the low frequency band using the scale factor determined in this way, and a 4-bit quantized value as a quantization result and the 8-bit scale factor. And the difference between the scale factors are output to the first encoding unit 350 together with the sign information received from the data separation unit 330.
[0076]
The quantized value, scale factor, and the like output to the first encoding unit 350 are Huffman encoded and output to the stream output unit 390 as a first encoded signal equivalent to the case of downsampling.
[0077]
On the other hand, the second quantizing unit 345 generates auxiliary information based on the spectral data of the high frequency region.
FIG. 5 is a spectrum waveform diagram showing a specific example of the auxiliary information (scale factor) generated by the second quantization unit 345 shown in FIG. 1, and FIG. 6 is the second quantization unit shown in FIG. 345 is a flowchart illustrating an operation in the auxiliary information (scale factor) calculation process 345.
[0078]
In FIG. 5, the delimiters shown on the low-frequency part frequency axis represent the delimiters of the scale factor bands defined in the present embodiment. In addition, a break indicated by a broken line in the frequency direction in the high frequency part indicates a scale factor band break defined in the present embodiment. The same applies to the following waveform diagrams.
[0079]
Of the spectrum data output from the conversion unit 320, a low-frequency part having a reproduction band of 11.025 kHz or less indicated by a solid waveform in FIG. 5 is output to the first quantization unit 340 and quantized as usual. On the other hand, the high frequency region up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz indicated by the broken line waveform in FIG. 5 is represented by auxiliary information (scale factor) calculated by the second quantization unit 345. .
[0080]
Hereinafter, the procedure for calculating auxiliary information (scale factor) of the second quantizing unit 345 will be described with reference to the flowchart of FIG. 6 using the specific example of FIG.
The second quantizing unit 345 sets the quantized value of the absolute maximum spectrum data in each scale factor band to “1” for all scale factor bands in the high frequency range up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz. Is calculated according to the following procedure (S11).
[0081]
The second quantizing unit 345 specifies the absolute maximum spectrum data (peak) in the first scale factor band of the high frequency region exceeding the reproduction band of 11.0525 kHz (S12). In the specific example of FIG. 5, it is assumed that the position of the peak specified in the first scale factor band is (1) and the peak value at that time is “256”.
[0082]
Similar to the procedure shown in the flowchart of FIG. 4, the second quantization unit 345 assigns the peak value “256” and the initial scale factor value to the formula for calculating the quantized value, and obtains it from the formula. The value of the scale factor sf with which the quantized value to be “1” is calculated (S13). For example, in this case, a value of the scale factor sf that sets the quantization value of the peak value “256” to “1”, for example, sf = 24, is calculated.
[0083]
When the scale factor value sf = 24 for setting the peak quantized value to “1” is obtained for the first scale factor band (S14), the second quantizing unit 345 determines the spectrum for the next scale factor band. The peak of the data is specified (S12). For example, when the specified peak position is (2) and the value is “312”, the quantized value of the peak value “312” is “1”. A value of the scale factor sf, for example, sf = 32 is calculated (S13).
[0084]
Similarly, the second quantizing unit 345 sets the value of the scale factor sf for setting the quantized value of the value “288” of the peak (3) to “1” for the third scale factor band in the high frequency range, For example, sf = 26 is calculated, and the value of the scale factor sf, for example, sf = 18, for setting the quantized value of the value “203” of the peak (4) to “1” is calculated for the fourth scale factor band.
[0085]
In this way, when the scale factor for setting the quantized value of the peak value to “1” is calculated for all the scale factor bands in the high frequency part (S14), the second quantizing unit 345 calculates by the calculation The obtained scale factor of each scale factor band is output to the second encoding unit 355 as auxiliary information of the high frequency part, and the process is terminated.
[0086]
As described above, the auxiliary information (scale factor) is generated by the second quantizing unit 345. The auxiliary information (scale factor) is obtained by converting the high frequency portion represented by the spectrum data of 1024 points into each of the high frequency portions. If the value of the scale factor is represented by a value from 0 to 255, it can be represented by 8 bits for each scale factor band (four in this case) in the high frequency region. Further, if the difference between the scale factors is Huffman-encoded, the amount of data may be further reduced. On the other hand, if the spectrum data of 1024 points in the high frequency part is quantized and Huffman encoded by the conventional method as in the low frequency part, it is predicted that the data amount will be at least about 300 bits. Therefore, this auxiliary information shows only one scale factor for each scale factor band in the high frequency part, but the amount of data is larger than when the high frequency part is quantized according to the conventional method. It can be seen that it has been reduced.
[0087]
The scale factor is a value that is substantially proportional to the peak value (absolute value) in each scale factor band, and is one of the spectrum data that takes a constant value at 1024 points in the high frequency region or the spectrum data in the low frequency region. It can be said that the spectrum data obtained by multiplying a copy of all or all copies by the scale factor roughly restores the spectrum data obtained based on the input acoustic signal. Also, for each scale factor band, the coefficient is the ratio between the absolute maximum value of the spectrum data copied in the band and the value obtained by dequantizing the quantized value “1” using the scale factor value corresponding to the band. As described above, the spectral data can be restored more accurately by multiplying each spectral data in the band. Furthermore, since the difference in the waveform of the high frequency part is not as clearly audibly identified as the low frequency part, the auxiliary information obtained in this way is sufficient as information representing the waveform of the high frequency part. It can be said.
[0088]
Here, the scale factor is calculated so that the quantized value of the spectrum data in each scale factor band in the high frequency region is “1”. However, the scale factor is not necessarily “1” and is set to another value. You may keep it.
[0089]
The auxiliary information generated by the second quantization unit 345 is Huffman encoded by the second encoding unit 355, and the conventional decoding apparatus in the bit stream as the second encoded signal by the stream output unit 390. It is ignored or stored in an area where its operation is not specified.
[0090]
FIG. 7 is a diagram illustrating a position in the bit stream in which auxiliary information is stored by the stream output unit 390 illustrated in FIG. In FIG. 7, after the auxiliary information representing the spectrum of the high frequency band is encoded, it is stored as a second encoded signal in a region that is not recognized as an acoustic encoded signal in the bit stream.
[0091]
In FIG. 7A, a hatched portion is, for example, a region (Fill Element) that is filled with “0” in order to match the data length of the bit stream. Even if the information, that is, the second encoded signal is stored, the conventional decoding apparatus 2000 does not recognize the encoded signal to be decoded and ignores it.
[0092]
In FIG. 7B, the hatched portion is, for example, an area called Data Stream Element (DSE), and this area has only a physical structure such as a bit length according to the AAC standard for future expansion. This is a defined area. Similar to the Fill Element, this area is read even if auxiliary information representing the spectrum in the high frequency part is stored here or ignored by the conventional decoding apparatus 2000 or even if the data is read. This is an area where the operation of the decryption apparatus 2000 for data is not defined.
[0093]
In the above description, the second encoded signal is stored in an area in the bitstream that is ignored by the conventional decoding apparatus 2000 according to the MPEG-2 AAC standard. The second encoded signal may be incorporated at a predetermined position in the first encoded signal, or may be incorporated across both. In addition, since the second encoded signal is stored in the bit stream, it is not necessary to secure a continuous area in the header and the first encoded signal. That is, as shown in FIG. 7C, the second encoded signal may be incorporated discontinuously in the header information and the first encoded information.
[0094]
FIG. 8 is a diagram illustrating another example when the stream output unit 390 illustrated in FIG. 1 stores auxiliary information. FIG. 8A shows a stream 1 in which only the first encoded signal is stored continuously for each frame. FIG. 8B shows the stream 2 in which only the second encoded signal in which the auxiliary information is encoded is continuously stored for each frame corresponding to the stream 1.
[0095]
The stream output unit 390 may store the second encoded signal in a stream 2 that is completely different from the stream 1 that is a bit stream in which the first encoded signal is stored. For example, stream 1 and stream 2 are bit streams transmitted on different channels.
[0096]
In this way, by transmitting the first encoded signal and the second encoded signal in completely different bit streams, a low-frequency portion representing basic information of the input acoustic signal is transmitted or accumulated in advance, There is an effect that the high frequency band information can be added later if necessary.
[0097]
In the format shown in FIGS. 7 and 8, the information indicating the sampling frequency of the bitstream stored in the header stores information of 22.05 kHz, which is half of the actual sampling frequency. As a result, the decoding apparatus 2000 of the first conventional example can also decode the bit stream in the 0 to 11.025 kHz band and reproduce it in the same manner as in the case of downsampling.
[0098]
Next, differences between the scheme of encoding apparatus 300 according to the embodiment of the present invention and the scheme of encoding apparatus 1000 of Conventional Example 1 will be described with reference to FIG. FIG. 9 is a diagram showing a comparison between the method according to the embodiment and the method according to Conventional Example 1 (downsampling method), and particularly FIG. 9A is a diagram showing the method according to the embodiment. FIG. 9B is a diagram showing a method according to Conventional Example 1.
[0099]
In the method according to the present embodiment, by setting the sampling frequency to 44.1 kHz, an acoustic data sequence is acquired every 22.7 μsec, and 2048 included in the encoding target frame and 1024 before and after that are included. A total of 4096 pieces are cut out and 2048 pieces of spectral data are obtained by MDCT. The reproduction band of this spectrum data represents 22.05 kHz. The 2048 spectral data are separated into spectral data (1024) in the low band part and 1024 spectral data (1024) in the high band part with a boundary of 11.25 kHz. The low-frequency spectrum data (1024) is subjected to normal quantization and encoding, and a first encoded signal having a high quality and a low bit rate equivalent to that of downsampling is obtained. In addition, 1024 pieces of spectral data are acquired for the high frequency region. If this is normally quantized and encoded, a low bit rate cannot be realized. Therefore, in the method according to the present embodiment, auxiliary information is generated from 1024 pieces of spectrum data in the high frequency band, and only the auxiliary information is encoded to obtain the second encoded signal. Therefore, it is possible to encode a high-quality acoustic signal within a range where the total amount of information does not increase significantly.
[0100]
On the other hand, in the method by the downsampling in the conventional example 1, by setting the sampling frequency to 22.05 kHz, an acoustic data string is obtained every 45 μsec, and 1024 included in the encoding target frame, A total of 2048 pieces including 512 pieces before and after are cut out and MDCT is performed to obtain 1024 pieces of spectrum data. The reproduction band of this spectrum data represents 11.025 kHz. The 1024 pieces of spectrum data are subjected to normal quantization / encoding. Therefore, although a high-quality encoded signal can be acquired at 11.025 kHz or less, there is no high-frequency spectrum data exceeding 11.025 kHz, and thus a high-frequency encoded signal cannot be acquired.
[0101]
Next, the difference between the method of encoding apparatus 300 according to the embodiment of the present invention and the method of encoding apparatus 1000 of Conventional Example 2 will be described with reference to FIG.
FIG. 10 is a diagram showing a comparison between the method according to the embodiment and the method according to the second conventional example, and in particular, FIG. 10A is a diagram showing the method according to the embodiment, and FIG. ) Is a diagram showing a method according to Conventional Example 2. FIG. Since the method according to the present embodiment has been described above, the description thereof is omitted.
[0102]
In the sampling method of Conventional Example 2, an acoustic data string is acquired every 22.7 μsec by setting the sampling frequency to 44.1 kHz as in the case of the present embodiment, and is included in the encoding target frame. A total of 2048 data including 1024 data and 512 data before and after that data are cut out and MDCT is performed to obtain 1024 spectral data. The reproduction band of this spectrum data represents 22.05 kHz. The 1024 pieces of spectrum data are subjected to normal quantization / encoding. That is, 1024 pieces of spectrum data (512 in the low range below 11.025 kHz, 512 in the high range exceeding 11.025 kHz) are acquired every half time (about 22.7 msec) of the present embodiment. ing.
[0103]
Here, it is assumed that the auxiliary information is generated from the high-frequency spectrum data of 11.025 to 22.05 kHz in the encoding apparatus 1000 of the conventional example 2 as in the embodiment of the present invention. In this case, if the number of bits that can be used for quantization at about every 22.7 msec is n and the number of bits that can be used as auxiliary information is m1, 512 samples in the low band (0 to 11.025 kHz) are (n−m1). ) It needs to be quantized with bits. On the other hand, in the embodiment of the present invention, when the number of bits per about 45.4 msec that can be used for quantization is 2 × n and the number of bits that can be used as auxiliary information is m2, It is only necessary to quantize 1024 samples (11.025 kHz) with (2 × n−m 2) bits.
[0104]
By the way, in AAC, it is generally known that high coding efficiency cannot be obtained unless a certain number of samples (threshold) or more are collected. In the case of 1024 samples of this embodiment, the threshold value is sufficiently exceeded.
[0105]
Therefore, it is higher to quantize 1024 samples with (2 × n−m2) bits as in the present embodiment than to quantize 512 samples with (n−m1) bits as in Conventional Example 2. Encoding efficiency is obtained. Further, as a result of obtaining higher encoding efficiency in the present embodiment, m2 can be increased (m2> 2 × m1), and the sound quality in the high frequency region can be improved.
[0106]
FIG. 11 is a diagram comparing spectral data and characteristics of the coding method according to the present embodiment and the coding methods of the first and second conventional examples.
In this embodiment, the sampling frequency is 44.1 kHz and the frame length is 2048. For this reason, in the spectral data, 1024 pieces of spectral data are acquired in the low frequency range of 0 to 11.025 kHz, and auxiliary information based on 1024 spectral data of the high frequency range of 11.025 to 22.05 kHz is acquired. . As a result, in the present embodiment, the bandwidth is substantially the same as that of Conventional Example 2, but is wider than that of Conventional Example 1. Further, in the present embodiment, in terms of sound quality, the low-frequency part of 0 to 11.025 kHz is the same quality as the conventional example 1, but the high-frequency part of 11.025 to 22.05 kHz is auxiliary. Since there is information, it becomes high quality as a whole, and with respect to the conventional example 2, it is possible to obtain substantially the same quality because there is auxiliary information for the high frequency part of 11.025 to 22.05 kHz, and 0 to 11.025 kHz. Since the low-frequency part is of high quality because the number of spectrum data is doubled, the overall quality is high.
[0107]
On the other hand, in the conventional example 1, the sampling frequency is 22.05 kHz, the frame length is 1024, and 1024 pieces of spectrum data are acquired in the band of 0 to 11.025 kHz as the spectrum data. As a result, the conventional example 1 is halved and narrower in terms of the bandwidth than the present embodiment. For this reason, in terms of sound quality, the low frequency range of 0 to 11.25 kHz is the same quality, but the high frequency range of 11.025 to 22.05 kHz deteriorates because there is no spectral data at all, and is low overall. Become.
[0108]
In Conventional Example 2, the sampling frequency is 44.1 kHz, the frame length is 1024, and 1024 pieces of spectral data are acquired in the entire band from 0 to 22.05 kHz. As a result, the conventional example 2 is the same as the present embodiment in terms of bandwidth, but in terms of sound quality, spectrum data is encoded for the high frequency range of 11.025 to 22.05 kHz. Although the quality is high, the number of spectrum data is halved in the low frequency band of the band 0 to 11.025 kHz, so that the quality is lowered and the quality is lowered as a whole.
[0109]
Therefore, according to the present embodiment, the low-frequency part performs the conventional encoding, and the high-frequency part is encoded with a very small amount of information, so that the total amount of information is significantly increased compared to the conventional case. It is possible to encode a high-quality acoustic signal within a range that does not.
[0110]
Next, the encoding process of each unit of the decoding device 400 in the broadcast system 1 will be specifically described.
The first encoded signal output from the stream input unit 410 is decoded into quantized data or the like by the first decoding unit 420, and the first dequantization unit 430 encodes the low band spectral data. It becomes. On the other hand, the second decoded signal output from the stream input unit 410 is decoded into auxiliary information by the second decoding unit 425. The second inverse quantization unit 435 generates spectral data of a high frequency part based on the auxiliary information. The processing in the second inverse quantization unit 435 will be described in detail.
[0111]
FIG. 12 is a flowchart illustrating a procedure in which the low frequency band 1024 spectrum is copied to the high frequency band in the forward direction by the second inverse quantization unit 435 illustrated in FIG. 1. Such a copy of the spectrum data of the low frequency part is executed when generating the spectrum data of the high frequency part.
[0112]
In FIG. 12, inv_spec1 [i] indicates the value of the i-th spectrum in the output data of the first inverse quantization unit 430, and inv_spec2 [j] is the input data of the second inverse quantization unit 435. Among these, the value of the j-th spectrum is shown.
[0113]
First, since the second inverse quantization unit 435 inputs the 0th spectrum to the 1023rd spectrum in the same direction, the initial values of the counters i and j for counting the number of spectra are set to “0”, respectively. (S71). Next, the second inverse quantization unit 435 checks whether or not the value of the counter i is less than “1024” (S72), and if the value of the counter i is less than “1024”, the first inverse quantum The value of the i-th (0th in this case) spectrum of the quantization unit 430 is input as the value of the j-th (0th in this case) spectrum of the second inverse quantization unit 435. (S73). Thereafter, the second inverse quantization unit 435 increments the values of the counters i and j by “1” (S74), and checks whether the value of the counter i is less than “1024” (S72). .
[0114]
The second inverse quantization unit 435 repeats the above process while the value of the counter i is less than “1024”, and ends the process when the value of the counter i becomes “1024” or more.
[0115]
As a result, the entire spectrum of the 0 to 1023rd low-frequency part, which is the inverse quantization result of the first inverse quantization unit 430, is directly copied as the high-frequency part spectrum of the second inverse quantization unit 435. .
[0116]
The spectrum data thus copied is adjusted in the amplitude of the copied spectrum data according to the auxiliary information decoded by the second decoding unit 425, that is, the value of the scale factor for setting the peak value to “1”. And output as high-frequency spectrum data. The amplitude is adjusted by dequantizing the quantized value “1” for each scale factor band using the absolute maximum value of the spectrum data copied in the band and the scale factor value corresponding to the band. This is achieved by multiplying each spectrum data in the band by using the ratio of the above as a coefficient. Here, the maximum number of samples of the spectrum data output from the second inverse quantization unit 435 is 1024, which represents a reproduction band exceeding 11.025.
[0117]
In FIG. 12, the procedure for copying the low-frequency part 1024 spectrum to the high-frequency part in the forward direction in the frequency axis direction is shown, but it may be copied in the opposite direction as shown in FIG.
[0118]
FIG. 13 is a flowchart illustrating a procedure in which the low frequency band 1024 spectrum is copied to the high frequency band in the reverse direction of the frequency axis direction by the second inverse quantization unit 435 illustrated in FIG. Similar to FIG. 12, inv_spec1 [i] in FIG. 13 indicates the value of the i-th spectrum in the output data of the first inverse quantization section 430, and inv_spec2 [j] is the second inverse quantization. The value of the j-th spectrum in the input data of the part 435 is shown.
[0119]
First, since the second inverse quantization unit 435 inputs the 0th spectrum to the 1023rd spectrum in the reverse direction, the initial value of the counter i for counting the number of spectra is set to “0”, and the initial value of j The value is set to “1023” (S81). Next, the second inverse quantization unit 435 checks whether or not the value of the counter i is less than “1024” (S82), and if the value of the counter i is less than “1024”, the first inverse quantum The value of the i-th (0th in this case) spectrum of the quantization unit 430 is input as the value of the j-th (1023th) spectrum of the second inverse quantization unit 435. (S83). Thereafter, the second inverse quantization unit 435 increments the value of the counter i by “1”, decrements the value of j by “1” (S84), and the value of the counter i is less than “1024”. It is checked whether or not there is (S82).
[0120]
The second inverse quantization unit 435 repeats the above process while the value of the counter i is less than “1024”, and ends the process when the value of the counter i becomes “1024” or more.
[0121]
As a result, the entire spectrum of the 0 to 1023 low frequency band, which is the inverse quantization result of the first dequantization unit 430, is the 1023 to 0th spectrum of the high frequency band of the second dequantization unit 435. Is copied in the reverse direction.
[0122]
Even in this case, the copied spectrum data has auxiliary information decoded by the second decoding unit 425, that is, the amplitude of the spectrum data copied according to the value of the scale factor that makes the peak value “1”. It is adjusted and output as high-frequency spectrum data. The amplitude is adjusted by dequantizing the quantized value “1” for each scale factor band using the absolute maximum value of the spectrum data copied in the band and the scale factor value corresponding to the band. This is achieved by multiplying each spectrum data in the band by using the ratio of the above as a coefficient. Here, the maximum number of samples of the spectrum data output from the second inverse quantization unit 435 is 1024, which represents a reproduction band exceeding 11.025.
[0123]
Note that the second inverse quantization unit 435 copies all the spectral data in the low-frequency part to the high-frequency part, but may copy only a part.
In addition, as an example of the procedure for copying the entire high frequency band and low frequency band at the same time, the case of FIG. 12 and FIG. 13 is given as an example. May be.
[0124]
Also, some or all of them may be copied with the positive and negative signs reversed. Furthermore, these copying procedures may be determined in advance, may be changed according to the data in the low frequency region, or may be transmitted as auxiliary information.
[0125]
Further, here, the spectral data on the low frequency side is copied as the spectral data on the high frequency side, but this is not limiting, and the spectral data on the high frequency side is generated only from the second encoded information. May be.
[0126]
Furthermore, in the present embodiment, a case has been described in which spectral data obtained mainly from the first inverse quantization unit 430 is copied as noise generation in the second inverse quantization unit 435, but the present invention is limited to this. In addition, spectral data having a constant value within each scale factor band of the high frequency part, white noise, pink noise, etc. may be generated independently by the second inverse quantization unit 435, and supplementary information It may be generated accordingly.
[0127]
The 1024 pieces of spectrum data output from the second inverse quantization unit 435 are combined with the spectrum data (1024 pieces) output from the first inverse quantization unit 430 in the inverse quantization data combining unit 440, and the IMDCT By doing so, it is inversely converted into acoustic data on the time axis, and by performing D / A conversion at a sampling frequency of 44.1 kHz, an audio signal having a reproduction band of 0 to 22.05 kHz is reproduced.
[0128]
As described above, according to the present invention, MDCT and IMDCT having twice the conversion length of the conventional method are used, and the conventional coding is performed only on the first 1024 samples with respect to the spectrum data of 2048 samples, and the remaining second 1024. The samples were encoded with a smaller amount of information than before, and were synthesized at the time of decoding.
[0129]
By reducing the amount of information necessary for encoding the spectral data of the latter half 1024 samples, the amount of information necessary for encoding the spectral data of the first 1024 samples can be increased, and the accuracy of the original signal in the low frequency region is improved. However, wideband encoding could be performed.
[0130]
In addition, the bit stream generated by the encoding apparatus according to the present embodiment can be decoded by a conventional decoding apparatus.
Next, various modifications of auxiliary information and its decoding will be described.
[0131]
FIG. 14 is a spectrum waveform diagram showing a specific example of other auxiliary information (quantized value) generated by the second quantizing unit 345 shown in FIG. FIG. 15 is a flowchart showing an operation in another auxiliary information (quantization value) calculation process of the second quantization unit 345 shown in FIG.
[0132]
The second quantization unit 345 predetermines a common scale factor value, for example, “18”, for all scale factor bands in the high frequency range up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz. Using the scale factor value “18”, the quantization value of the absolute maximum value spectrum data (peak) in the scale factor band is calculated for each scale factor band (S21).
[0133]
The second quantizing unit 345 specifies the absolute maximum spectrum data (peak) in the first scale factor band in the high frequency part exceeding the reproduction band of 11.0525 kHz (S22). In the specific example of FIG. 14, it is assumed that the peak position specified in the first scale factor band is {circle around (1)} and the peak value at that time is “256”.
[0134]
The second quantization unit 345 applies a common scale factor value “18” and a peak value “256” to the formula for calculating the quantization value, and calculates the quantization value (S23). For example, in this case, when the peak value “256” is quantized with the scale factor value “18”, the quantized value “6” is calculated.
[0135]
When the quantized value “6” of the peak value “256” is obtained for the first scale factor band (S24), the second quantizing unit 345 specifies the peak of the spectrum data for the next scale factor band. (S22) For example, if the specified peak position is (2) and the value is “312”, the quantized value of the peak value “312” with the scale factor value “18”, for example, “10” is calculated (S23).
[0136]
Similarly, the second quantizing unit 345 has the quantized value “9” of the peak “3” value “288” with the scale factor value “18” for the third scale factor band in the high frequency region. ”And the quantized value“ 5 ”of the value“ 203 ”of the peak (4) with the scale factor value“ 18 ”is calculated for the fourth scale factor band.
[0137]
In this manner, when the quantized value of the peak value when the scale factor is fixed to “18” is calculated for all the scale factor bands in the high frequency part (S24), the second quantizing unit 345 Then, the quantized value of each scale factor band obtained by the calculation is output to the second encoding unit 355 as auxiliary information of the high frequency part, and the process is terminated.
[0138]
As described above, the auxiliary information (quantized value) is generated by the second quantizing unit 345. The auxiliary information (quantized value) is obtained by converting the high-frequency part represented by the spectrum data of 1024 points. Each of the four scale factor bands is represented by a 4-bit quantized value. On the other hand, in the auxiliary information (scale factor) described above, the high frequency region is expressed by an 8-bit scale factor for each of the four scale factor bands. It has been reduced more. Further, this quantized value roughly represents the amplitude of the peak value (absolute value) in each scale factor band, and is one of spectral data that takes a constant value at 1024 points in the high frequency region or spectral data in the low frequency region. Even spectral data obtained by simply multiplying a copy of all or part of the copy can be said to roughly restore the spectral data obtained based on the input acoustic signal. Also, for each scale factor band, the ratio between the absolute maximum value of the spectrum data copied in the band and the value obtained by dequantizing the quantized value corresponding to that band using a predetermined scale factor value By multiplying each spectrum data in the band by using as a coefficient, the spectrum data can be restored with higher accuracy.
[0139]
In this embodiment, the scale factor value corresponding to the quantized value transmitted as the second encoded information is set in advance. However, an optimal scale factor value is calculated and the second code is calculated. It may be transmitted in addition to the conversion information. For example, if the scale factor is selected so that the maximum value of the quantized value is 7, the number of bits representing the quantized value can be 3 bits, so that the amount of information necessary for transmitting the quantized value can be reduced. .
[0140]
FIG. 16 is a spectrum waveform diagram showing a specific example of other auxiliary information (position information) generated by the second quantization unit 345 shown in FIG. FIG. 17 is a flowchart showing an operation in another auxiliary information (position information) calculation process of the second quantization unit 345 shown in FIG.
[0141]
The second quantization unit 345 performs the following procedure for the position of the absolute maximum spectral data in each scale factor band for all the scale factor bands in the high frequency range up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz. Therefore, it specifies (S31).
[0142]
The second quantization unit 345 specifies the absolute maximum spectrum data (peak) in the first scale factor band in the high frequency region exceeding the reproduction band of 11.0525 kHz (S32). In the specific example of FIG. 16, it is assumed that the position of the peak specified in the first scale factor band is {circle around (1)} and is the 22nd spectrum data from the head of this scale factor band. The second quantization unit 345 holds the position of the identified peak “the 22nd spectrum data from the head of the scale factor band” (S33).
[0143]
When the peak position is specified and held for the first scale factor band (S34), the second quantization unit 345 specifies the peak of the spectrum data for the next scale factor band (S32). For example, assume that the specified peak position is {circle around (2)} and is the 60th spectrum data from the head of the scale factor band. The second quantizing unit 345 holds the position of the specified peak “60th spectrum data from the head of the scale factor band” (S33).
[0144]
In the same manner, the second quantizing unit 345 specifies and holds the position of the peak {circle around (3)} "the first spectral data of the scale factor band" for the third scale factor band in the high frequency part. For the fourth scale factor band, the position of peak (4) “25th spectrum data from the beginning of the scale factor band” is specified and held.
[0145]
In this way, when the peak positions are specified and held for all the scale factor bands in the high frequency region (S34), the second quantizing unit 345 causes the peak of each scale factor band held. Is output to the second encoding unit 355 as auxiliary information of the high frequency region, and the process ends.
[0146]
As described above, auxiliary information (position information) is generated by the second quantizing unit 345. This auxiliary information (position information) is obtained by converting the high frequency part represented by the spectrum data of 1024 points to 4 Each scale factor band is represented by 6-bit position information.
[0147]
In this case, in the decoding device 400, the second inverse quantization unit 435 receives a part or all of the spectrum data for 1024 samples in the low band part as auxiliary information (from the second decoding unit 425 ( Is copied as 1024 sample data on the high frequency side according to the position information. In the copying procedure, based on the peak information of the spectrum data in one or more scale factor bands, similar data is extracted from the spectrum data output from the first inverse quantization unit 430, and part or all of the data is extracted. This is achieved by copying. In addition, the second inverse quantization unit 435 adjusts the amplitude of the copied spectral data as necessary. Amplitude adjustment is achieved by multiplying each spectral data by a predetermined coefficient, for example, “0.5”. This coefficient may be a fixed value, may be changed for each band or for each scale factor band, or may be changed according to spectrum data output from the first inverse quantization unit 430.
[0148]
In the above description, a predetermined coefficient is used, but the value of this coefficient may be added to the second encoded information as auxiliary information. Alternatively, the scale factor value may be added to the second encoded information as a coefficient, or the quantized value of the peak in the scale factor band may be added to the second encoded information as the coefficient. The amplitude adjustment method is not limited to this, and other methods may be used.
[0149]
In this embodiment, only position information or only position information and coefficient information is encoded as auxiliary information. However, the present invention is not limited to this, and scale factor, quantized value, spectrum sign information, and A noise generation method or the like may be encoded. Moreover, you may encode combining these 2 or more.
[0150]
Moreover, although the spectrum data on the low band side is copied as the spectrum data on the high band section side, the present invention is not limited to this, and the spectrum data on the high band section side may be generated only from the second encoded information. .
[0151]
FIG. 18 is a spectrum waveform diagram showing a specific example of other auxiliary information (sign information) generated by the second quantization unit 345 shown in FIG. FIG. 19 is a flowchart showing an operation in another auxiliary information (sign information) calculation process of the second quantization unit 345 shown in FIG.
[0152]
The second quantizing unit 345 has a predetermined position of each scale factor band, for example, at the center of the scale factor band, for all the scale factor bands in the high frequency range up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz. The signature information of the spectrum data is specified according to the following procedure (S41).
[0153]
The second quantizing unit 345 checks the sign information of the spectrum data at the center position of the first scale factor band in the high frequency region exceeding the reproduction band of 11.0525 kHz (S42), and holds the value. For example, the sign code of the spectrum data at the center position of the first scale factor band is “+”. The second quantizing unit 345 represents this code “+” as a 1-bit value “1” and holds it. If this sign is “−”, it is represented by “0” and held.
[0154]
When the sign information of the spectrum data at the center position of the scale factor band is held for the first scale factor band (S43), the second quantization unit 345 determines the spectrum data at the center position for the next scale factor band. The sign is checked (S42). For example, if the checked code is “+”, the second quantization unit 345 holds “1” as the sign information of the spectrum data at the center position of the second scale factor band.
[0155]
Similarly, the second quantization unit 345 checks the sign “+” of the spectrum data at the center position of the third scale factor band in the high frequency region, holds the sign information “1”, and The sign “+” of the spectral data at the center position of the scale factor band is checked, and the sign information “1” is retained.
[0156]
In this way, when the sign information of the spectrum data at the center position is held for all the scale factor bands in the high frequency part (S43), the second quantizing unit 345 holds each scale factor band held. Is output to the second encoding unit 355 as auxiliary information of the high frequency band, and the process is terminated.
[0157]
As described above, auxiliary information (sign information) is generated by the second quantizing unit 345. This auxiliary information (sign information) is obtained by converting the high frequency portion represented by 1024 points of spectral data into 4 Each scale factor band is represented by 1-bit sine information, and the spectrum of the high frequency band can be represented with a very short data length.
[0158]
In this case, in the decoding device 400, the second inverse quantization unit 435 copies a part or all of the spectrum data for 1024 samples of the low band part as the high band side spectrum, and the second decoding unit The sign of the spectrum data at a predetermined position is determined according to the sign information input from 425.
[0159]
Here, the sign information representing the sign of the center position of each scale factor band in the high frequency region is used as auxiliary information (sign information), but is not limited to the position of the center of the scale factor band. It may be sign information, may be sign information at the beginning of a scale factor band, or may be a predetermined position other than that.
[0160]
Here, the position of the spectrum data corresponding to the code (sign information) to be transmitted is predetermined, but this may be changed according to the output of the first inverse quantization unit 430. The position information indicating the position of the sign information of each scale factor band may be added to the second encoded information and transmitted.
[0161]
The second inverse quantization unit 435 adjusts the amplitude of the copied spectral data as necessary. The adjustment of the amplitude can be achieved by multiplying each spectrum data by a predetermined coefficient, for example, a value of “0.5” and multiplying the coefficient.
[0162]
This coefficient may be a fixed value, may be changed for each band or for each scale factor band, or may be changed according to the spectrum data output from the first inverse quantization unit 430. The amplitude adjustment method is not limited to this, and other methods may be used.
[0163]
In this embodiment, a predetermined coefficient is used, but the value of this coefficient may be added to the second encoded information as auxiliary information. Further, a scale factor value may be added to the second encoded information as the coefficient, or a quantized value may be added to the second encoded information as a coefficient.
[0164]
In addition, only sign information, only sign information and coefficient information, only sign information and position information, or only sign information, position information, and coefficient information are encoded as auxiliary information. Alternatively, the quantization value, scale factor, characteristic spectrum position information, noise generation method, and the like may be encoded. Two or more of these may be combined and encoded.
[0165]
Further, in the present embodiment, the spectrum data on the low frequency band side is copied as the spectrum data on the high frequency band side. However, the present invention is not limited to this, and the spectrum data on the high frequency band side is the second encoded information. You may generate from only.
[0166]
In the above description, the code “+” is represented by a 1-bit value “1”, and the code “−” is represented by “0”. It is not limited and may be expressed by other values.
[0167]
FIG. 20 is a spectrum waveform diagram showing an example of a method of creating other auxiliary information (copy information) generated by the second quantization unit 345 shown in FIG. FIG. 20A is a waveform diagram showing a spectrum in the first scale factor band in the high frequency region. FIG. 20B is a waveform diagram showing an example of the spectrum waveform of the low frequency region specified by the auxiliary information (copy information). FIG. 21 is a flowchart showing an operation in another auxiliary information (copy information) calculation process of the second quantization unit 345 shown in FIG.
[0168]
The second quantizing unit 345, for all the scale factor bands in the high frequency range up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz, is the peak position n from the head of the scale factor band (nth from the beginning). ), The scale factor band number N at which the peak position from the head of the scale factor band is the closest to n in the low frequency region is specified according to the following procedure (S51).
[0169]
The second quantizing unit 345 specifies the position n of the absolute maximum spectral data (peak) in the first scale factor band in the high frequency region exceeding the reproduction band of 11.0525 kHz (S52). As a result, for example, as shown in FIG. 20A, it is assumed that the specified peak position is {circle around (1)} and the spectrum is spectrum data of n = 22 in this scale factor band.
[0170]
The second quantizing unit 345 specifies the positions of all peaks (including both positive and negative) of the spectrum in the low frequency region where the frequency of the spectrum is not more than the reproduction band of 11.0525 kHz (S53).
[0171]
Next, the second quantizing unit 345 searches for all the peaks specified in the low frequency part for a scale factor band whose position from the peak to the beginning of the scale factor band is closest to n, and for the scale factor band. The number N, the search direction, and the peak sign information are specified (S54).
[0172]
Specifically, the second quantizing unit 345 sequentially selects all of the specified peaks (including both positive and negative) from the peak on the low frequency side in the scale factor band whose position from the peak is closest to n. Search for the beginning. There are two search directions: a case where the search is further performed from the peak toward the lower frequency direction (1), and a case where the search is performed further from the peak toward the higher frequency direction (2). In addition, the search for the high frequency peak and the low frequency peak in which the positive / negative sign is reversed is also performed when searching from the peak toward the lower frequency (3), and from the peak toward the higher frequency. There are two ways of searching (4).
[0173]
Among these, when the search directions are (2) and (4), when the spectrum waveform of the low band is copied based on this peak information, the high band is shown as shown in FIG. Since the waveform in which the position of the peak and the position of the low-frequency peak are reversed left and right (frequency axis direction) in the scale factor band is copied, for example, the search directions of (1) and (3) are forward And (2) and (4) as reverse directions, it is necessary to attach information indicating the forward and reverse search directions. When the search directions are (3) and (4), as shown in FIG. 20 (b), the position of the peak in the high frequency region and the position of the peak in the low frequency region are vertically (vertical direction). Since the inverted waveform is copied, it is necessary to attach information indicating whether or not the positive and negative signs of the high band peak value and the low band peak value are reversed.
[0174]
The second quantizing unit 345 determines that the peak specified in the low band is negative in the search directions of (1) and (2) if the peak specified in the low band has a positive value. If the peak has the value of (3) and (4), a search is performed in four directions, and among the search results, the number of the scale factor band whose position from the peak is closest to n is specified. To do. In this case, an error range with n is set to a predetermined value, for example, “5” in advance, and a scale factor band whose position from the peak is closest to n is selected from the four search results. The number N of the scale factor band is specified. At the same time, sign information indicating whether or not the positive and negative signs of the peak value in the high frequency region and the peak value in the low frequency region are inverted, and information indicating the reverse of the search direction are specified.
[0175]
For example, in the search direction (1), corresponding to the low-frequency spectrum as shown in (1) of FIG. 20B, the position error from the peak is “1” and the scale factor band number N = Assume that 3 is specified. Further, in the search direction (2), the scale factor band number N == 5 with the position error from the peak “5” corresponding to the spectrum in the low band as shown in (2) of FIG. 18 is specified, and similarly, in the search direction (3), the error is “4” corresponding to the spectrum of the low frequency region as shown in (3) of FIG. In the search direction (4) with the number N = 12, the scale factor band number N = 10 has an error of “2” corresponding to the low-frequency spectrum as shown in (4) of FIG. Suppose that it was identified. The second quantizing unit 345 has an error in the position from the peak of “4” of the specified scale factor band numbers, and the number N = 3 of the scale factor band closest to the position from the peak n. Select. At the same time, sign information “1” indicating the sign “+” of the peak in the low frequency region and search direction information “1” indicating that the search is further performed from the peak toward the low frequency direction are generated. In this case, if the sign of the peak is “−”, the sign information is “1”, and if searching from the peak further toward the high frequency direction, the search direction information is represented as “0”.
[0176]
When the scale factor band number N = 3, the sign information “1”, and the search direction information “1” are specified for the first scale factor band in the high frequency range (S55), the second quantization unit 345 In the same manner as described above, the scale factor band number N, its sign information, and its search direction information are specified for the next scale factor band.
[0177]
In this way, for all scale factor bands in the high frequency band, the low frequency where the peak position from the head of the scale factor band is closest to n with respect to the peak position n from the head of the scale factor band. When the number N of the scale factor band of the part, the sign information thereof, and the search direction information thereof are specified (S55), the second quantizing part 345 corresponds to each scale factor band of the specified high frequency part. The scale factor band number N in the low frequency part, the sign information, and the search direction information are output to the second encoding unit 355 as auxiliary information (copy information) in the high frequency part, and the process ends.
[0178]
In this case, when the decoding apparatus 400 decodes the first encoded signal according to the conventional procedure, spectrum data of 1024 samples on the low frequency side is obtained. The second inverse quantization unit 435 copies a part or all of the spectrum data corresponding to the scale factor band number output from the second decoding unit 425 as a high frequency band side spectrum. The second inverse quantization unit 435 adjusts the amplitude of the copied spectral data as necessary. The amplitude can be adjusted by multiplying each spectrum by a predetermined coefficient, for example, by setting the value to “0.5” and multiplying the coefficient.
[0179]
This coefficient may be a fixed value, may be changed for each band, for each scale factor band, or may be changed according to the spectrum data output from the first inverse quantization unit 430.
[0180]
In this embodiment, a predetermined coefficient is used for amplitude adjustment. However, the value of this coefficient may be added to the second encoded information as auxiliary information. Further, a scale factor value may be added to the second encoded information as a coefficient, or a quantized value may be added to the second encoded information as a coefficient. The amplitude adjustment method is not limited to this, and other methods may be used.
[0181]
In addition to the scale factor band number N, the sign information and the search direction information are extracted as auxiliary information (copy information) for the high frequency band, but the sign information depends on the amount of information that can be transmitted for the high frequency band. And search direction information may be omitted. The sign information is “1” if the sign of the low-frequency peak is “+”, and “0” if it is “−”, and the search direction information is further from the peak toward the lower frequency. The search is represented as “1”, and the search from the peak toward the higher frequency direction is represented as “0”. However, the sign of the low-frequency peak in the sign information and how to express the search direction of the search direction information are as follows. These are not limited to these, and may be expressed by other values.
[0182]
In addition, the head of the scale factor band whose distance is closest to n from the position of each peak specified in the low frequency part has been searched. However, the present invention is not limited to this example, and the head of each scale factor band in the low frequency part is searched. The peak whose distance is closest to n may be searched.
[0183]
FIG. 22 is a spectrum waveform diagram showing a second example of a method of creating other auxiliary information (copy information) generated by the second quantization unit 345 shown in FIG. FIG. 23 is a flowchart showing an operation in the second calculation process of other auxiliary information (copy information) of the second quantizing unit 345 shown in FIG.
[0184]
The second quantizing unit 345 performs spectral difference (energy difference) with respect to all the spectrums within the scale factor band for all the scale factor bands in the high frequency range up to the reproduction band 22.05 kHz exceeding the reproduction band 11.0525 kHz. The number N of the scale factor band of the low-frequency part that minimizes () is specified according to the following procedure (S61). However, the number of spectra taking the difference from the high frequency region in the low frequency region is equal to the number of spectra in the scale factor band of the high frequency region, and the specified scale factor band number N is the head of the spectrum. The scale factor band number.
[0185]
The second quantizing unit 345 performs processing for all the scale factor bands in the low band part (S62), with the frequency composed of the same number of spectrum data as the spectrum data in the scale factor band in the high band part from the head of the scale factor band. A difference between the spectrum of the high frequency region and the spectrum of the low frequency region is obtained by the width (S63). For example, in the waveform diagram shown in FIG. 22, if the first scale factor band in the high frequency part is a scale factor band with the number of spectral data = 48, the second quantizing unit 345 uses the number N of the low frequency part. For 48 spectral data from the head of the scale factor band of = 1, the difference in spectrum between the high frequency region and the low frequency region is obtained sequentially.
[0186]
When the second quantizing unit 345 obtains the spectrum difference between the high-frequency part and the low-frequency part for the same number of spectra as the scale factor band of the high-frequency part (S65), the second quantizing part 345 holds the value, A difference between the high-frequency spectrum and the low-frequency spectrum is obtained from the beginning of the low-frequency scale factor band with the same frequency width of the spectrum data as the spectrum in the high-frequency scale factor band (S64). For example, when the difference of the spectrum is obtained with the width of 48 spectrum data from the beginning of the scale factor band of the low band number N = 1, the value of the obtained difference is held, A spectrum difference is obtained with a width of 48 spectrum data from the head of the scale factor band of number N = 2. Similarly, the low-frequency part N = 3 scale factor band, the number N = 4 scale factor band,..., The number N = 28 (the last low-frequency part) scale factor band, and so on. For all the scale factor bands in the region, the difference between the 48 spectral data in the high region and the low region is sequentially added to obtain the spectrum difference.
[0187]
For all scale factor bands in the low frequency band, the width of the high frequency spectrum and the low frequency spectrum is the same as the spectrum data in the high frequency scale factor band from the beginning of the scale factor band. When the difference is obtained (S64), the second quantization unit 345 specifies the number N of the scale factor band that minimizes the obtained difference (S65). For example, in the spectrum waveform diagram shown in FIG. 22, it is assumed that the scale factor band of number N = 8 in the low frequency region is specified. This indicates that the spectrum of the portion indicated by the oblique line in the low frequency part has the smallest difference from the spectrum of the part indicated by the oblique line in the high frequency part, and the energy difference between the spectra is the smallest. That is, when 48 spectral data from the head of the scale factor band of number N = 8 are copied to the first scale factor band of the high band starting from 11.25 kHz, they are indicated by a one-dot chain line in the high band of FIG. It becomes a waveform, and can express the energy within the scale factor band of the high frequency part approximately with respect to the original spectrum.
[0188]
When the second quantization unit 345 specifies the number N of the low-frequency scale factor band that minimizes the difference in spectrum for the high-frequency scale factor band, the second quantization unit 345 holds the specified scale factor band number N. In the same manner as described above, the number N of the corresponding scale factor band is specified for the scale factor band of the next high frequency part (S66). Hereinafter, this process is sequentially repeated for each scale factor band of the high frequency part, and when the number N of the low frequency scale factor band having the smallest spectrum difference is specified in all the scale factor bands of the high frequency part, The stored low-band scale factor band number N is output to the second encoding unit 355 as auxiliary information (copy information) for the high band, and the process is terminated.
[0189]
In this case, the low-frequency side spectrum copy method and amplitude adjustment method in decoding apparatus 400 are the same as those in the case of auxiliary information (copy information) described with reference to FIGS.
[0190]
Further, in the flowchart of FIG. 23, when calculating the energy difference between the high frequency region and the low frequency region, the calculation is performed in the same direction and in the same direction on the frequency axis. However, the encoding device of the present invention is limited to this. Instead, as described with reference to FIGS. 20 and 21, the energy difference between the high frequency region and the low frequency region may be calculated using any of the following three methods. (1) The low-frequency scale for the high-frequency spectrum data that is selected from the low-frequency side to the high-frequency side, with the same spectral data values for the high-frequency part and the low-frequency part. Spectral data is sequentially selected from the high frequency side to the low frequency side (that is, in the reverse direction on the frequency axis) for the same number of spectral data as the high frequency region from the beginning of the factor band, and the difference is calculated. (2) The sign of the low frequency spectrum is inverted (minus) and calculated in the same direction on the frequency axis. (3) The sign of the low frequency spectrum is inverted (minus) and the calculation is performed in the reverse direction on the frequency axis. Further, after performing the calculation by all these four methods, the number N of the scale factor band of the low-frequency spectrum that minimizes the energy difference among these may be used as auxiliary information. In this case, in order to correctly copy the low band spectrum with the smallest energy difference to the high band, information indicating the sign relationship between the low band spectrum and the high band spectrum and the low band spectrum Information indicating the direction on the frequency axis for copying the region spectrum is included in the auxiliary information for each scale factor band. The information indicating the relationship between the codes of the low-frequency spectrum and the high-frequency spectrum is, for example, “1” when the difference is obtained with the same code and “0” when the difference is obtained with the opposite code. expressed. The information indicating the direction on the frequency axis when copying the low-frequency spectrum to the high-frequency section is, for example, selecting spectral data in the high-frequency area and the low-frequency area when copying in the forward direction. When the direction is forward, “1”, when copying in the reverse direction, that is, when the direction in which spectrum data is selected in the high-frequency part and the low-frequency part is the reverse direction, “0” is set to 1 bit. expressed.
[0191]
In addition, although the case where the example of the acoustic data distribution system which concerns on the said embodiment was applied to the broadcasting system was demonstrated, the acoustic data distribution system which distributes acoustic data with a bit stream from a server via a transmission medium, such as the internet, to a terminal The bit stream output from the encoding device 300 is temporarily recorded on a recording medium such as an optical disk such as a CD or DVD, a semiconductor, or a hard disk, and is reproduced by the decoding apparatus 400 via this recording medium. You may apply to such an acoustic data delivery system.
[0192]
In the above embodiment, the LONG block is used. However, the SHORT block may be used, and the SHORT block may be processed in the same way as the LONG block.
[0193]
In the encoding process, tools such as Gain Control, TNS (TEMPORAL NOISE SHAPING), psychoacoustic model, M / S Stereo, Intensity Stereo, Prediction, block size switching, bit reservoir, etc. may be used. .
[0194]
In the above embodiment, the auxiliary information is generated based on the spectral data of the high frequency band separated by the data separation unit 330, but the value obtained by dequantizing the output of the first quantization unit 340 is set to the high frequency band. The auxiliary information may be generated based on the spectral data of the high frequency region.
[0195]
In addition, as auxiliary information, a scale factor such that the quantized value of spectrum data in each scale factor band in the high frequency part is “1”, a quantized value, characteristic spectral position information, and positive / negative of the spectrum. Although it implemented with the sign information showing a code | symbol, what combined these 2 or more is good also as auxiliary information. In this case, it is particularly effective if the auxiliary information is encoded in combination with the scale factor such as a coefficient representing the amplitude ratio and the position of the absolute maximum spectrum data. In the present embodiment, one auxiliary information is encoded for each scale factor band as the second encoded signal. However, one auxiliary information is encoded for each of two or more scale factor bands. Alternatively, two or more pieces of auxiliary information may be encoded in one scale factor band. Moreover, the auxiliary information in this Embodiment may encode auxiliary information for every channel, and may encode one auxiliary information with respect to two or more channels.
[0196]
In the present embodiment, the number of quantization units and the number of coding units in coding apparatus 300 are two, but the present invention is not limited to this, and three or more quantization units and decoding units are provided. May be.
[0197]
Further, in the present embodiment, two decoding units and two inverse quantization units are included in decoding apparatus 400, but the present invention is not limited to this, and three or more decoding units and dequantization units are not limited thereto. May be provided.
[0198]
Further, the above processing can be realized not only by hardware but also by software, and a configuration in which a part is realized by hardware and the rest is realized by software may be adopted.
In the above embodiment, the sampling frequency is 44.1 kHz. However, the sampling frequency may be 32 kHz, 48 kHz, or the like, and the separation frequency in the data separation unit 330 is set to a frequency other than 11.25 kHz. It may be changed.
[0199]
Furthermore, although the above embodiment has been described with respect to AAC, the same may be applied to encoding apparatuses and decoding apparatuses of other systems (for example, MP3, AC3, etc.).
[0200]
The encoding device may be configured as follows.
That is, an encoding device that encodes acoustic data, a cutout unit that continuously cuts out a number m2 of acoustic data that exceeds the required number m1 from the acoustic data string, and the acoustic data cut out by the cutout unit that cuts out the acoustic data. Conversion means for converting into spectrum data on the frequency axis, and m2 spectrum data obtained by the conversion are separated into the m1 low-frequency spectrum data and (m2-m1) high-frequency spectrum data. Separation means, low-band coding means for quantizing and coding the separated low-frequency spectrum data, and auxiliary information indicating characteristics of the high-frequency spectrum from the separated high-frequency spectrum data Auxiliary information generating means for generating, a high band encoding means for encoding the generated auxiliary information, and a code obtained by the low band encoding means It may be characterized in that the sign obtained in the serial high frequency portion encoding means synthesizes and an output means for outputting.
[0201]
In this case, the auxiliary information generation means calculates a normalization coefficient for the spectrum data divided into a plurality of groups, such that the value obtained by quantizing the peak spectrum data in each group in the high frequency region becomes a constant value. In addition, the calculated normalization coefficient may be generated as the auxiliary information.
[0202]
Further, the auxiliary information generation means, for the spectrum data divided into a plurality of groups, quantize the spectrum data that becomes a peak in each group of the high frequency region using a normalization coefficient common to each group, The quantized value is generated as the auxiliary information.
[0203]
Further, the auxiliary information generating means generates, as the auxiliary information, the frequency position of the spectrum data that becomes a peak in each group of the high frequency part for the spectrum data divided into a plurality of groups. I can do it.
[0204]
Further, the spectrum data is an MDCT coefficient, and the auxiliary information generating means assigns a sign indicating whether the spectrum data is positive or negative at a predetermined frequency position in a high frequency region for the spectrum data divided into a plurality of groups. It can be set as the characteristic characterized by producing | generating as.
[0205]
In addition, the auxiliary information generation means includes, for the spectrum data divided into a plurality of groups, information for specifying a spectrum in a low frequency portion that most closely approximates a spectrum in the group in each high frequency region group. It can be set as the characteristic characterized by producing | generating as information. In this case, the auxiliary information generating means is configured such that the distance on the frequency axis from the group break to the peak of the high band spectrum in the group and the frequency from the low band group break to the peak of the low band spectrum. It can be set as the structure characterized by pinpointing the low frequency part spectrum from which the difference with the distance on an axis | shaft becomes the minimum. Further, the auxiliary information generating means identifies a spectrum in a low frequency region where a difference value is minimized when an energy difference is taken with the same frequency width as a spectrum in the group in the high frequency region. Can be configured. Further, the information specifying the spectrum of the low-frequency part may be a number specifying the group of the specified low-frequency spectrum.
[0206]
The auxiliary information generating means may generate a predetermined coefficient representing the amplification ratio of the amplitude of the high band spectrum as the auxiliary information.
[0207]
Further, the output means further converts the data encoded by the low frequency band encoding means into an encoded audio stream defined in a predetermined format, and is an area in the encoded audio stream. A stream output unit that stores and outputs the data encoded by the high frequency band encoding means in an area that is not restricted by the encoding protocol can be used. In this case, the stream output unit may be configured to write information representing f1 Hz as a sampling frequency.
[0208]
Further, the output means further converts the data encoded by the low frequency band encoding means into an encoded acoustic stream defined in a predetermined format and is encoded by the high frequency band encoding means. A second stream output unit that stores and outputs the data in a stream different from the encoded audio stream can be provided.
[0209]
It is realized as a communication system comprising the encoding device and the decoding device of this modification, or realized as an encoding method and communication method using characteristic means constituting the encoding device and the communication system as steps. Can be realized as an encoding program that causes the CPU to execute characteristic means and steps constituting the encoding device, or can be realized as a computer-readable recording medium on which these programs are recorded. Needless to say.
[0210]
【The invention's effect】
As is clear from the above description, the encoding device according to the present invention is an encoding device that encodes acoustic data, and includes a cutting-out unit that cuts out a predetermined number of pieces of acoustic data from the acoustic data sequence, and a cutting-out unit. The conversion means for converting the acquired acoustic data into spectral data on the frequency axis, and the spectral data obtained by the conversion are separated into low-frequency spectral data up to f1 Hz and high-frequency spectral data higher than f1 Hz. Separation means, low-band coding means for quantizing and coding the separated low-frequency spectrum data, and auxiliary information indicating characteristics of the high-frequency spectrum from the separated high-frequency spectrum data Auxiliary information generating means for generating, a high frequency band encoding means for encoding the generated auxiliary information, and a code obtained by the low frequency band encoding means Output means for synthesizing and outputting the code obtained by the high frequency band encoding means, wherein f1 is less than or equal to half the sampling frequency f2 when the acoustic data string is created, To do.
[0211]
In the encoding device of the present invention, the converting means outputs a large number of low-frequency part spectrum data of f1 or less from the acoustic data sequence cut out by the cut-out means, and simultaneously outputs high-frequency part spectrum data exceeding f1. Then, the low-frequency spectrum data separated by the separating means is quantized and encoded, and the high-frequency spectrum data is encoded into auxiliary information representing the characteristics of the high-frequency portion of the frequency, The encoding means encodes the generated auxiliary information.
[0212]
Therefore, in a range where the total amount of information does not increase significantly, the low-frequency part is equivalent to downsampling, and the acoustic signal can be encoded with high quality so that the high-frequency part can be reproduced.
[0213]
Here, f1 is f2 / 4, the converting means converts the acoustic data into spectral data of 0 to 2 × f1 Hz, and the separating means includes low-frequency spectral data of 0 to f1 Hz and f1 ˜2 × f1 Hz high-frequency spectrum data, or the low-frequency spectrum data up to f1 Hz is composed of n spectral data, and the clipping means is 2 × n The number of pieces of acoustic data necessary to generate spectrum data is cut out, the converting unit converts the cut out acoustic data into 2 × n pieces of spectral data, and the separating unit uses n pieces of low-frequency spectrums. Data and n high-frequency spectrum data are separated, or the cut-out means includes n acoustic data corresponding to one frame which is a unit of encoding. And 2 × n pieces of spectrum data that are combined with n / 2 pieces of sound data belonging to each of two frames adjacent to the frame, and the converting means cuts out 2 × n pieces of sound data. Alternatively, the conversion may be performed by MDCT and converted to a spectrum of 0 to 2 × f1 Hz including 2 × n spectrum data.
[0214]
Furthermore, a decoding device according to the present invention is a decoding device that decodes encoded data input via a recording medium or a transmission medium, and includes a low-band encoded data included in the encoded data and a high-frequency encoded data. Extraction means for extracting each of the band encoded data, and the low band encoded data extracted by the extraction means are decoded and inverse quantized to output spectrum data of the low band below the frequency f1 Low-frequency dequantizing means, and auxiliary information decoding means for generating auxiliary information representing characteristics of high-frequency spectrum data by decoding high-frequency data extracted by the extracting means, A high frequency band inverse quantization means for outputting high band spectrum data based on the auxiliary information generated by the auxiliary information decoding means, and a low frequency band spectrum output by the low frequency band inverse quantization means. Synthesis means for synthesizing the data and the high-frequency spectrum data output by the high-frequency inverse quantization means, and inverse transform for inversely transforming the spectrum data synthesized by the synthesis means into acoustic data on the time axis And acoustic data output means for outputting the acoustic data inversely transformed by the inverse transform means in time order.
[0215]
In the decoding apparatus of the present invention, the extraction means extracts low-frequency encoded data and high-frequency encoded data from the input encoded data, and the low-frequency inverse quantization means has a frequency f1 or less. The spectrum data of the low frequency part of is output. The auxiliary information decoding means decodes the auxiliary information, and the high band inverse quantization means outputs high band spectrum data based on the auxiliary information. Therefore, it is possible to decode the amount of information that is significantly increased compared to the conventional amount from the same small amount of information as in the past, and to decode a high-quality sound signal.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a functional configuration of a broadcast system according to an embodiment of the present invention.
FIG. 2 is a diagram showing a change in state of an acoustic signal processed in the encoding device shown in FIG.
FIG. 3 is a flowchart showing an operation in a scale factor determination process of the first quantization unit shown in FIG. 1;
4 is a flowchart showing another operation in the scale factor determination process of the first quantization unit shown in FIG. 1; FIG.
FIG. 5 is a spectrum waveform diagram showing a specific example of auxiliary information (scale factor) generated by the second quantizing unit shown in FIG. 1;
6 is a flowchart showing an operation in auxiliary information (scale factor) calculation processing of the second quantization unit shown in FIG. 1; FIG.
7 is a diagram illustrating a position in a bit stream in which auxiliary information is stored by the stream output unit illustrated in FIG. 1. FIG.
FIG. 8 is a diagram illustrating another example when the stream output unit illustrated in FIG. 1 stores auxiliary information.
9 is a diagram showing a comparison of processing between the encoding device shown in FIG. 1 and a conventional example 1. FIG.
10 is a diagram showing a comparison of processing between the encoding device shown in FIG. 1 and a conventional example 2. FIG.
11 is a diagram showing a comparison of spectral data and characteristics between the encoding device shown in FIG. 1 and conventional examples 1 and 2. FIG.
12 is a flowchart showing a procedure in which a low-frequency part 1024 spectrum is copied in a forward direction to a high-frequency part by the second inverse quantization unit shown in FIG. 1;
13 is a flowchart showing a procedure of copying a low frequency band 1024 spectrum to a high frequency band in the reverse direction of the frequency axis direction by the second inverse quantization unit shown in FIG. 1;
14 is a spectrum waveform diagram showing a specific example of other auxiliary information (quantized value) generated by the second quantization unit shown in FIG. 1; FIG.
15 is a flowchart showing an operation in another auxiliary information (quantization value) calculation process of the second quantization unit shown in FIG. 1;
FIG. 16 is a spectrum waveform diagram showing a specific example of other auxiliary information (position information) generated by the second quantization unit shown in FIG. 1;
FIG. 17 is a flowchart showing an operation in another auxiliary information (position information) calculation process of the second quantization unit shown in FIG. 1;
FIG. 18 is a spectrum waveform diagram showing a specific example of other auxiliary information (sign information) generated by the second quantization unit shown in FIG. 1;
FIG. 19 is a flowchart showing an operation in another auxiliary information (sign information) calculation process of the second quantization unit shown in FIG. 1;
20 is a spectrum waveform diagram showing an example of a method for creating other auxiliary information (copy information) generated by the second quantization unit shown in FIG. 1; FIG.
FIG. 21 is a flowchart showing an operation in another auxiliary information (copy information) calculation process of the second quantization unit shown in FIG. 1;
22 is a spectrum waveform diagram showing a second example of a method for creating other auxiliary information (copy information) generated by the second quantization unit shown in FIG. 1; FIG.
FIG. 23 is a flowchart showing an operation in a second calculation process of other auxiliary information (copy information) of the second quantization unit shown in FIG. 1;
FIG. 24 is a block diagram showing a configuration of a conventional AAC encoding apparatus and decoding apparatus.
[Explanation of symbols]
1 Broadcasting system
300 Encoder
305 A / D converter
310 Acoustic data input unit
320 Conversion unit
330 Data separator
340 First quantization unit
345 Second quantization unit
350 1st encoding part
355 Second encoding unit
390 stream output unit
400 Decoding device
410 Stream input section
420 First decoding unit
425 Second decoding unit
430 First inverse quantization unit
435 Second inverse quantization unit
440 Inverse Quantized Data Synthesizer
480 Inverse conversion unit
490 Sound data output unit
495 D / A converter

Claims (12)

An encoding device for encoding acoustic data,
A cutting means for cutting out a fixed number of continuous acoustic data from the acoustic data string,
Conversion means for converting the cut out acoustic data into spectrum data on the frequency axis;
Separating means for separating the spectral data obtained by the conversion into low-frequency spectral data up to the f1 Hz and high-frequency spectral data higher than the f1 Hz;
A low frequency band encoding means for quantizing and encoding the separated low frequency band spectrum data;
Auxiliary information generating means for generating auxiliary information indicating the characteristics of the high frequency part frequency spectrum from the separated high frequency part spectrum data,
High frequency band encoding means for encoding the generated auxiliary information;
Output means for combining and outputting the code obtained by the low frequency band encoding means and the code obtained by the high frequency band encoding means,
The f1 is less than or equal to half the sampling frequency f2 when the acoustic data string is created,
The auxiliary information generating means, respectively per the higher band spectrum data and the low frequency band spectrum data divided into a plurality of groups, each group of the high frequency band, with a spectrum that is most approximate to the spectrum in the group An encoding device characterized in that information specifying a group of low-frequency spectrum data is generated as the auxiliary information. The auxiliary information generating means includes a distance on the frequency axis from the group break to the high-frequency spectrum peak in the group, and a frequency-axis from the low-band group break to the low-frequency spectrum peak. The encoding device according to claim 1, wherein a low-frequency spectrum having a minimum difference from the distance is specified. The auxiliary information generation means specifies a spectrum in a low frequency region where a difference value is minimum when an energy difference is taken with the same frequency width as a spectrum in the group in the high frequency region. The encoding device according to 1. The encoding device according to claim 1, wherein the information for specifying the spectrum of the low-frequency part is a number for specifying the group of the specified low-frequency part spectrum. A decoding device that decodes encoded data input via a recording medium or a transmission medium,
Extraction means for extracting low-band encoded data and high-band encoded data included in the encoded data,
Low-frequency band inverse quantization means for decoding low-frequency band encoded data extracted by the extraction means and dequantizing to output low-frequency spectrum data having a frequency of f1 Hz or less;
Auxiliary information decoding means for generating auxiliary information representing the characteristics of the high frequency band spectrum data by decoding the high frequency band data extracted by the extraction means,
High-band inverse quantization means for outputting high-band spectrum data based on the auxiliary information generated by the auxiliary information decoding means;
Combining means for synthesizing the low-frequency part spectrum data output by the low-frequency part inverse quantization means and the high-frequency part spectral data output by the high-frequency part inverse quantization means;
Inverse conversion means for inversely converting spectral data synthesized by the synthesis means into acoustic data on a time axis;
Acoustic data output means for outputting the acoustic data inversely transformed by the inverse transformation means in order of time,
Low-frequency auxiliary information, respectively per the high frequency portion spectral data and the low frequency band spectrum data divided into a plurality of groups, with each group of the high frequency band, the spectrum which is most approximate to the spectrum in the group Information identifying a group of partial spectral data ,
The high frequency band inverse quantization means generates predetermined noise in each group of high frequency bands based on the auxiliary information, and adds the generated noise to the spectrum data to generate high frequency spectrum data. Decoding device to perform. An acoustic data distribution system for distributing acoustic data compression-encoded into a low bit rate bit stream via a recording medium or a transmission medium,
An acoustic data distribution system comprising: the encoding device according to claim 1; and the decoding device according to claim 5. An encoding method for encoding acoustic data, comprising:
A step of cutting out a certain number of continuous acoustic data from the acoustic data sequence;
A conversion step of converting the cut out acoustic data into spectrum data on the frequency axis;
A separation step of separating the spectral data obtained by the conversion into low-frequency spectral data up to the f1 Hz and high-frequency spectral data higher than the f1 Hz;
A low band encoding step for quantizing and encoding the separated low band spectrum data;
Auxiliary information generating step for generating auxiliary information indicating the characteristics of the high frequency part frequency spectrum from the separated high frequency part spectrum data,
A high band encoding step for encoding the generated auxiliary information;
An output step of combining and outputting the code obtained in the low frequency band encoding step and the code obtained in the high frequency band encoding step;
The f1 is less than or equal to half the sampling frequency f2 when the acoustic data string is created,
In the auxiliary information generation step, each of the high-frequency part of each of the high-frequency part spectrum data and the low-frequency part spectral data divided into a plurality of groups has a spectrum that most closely approximates the spectrum in the group. The information which specifies the group of low frequency part spectrum data is generated as the auxiliary information. An encoding method characterized by things. A decoding method for decoding encoded data input via a recording medium or a transmission medium,
An extraction step for extracting low-band encoded data and high-band encoded data included in the encoded data,
A low-band inverse quantization step for outputting low-band spectrum data having a frequency of f1 Hz or less by decoding and inverse-quantizing the low-band encoded data extracted by the extraction step;
An auxiliary information decoding step for generating auxiliary information representing the characteristics of the high-frequency spectrum data by decoding the high-frequency data extracted in the extraction step;
A high-band inverse quantization step for outputting high-band spectral data based on the auxiliary information generated by the auxiliary information decoding step;
A synthesis step of synthesizing the low-frequency part spectrum data output by the low-frequency part inverse quantization step and the high-frequency part spectrum data output by the high-frequency part inverse quantization step;
An inverse transformation step for inversely transforming the spectral data synthesized by the synthesis step into acoustic data on a time axis;
An acoustic data output step for outputting the acoustic data inversely transformed by the inverse transformation step in time order,
Low-frequency auxiliary information, respectively per the high frequency portion spectral data and the low frequency band spectrum data divided into a plurality of groups, with each group of the high frequency band, the spectrum which is most approximate to the spectrum in the group Information identifying a group of partial spectral data ,
In the high frequency band inverse quantization step, based on the auxiliary information, a predetermined noise is generated in each group of the high frequency band, and is added to the spectrum data to generate high frequency spectrum data. Decryption method to do. A program used in an encoding device that encodes an input acoustic signal,
A program that causes a computer to function as each unit included in the encoding device according to claim 1. A program used in a decoding device for decoding encoded data input via a recording medium or a transmission medium,
A program that causes a computer to function as each unit included in the decoding device according to claim 5.   A computer-readable recording medium on which the program according to claim 9 is recorded.   A computer-readable recording medium on which the program according to claim 10 is recorded.

Images