JP2003029797A - Encoder, decoder and broadcasting system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an encoder, a decoder and a broadcasting system for transmitting a high-quality encoded acoustic signal. SOLUTION: The encoder 111 for encoding an inputted acoustic signal is provided with an acoustic signal input part 112 for segmenting the inputted acoustic signal into frame for each fixed time, a converting part 113 for converting the segmented acoustic signal to a frequency spectrum data group, a spectrum data consolidating part 114 for consolidating at least two pieces of spectrum data to less pieces of spectrum data representative in these spectrum data in a specified frequency band included in the frequency spectrum data group and outputting the data, a quantizing part 115 for quantizing the merged spectrum data,and an encoding part 116 for encoding the quantized data and outputting encoded data.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high quality audio encoding and decoding technique for digital audio data.

[0002]

2. Description of the Related Art Currently, various acoustic compression methods for compressing and encoding acoustic data have been developed. MPEG-2 Advanced Au
dio Coding (hereinafter abbreviated as MPEG-2 AAC)
Is one of the methods. For details of MPEG-2 AAC, refer to "ISO / IEC 13818-7 (MPEG-2 Advanced Audio Codin
g, AAC) ”, but below,
A portion related to the present invention will be summarized and described.

First, the procedure of encoding and decoding by the conventional encoder and decoder will be described. First, the encoding device cuts out digital acoustic data as an input signal at regular time intervals, and extracts the cut-out sample data on the time axis from an MDCT (Modified Discrete Cosine Transfor
Convert to spectrum data on the frequency axis by m). The spectrum data on the frequency axis is classified into a plurality of groups, and then normalized and quantized for each group. The quantized data is Huffman encoded into a stream format. The coded signal obtained by the Huffman coding is converted into an MPEG-2 AAC bitstream and output. The bit stream output from the encoding device is transmitted to the decoding device via a transmission medium such as a broadcast wave or a communication network, or recorded on a recording medium such as an optical disk such as a CD or a DVD, a semiconductor, a hard disk or the like. .

In the decoding device, the bit stream coded by the coding device is received through the transmission line or reproduced from the recording medium and input, and the coded signal is extracted from the input bit stream. To decrypt. Specifically, first, the extracted coded signal is converted from a stream format to a data processing format, decoded into quantized data, and the decoded quantized data is further dequantized. Then
The spectrum data on the frequency axis obtained by the inverse quantization is IMDCT (Inverse Modified Discrete Cosine).
Transform) is transformed into sample data on the time axis. The sample data on the time axis obtained by the inverse transformation are sequentially combined and output as digital acoustic data.

In the actual MPEG-2 AAC encoding process, other than this, Gain Control and TNS (Temporal) are also used.
Noise Shaping), psychoacoustic model, M / S Stereo, Inte
There are cases where tools such as nsity Stereo and Prediction are used, and bit reservoirs are used.

The reproduction band after encoding is one measure of how much the sound quality of the acoustic data encoded by the encoding device and transmitted to the decoding device is maintained. . For example, when the sampling frequency of the input signal is 44.1 kHz, the reproduction band is 22.05 kHz
z. This 22.05 kHz, or 22.05 kHz
By encoding a wideband acoustic signal close to Hz without deterioration and transmitting all the encoded data,
High-quality sound signal transmission can be achieved. However, the width of the playback band affects the number of spectral data,
The number of spectrum data affects the amount of transmitted data. For example, when the sampling frequency of the input signal is 44.1 kHz, the spectrum data of 1024 samples is 22.05.
It is necessary to transmit all spectrum data of 1024 samples in order to secure the reproduction band of 22.05 kHz corresponding to the data of kHz. For this purpose, it is necessary to efficiently encode the acoustic signal and keep the data amount within the transfer rate range of the transmission path.

However, in consideration of a low transfer rate transmission line such as a mobile phone, it is not realistic to actually transmit all the spectrum data of 1024 samples because the data amount is too large. In other words, if you try to transfer all spectrum data in this playback band with a data amount that matches the transfer rate, the amount of information that can be assigned to each frequency band will be small, and as a result, the effect of quantization noise will increase, There is a problem that the sound quality is deteriorated due to the encoding.

For this reason, not only MPEG-2 AAC but also many acoustic signal coding systems perform auditory weighting on spectrum data and do not transmit low priority data, so that an efficient acoustic signal is generated. Transmission is realized. According to this, in order to improve the coding accuracy of the low-frequency part having a high auditory priority, a sufficient amount of data is assigned to the coded information of the low-frequency part, and the high-frequency part of low priority is not transmitted. There is a high probability that

[0009]

[Problems to be Solved by the Invention] However, MPE
Despite such ingenuity in the G-2 AAC system, further improvement in quality and improvement in compression efficiency are demanded for audio signal encoding. That is, there is an increasing demand for transmitting high-frequency acoustic signals even at low transfer rates.

[0010]

In order to meet the above-mentioned demands, an encoding apparatus of the present invention is an encoding apparatus for encoding an input acoustic signal, wherein the input acoustic signal is framed at regular intervals. A conversion means for cutting out into a frequency spectrum data group and converting it into a frequency spectrum data group, and within a specific frequency band included in the frequency spectrum data group, in accordance with a predetermined function, two or more spectrum data are represented in a smaller number And an encoding unit that quantizes and encodes the integrated spectrum data and outputs the encoded data.

The decoding device of the present invention is a decoding device for decoding input coded data to restore an acoustic signal, wherein the input coded data is decoded and dequantized to obtain a frequency. Dequantizing means for converting into spectral data group, and expanding means for expanding the spectral data contained in the frequency spectral data group into two or more spectral data respectively by using a predetermined inverse function in a specific band. , And inverse conversion means for converting the expanded spectrum data into an acoustic signal on the time axis and outputting the acoustic signal.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION A broadcasting system 100 according to an embodiment of the present invention will be described below with reference to the drawings. FIG. 1 is a block diagram showing the configuration of a broadcasting system 100 of this embodiment. The broadcasting system 100 encodes an acoustic signal according to the encoding method of the present invention and delivers it via a satellite broadcasting radio wave, and a broadcasting station 110 and a broadcasting satellite 13.
It is composed of a plurality of homes 120 that receive satellite broadcast radio waves relayed by 0, decode and reproduce encoded audio data included in the received broadcast radio waves, and enjoy sounds and voices such as movies and music.

<Broadcasting Station> The broadcasting station 110 is a coding device 1.
11 and the transmitter 118. Encoding device 111
Reduces the data amount of an acoustically encoded bitstream that has been conventionally transmitted, and if the data amount is the same, outputs an acoustically encoded bitstream that allows a decoding device to restore an audio signal with higher sound quality than before. . Encoding device 11
1 is realized by a program executed by a general-purpose computer or hardware such as a dedicated circuit board or LSI, and an acoustic signal input unit 112, a conversion unit 113, a spectrum data integration unit 114, a quantization unit 11
5, a coding unit 116 and a stream output unit 117. The acoustic signal input unit 112 is, for example, 44.1 kH.
Digital acoustic data, which is an input signal sampled at a sampling frequency of z, is cut out for every 1024 continuous samples. The coding unit of 1024 samples is called "frame". Specifically, the extraction method is about 22.7 mse in which 1024 samples of digital acoustic data with a sampling frequency of 44.1 kHz are input.
For each c, 512 samples before and after that are overlapped, and 2048 samples of digital acoustic data are cut out.

The conversion unit 113 converts the sample data on the time axis cut out by the acoustic signal input unit 112 into spectrum data on the frequency axis. Converter 113
Is the time axis data of the extracted 2048 samples,
Although the spectrum data of 2048 samples is converted using MDCT, since the spectrum data is symmetrical in MDCT, only one 1024 samples is used in the subsequent processing. The conversion unit 120 converts the spectrum data of 1024 samples after conversion obtained in this way into
Classify into multiple groups. Each of these groups simulates a critical band in human hearing, and each of the groups is called a “scale factor band”. Each of the scale factor bands is defined to include spectral data of one sample or more (practically a multiple of 4). In MPEG-2 AAC,
Sampling frequency is 44.1kHz, 1 frame is 1
With 024 samples, the number of scale factor bands is set to 49 per frame. Further, the number of spectral data (samples) included in each scale factor band is not uniform among the scale factor bands and differs depending on the frequency. That is, the low frequency part includes a small number of spectral data, and the higher frequency part includes more spectral data.

The spectrum data integration unit 114 targets a specific band of the 1024 samples of spectrum data output by the conversion unit 113, and uses a predetermined function to convert the spectrum data of two or more samples into a smaller number of spectra. A processing unit that integrates the data. The spectral data integration unit 114 obtains 1 obtained by the conversion.
Of the spectrum data of 1024 sample frames,
For 512 samples in the high frequency band, the spectral data of 2 samples are integrated into the spectral data representing one integrated value. Specifically, the 512 samples in the low frequency band are directly output to the quantizing unit 115, and the 512 samples in the high frequency band are output.
For each sample, the absolute values of the spectrum data of two samples that are continuous in the front and back on the frequency axis are compared, and only the spectrum data having the larger absolute value is output to the quantization unit 115 as the integrated value of the two samples. As a result, the spectrum data of the two samples in the high frequency band are integrated into the spectrum data showing one integrated value. together,
The spectrum data integration unit 114 generates integration information indicating that continuous two samples of the high frequency band 512 samples are integrated into one sample, and outputs the integrated information to the encoding unit 116.

The quantizing unit 115 normalizes the spectrum data of 768 samples in one frame, which is composed of 512 samples in the low frequency band and 256 samples in the high frequency band, output from the spectral data integrating unit 114, one for each scale factor band. The coefficient is used to normalize the spectral data in the scale factor band, and each frame is quantized so as not to exceed a predetermined number of bits. This normalization coefficient is called "scale factor". In addition, specifically, the quantizing unit 115 outputs a bit amount for one frame to the spectrum data in each scale factor band when the spectrum data of the frame is finally converted into an acoustic bit stream. Each scale factor is determined while making an approximation so that it falls within the range of the transmission amount of the transmission path, and normalization and quantization are performed using the determined scale factor. Here, the quantization result of each spectrum data is called "quantization value". The quantization unit 115 outputs the scale factor and the quantization value used for normalization as the result of this quantization.

The encoder 116 Huffman-encodes the scale factor and the quantized value input from the quantizer 115, converts them into a predetermined stream format, and generates an encoded signal. In this process, for the scale factor, the difference in the scale factor is sequentially calculated between two continuous scale factor bands, and the scale factor in the leading scale factor band and the calculated difference are Huffman-encoded. Furthermore, the encoding unit 116 includes the spectrum data integration unit 114.
Huffman coding is performed on the integrated information input from the above, the converted information is converted into a predetermined stream format, and the integrated information is output to the stream output unit 117. The stream output unit 117 adds header information and other sub-information as necessary to the encoded signal input from the encoding unit 116 to make MPEG-2.
The integrated information is converted into an AAC encoded bit stream, and the encoded integrated information is stored in an area in the bit stream that is ignored by the conventional decoding device or its operation is not specified, and is output. The transmission device 118 transmits the encoded bit stream output from the stream output unit 117 of the encoding device 111 to the broadcasting satellite 130 by placing it on a satellite broadcast radio wave.

<Home> Each home 120 has a broadcasting satellite 130.
It is provided with a receiving device 121, a decoding device 122, and a speaker 129 for receiving a broadcast radio wave via, receiving a broadcast radio wave, extracting an encoded bit stream from the received broadcast radio wave, decoding the encoded bit stream, and reproducing an acoustic signal. The receiving device 121 uses the STB (Se
t Top Box) to receive the broadcast radio waves of satellite broadcasting, extract the encoded bit stream from the received radio waves, and output the encoded bit stream to the decoding device 122. Like the encoding device 111, the decoding device 122 is realized by a program executed by a general-purpose computer or hardware such as a dedicated circuit board or LSI, and outputs an encoded signal representing sound from an input encoded bit stream. , And decodes the integrated information indicating the contents of the integration, expands the integrated spectrum data based on the decoded integrated information, and restores the acoustic data. The decoding device 122 includes a stream input unit 123 and a decoding unit 1.
24, inverse quantization unit 125, spectrum data expansion unit 12
6, an inverse conversion unit 127 and an acoustic signal output unit 128.

The stream input section 123 is used by the receiving device 12
The encoded signal representing the acoustic data and the integrated information generated by the encoding device 111 are extracted from the encoded bit stream extracted by 1 and output to the decoding unit 124, respectively. The decoding unit 124 decodes the Huffman-encoded data input from the stream input unit 123 from the stream format into quantized data, and outputs the quantized data to the dequantization unit 125. At the same time, the Huffman-coded integrated information is decoded and output to the spectrum data expansion unit 126. The inverse quantization unit 125 inversely quantizes the quantized value of one frame of 768 samples, which includes 512 samples of the low frequency band and 256 samples of the high frequency band, decoded by the decoding unit 124, and scales each scale factor band. The factor is used to restore the spectral data of 512 samples in the low band part and 256 samples in the high band part integrated value. The spectral data expansion unit 126 holds in advance various expansion methods associated with the integrated information, and based on the integrated information input from the decoding unit 124, the spectral data expanded by the inverse quantization unit 125 The integrated value included is expanded in association with the spectrum data selected as the integration target in the encoding device 111, and the high-frequency part 512 samples are restored to restore the spectrum data of 1024 samples per frame. The inverse transformation unit 127 converts the spectrum data on the frequency axis into IMDCT according to MPEG-2 AAC.
(Inverse Modified Discrete Cosine Transform) is used to convert to sample data on the time axis. The acoustic signal output unit 128 sequentially combines the sample data on the time axis obtained by the inverse conversion unit 127 and outputs it as digital acoustic data. The speaker 129 inputs the digital acoustic data restored by the decoding device 122, and reproduces music and voice according to an analog acoustic signal obtained by D / A converting the input acoustic data. The broadcast satellite 130 receives the broadcast radio wave from the broadcast station 110 and sends it back to the ground.

The broadcasting system 10 configured as described above
The operation of the encoding device 111 in 0 will be described with reference to FIGS. 2A to 6. FIG. 2A is a diagram showing an example of a simplified waveform of acoustic data on the time axis cut out by the acoustic signal input unit 112 shown in FIG. 1. Figure 2
(B) is a diagram showing an example of spectrum data on the frequency axis when the acoustic data on the time axis is MDCT-transformed by the conversion unit 113 shown in FIG. 1. Note that FIG.
In FIG. 2A and FIG. 2B, the sample data and the spectrum data are shown as continuous waveforms, but in reality, both are discrete data groups.

As shown in FIG. 2A, a signal representing a sound is represented by a waveform of a voltage value that changes with time. Figure 2
In (a), the voltage value on the vertical axis corresponds to the sound intensity at that time. Generally, a sound waveform contains many frequency components. When a signal for a certain period of time is cut out from such an acoustic signal and the signal is MDCT-converted, as a conversion result, the ratio of each frequency component included in the cut-out signal has a positive and negative value in FIG. ) Is obtained as spectral data.

On the basis of the characteristics of such an acoustic signal and the characteristics of human hearing for it, MPEG-2 AA
In C, signal processing is performed using the scale factor band as one processing unit for quantization. FIG. 3 is a diagram showing an example of a scale factor band in which spectrum data is classified by the conversion unit 113 shown in FIG. In the figure, each spectrum data is represented by a bar graph. In MPEG-2 AAC, the number of scale factor bands included in one frame is determined by the difference between the LONG block and the SHORT block, and according to the sampling frequency of the input audio data. The LONG block is an MD by the conversion unit 113.
The one with a CT conversion length of 2048 samples, SHO
The RT block has a conversion length of 256 samples. For example, when the sampling frequency is 44.1 kHz in the LONG block as in this embodiment, the number of scale factor bands included in one frame is 49. In MPEG-2 AAC, the number of samples (spectral data) included in each scale factor band is also determined according to the frequency. Specifically, as shown in the figure, it is determined that a low frequency range includes a small number of samples, and a high frequency range includes a large number of samples. This is because, in general, human auditory characteristics are sensitive to low-frequency and mid-frequency components of an acoustic signal, and thus higher precision is required for low-frequency and mid-frequency encoding and decoding. Is based. Using such a scale factor band, the quantization unit 115 normalizes and quantizes the spectrum data included in the same scale factor band using the same scale factor.

The quantizing unit 115 determines the scale factor while calculating the amount of bits used for transmitting coded data for one frame when quantizing each scale factor band. At this time, if the calculated data amount is very large compared to the transmission rate of the transmission path, in order to reduce the data amount after encoding, the quantized value of each spectrum data becomes a small value, Determine the scale factor. This is remarkable especially in the high range. As a result, if the quantization unit 115 performs normalization and quantization as in the conventional case in such a case,
It often happens that the quantized value continuously becomes "0" in the high frequency region. However, even if the quantized value of the spectrum data is “0” in this way, the corresponding encoded data amount does not become “0”. Therefore, in the coding apparatus 111 of the present embodiment, the spectrum data output from the conversion unit 113 is subjected to the following integration processing by the spectrum data integration unit 114 before being quantized by the quantization unit 115. To do. FIG. 4A is a diagram showing an example of the spectrum data before integration which is the output of the conversion unit 113 shown in FIG. FIG. 4B is a diagram showing an example of spectrum data after the integration processing by the spectrum data integration unit 114 shown in FIG. 1. As shown in FIG. 4 (a), the spectrum data integration unit 114 displays 1 frame 1
Of the 024 samples of spectrum data, the 512 samples on the low frequency side are output to the quantizer 115 as they are, and the 512 samples on the high frequency side are integrated values from the spectrum data of the two samples continuous on the frequency axis. Seeking,
As shown in FIG. 4B, the continuous spectrum data of two samples is integrated into the spectrum data of one sample representing the calculated integrated value and output to the quantization unit 115. Here, the absolute values of the spectrum data of two consecutive samples are compared, and the value of the spectrum data with the larger absolute value is used as the integrated value. In FIG. 4A, the spectral data of the larger absolute value of the two consecutive samples is
It is shown by hatching. As shown in FIG. 4B, the integration processing of the spectrum data integration unit 114 causes the FIG.
The spectral data of the 512 high-frequency samples shown in
It can be seen that they are integrated into 256 samples consisting of integrated values only.

As described above, by integrating two consecutive samples into one sample, it is possible to reduce the data amount after encoding by the amount of the spectral data thinned by the integration. In addition, since the number of spectrum data to be quantized can be significantly reduced by the integration, it is possible to prevent the quantization value from becoming “0” in the adjustment of the scale factor in the high frequency band scale factor band. The effect is that you can do it.

In this way, by using the spectrum data of the larger absolute value of the continuous two samples as the integrated value, when the continuous two samples are both "0" as the spectrum value in the high frequency region. In addition to reducing the amount of data by not transmitting one of the repeated "0" as data, if one of the two consecutive samples has a value other than "0", then that value is There is an advantage that it can be left as an integrated value (that is, as data) without omission.

The above-mentioned integration processing is executed by the data processing in the spectrum data integration unit 114 in the following procedure. FIG. 5 is a spectrum data integration unit 1 in the spectrum data integration process shown in FIG.
14 is a flowchart showing the operation of No. 14. In the figure, i and j are ordinal numbers indicating the numbers of the respective spectrum data. The register is a storage area for temporarily storing the value of the variable. First, the spectrum data integration unit 1
Reference numeral 14 denotes spectrum data spectral [i], i = 0, of 1024 samples per frame from the conversion unit 113.
1, ..., 1023 are input (S501). These spectral data are i = 0, 1, ...,
It is stored in the storage area spectral [i] represented by the one-dimensional array specified by 1023. Next, the spectrum data integration unit 114 assigns “512” to the register i and the register j, and performs the subsequent processing as 51
This is performed for the second (the top of the high frequency band) and subsequent spectrum data (S502). Further, it is determined whether i is less than "1024" (S503). i is “1024”
If it is less than, since the integration process has not been completed yet,
Absolute value of i-th spectrum data abs (spect
ral [i]) is calculated and substituted in the register a, and the absolute value abs of the (i + 1) th spectrum data is obtained.
Calculate (spectral [i + 1]) and register b
To (S504). Here, the absolute value of the 512th spectrum data is assigned to the register a and the absolute value of the 513th spectrum data is assigned to the register b.

Further, the i-th spectrum data spectral [i] is previously substituted into the spectral [j] which is an area for storing the j-th spectrum data, and then the i-th spectrum data stored in the register a is stored. Absolute value of spectral data abs (spectra
l [i]) and (i + 1) stored in register b
Absolute value of the th spectral data abs (spectr
al [i + 1]). That is, spec
ral [j = 512] to spectral [i = 51
2] is substituted, and then the absolute value of the 512th spectrum data stored in the register a and the absolute value of the 513th spectrum data stored in the register b are compared. If the absolute value of the (i + 1) th spectrum data stored in the register b is larger than the absolute value of the i-th spectrum data stored in the register a as a result of the comparison, the jth spectrum Spectral [j], which is an area for storing data
Is the (i + 1) th spectrum data
The value of tral [i + 1] is overwritten (S505). That is, as a result of the comparison, 51 stored in the register b
If the absolute value of the third spectrum data is larger,
In spectral [j = 512], spectral [i + 1 = 51, which is the 513th spectrum data.
3] is overwritten. As a result, of the two adjacent spectrum data, the value of the spectrum data having the larger absolute value is written to spectral [j = 512]. Next, the spectrum data integration unit 11
4 increments i by "2" and j by "1" (S506), and returns to step S503.
As a result, i = 514 and j = 513.

In this way, the spectrum data integrating unit 114 repeats the processing of two successive consecutive steps by incrementing i by "2" and incrementing j by "1" and repeating the processing of the following steps S503 to S506. The absolute values of the spectrum data are compared, and the spectrum data with the larger absolute value is used as the array j.
Write to ral [j]. As a result, step S
When i becomes “1024” or more in 503, the storage area spectral [j], j = 512, 5
13, ..., 767, all the spectral data of 256 samples consisting of only the integrated value of two consecutive samples of the high frequency part 512 samples are held. Therefore, when i becomes “1024” or more in step S503, the spectrum data integration unit 114 stores the spectrum data from i = 0th to 511th stored in the storage area spectral [i] and the storage area spectral. Spectral data spectral consisting of j = 512th to 767th spectral data held in [j]
[K], k = 0, 1, ...
(S507), and the integration process in the frame except for the creation of the integration information ends.

As a result of the above integration processing, it is possible to reduce the spectrum data of 512 samples in the high frequency band to 256 samples and the spectrum data of 1024 samples in one frame to 768 samples. As described above, the spectrum data integration unit 114 can easily reduce the data amount of the high frequency spectrum data by a simple process. Furthermore, the spectral data of 768 samples in which two consecutive two in the high range are integrated into one is originally
By sequentially allocating a predetermined number of samples from the low frequency side to the scale factor bands allocated to the spectrum data of the frame 1024 samples, the number of scale factor bands to be quantized is reduced, As a result, not only the load on the quantization processing of the quantization unit 115 can be reduced, but also the number of scale factor bands is reduced.
The number of scale factors to be transmitted is reduced, and the data amount of the encoded signal can be further reduced. In this way, the number of samples of spectrum data can be greatly reduced by integration as compared with the conventional method, so if the data amount of the encoded bitstream is the same as the conventional method, each quantization value and scale factor will be increased. , The number of bits can be assigned, and encoding can be performed with higher accuracy.

The reason why the integration processing of the spectrum data is applied only to the spectrum data of the high frequency side half is that the sound quality deterioration of the reproduced sound caused by the integration of the spectrum data is more easily perceived on the low frequency side. Yes, so the low range remains the same and only the high range is integrated.

FIG. 6 is a diagram showing an example of integrated information 500 generated when the spectral data integration process of FIG. 4 is performed. The integrated information 500 includes a header 510 and one or more blocks 520. Header 510
Is a part that represents information about the integrated information 500 and includes items of integrated information ID 511, frame number 512, and data length 513. Integrated information ID 5
In the 11th item, an ID indicating that this data is the integrated information 500 is recorded. In frame number 512,
A frame number for indicating which frame the integrated information 500 represents, that is, a frame number for identifying the frame is recorded. Integrated information 5 is included in the data length 513.
00 has a variable length, the first block 520 to the last block 52 of the first block 520 included in the integrated information 500.
The bit length up to the end of 0 is recorded.

Each block 520 shows specific contents of the integration process, and each time the integration method in one frame is switched, the integration method in that portion is recorded. Specifically, each block 520 is divided into a part for designating an application range of the integration process and a part for designating a detailed integration method within the specified integration range. In each block 520, the part for designating the application range of the integration process is the designation method 5.
21, the start 522, and the end 523.
The designation method 521 is a portion showing what is designated by the integrated range, and either “sfb” or “SD” is selected and recorded. When "sfb" is selected here, the integrated range indicated by each item of the start 522 and the end 523 is described by the scale factor band number. When "SD" is selected, each item of the start 522 and the end 523 is described by the spectrum data number. In the item of start 522, the start position of the integrated range is described with a value according to the designation method 521. In the item of end 523, the end position of the integrated range is described with a value according to the designation method 521. For example,
If the serial numbers from “0” to “1023” are given to each spectrum data of each frame 1024 samples, and it is assumed that the high frequency band 512 samples are integrated according to the same integration method, the item of the designation method 521 is "S
If it is “D”, “511” is recorded in the item of start 522, and “1023” is recorded in the item of end 523.

The part for designating the integration method in each block 520 includes items of integration number 524, spectrum selection method 525, integration value 526 and weighting 527. In the item of integration number 524, the number of spectrum data to be integrated is recorded. Here, “2” is recorded, which indicates that the two pieces of spectral data are integrated. In addition, the spectrum selection method 525 includes a method of selecting spectrum data to be integrated, for example, whether the number of spectrum data indicated by the item of the integration number 524 is continuously selected or every other data is selected. The method is recorded. Here, "continuous" is recorded. Further, in the integrated value 526, a method for obtaining the integrated value from the selected spectrum data is recorded. Here, the “absolute maximum value” is recorded, and it is shown that, of the spectrum data selected as the integration target, the value of the spectrum data having the largest absolute value becomes the integration value. In the item of weighting 527, whether or not to perform some weighting, such as multiplying each spectrum data selected as the integration target by a coefficient, is recorded. When weighting is performed, the designation of the spectrum data to be weighted and the numerical value of the weighting coefficient are recorded. Here, “none” is recorded.

By referring to the integrated information 500 recorded in this way, on the decoding device 122 side, two consecutive spectral data among the spectral data of 512 samples in the high frequency band are absolute in this frame. It can be known that the value of the larger spectrum data is integrated into one spectrum data as an integrated value, and the spectrum data close to the original spectrum data can be restored corresponding to this.

In this example, one block 52
Since the integration range is specified up to the end of the frame by 0, the block 520 included in the integration information 500.
Is only one illustrated, but a plurality of blocks 520 are used when a plurality of integration methods are used in one frame.
Is included. Further, in this example, the integrated information 50 including at least one block 520 for one frame.
Although it has been described that 0 is generated, items that are predetermined between the encoding device 111 and the decoding device 122 may be deleted from the integrated information 500. For example,
For every frame, only the high-frequency part 512 samples,
In the case where the application range of the integration process using the same integration method is set, a part that specifies the application range of the integration process, that is, the integration information 500 in which the items of the specification method 521, the start 522, and the end 523 are deleted is generated. Good. Further, also in the following items, for example, in a frame in which weighting is not performed or in the integration processing target range, “none” is not described in the item of the weighting 527, but the item of the weighting 527 itself is deleted from the integration information 500. It is also possible to insert the item of the weighting 527 describing the weighting coefficient into the integrated information 500 and transmit it only in the frame to be transmitted and weighted or in the range to be integrated.

Integrated information 500 generated as described above
Is Huffman-encoded, converted into a format for a stream, and then converted from an encoded signal MPEG-
2 It is ignored in a conventional decoding device in a coded bitstream of 2 AAC or is incorporated in a region whose operation is not specified.

FIG. 7A is a diagram showing an example of the data structure of an MPEG-2 AAC audio coded bit stream 600 in which integrated information 500 is incorporated. Figure 7 (b)
Is another MPEG-2 with integrated information 500 incorporated.
It is a figure which shows the example of a data structure of AAC audio encoding bit stream 610. The integrated information 500 of the present invention is incorporated in each of the hatched portions in FIGS. 7A and 7B. As shown in FIG. 7A, the acoustic coded bit stream 600 includes items of a header 601 and a coded signal 602, and further, a Fill Element.
Or area 60 such as Data Stream Element (DSE)
Includes 3. The header 601 contains information about the audio coded bitstream 600, for example, an identifier indicating that this data is an audio coded bitstream conforming to MPEG-2 AAC, a data length of the audio coded bitstream 600, and a code. The frame number corresponding to the encoded signal 602, the number of scale factor bands corresponding to the encoded signal 602, and the like are recorded. In the item of the coded signal 602, the spectrum data integrated by the spectrum data integration unit 114 is quantized and coded, and the format-converted coded signal is recorded.

The Fill Element is "0" in order to adjust the data length of the acoustically encoded bit stream 600 to a fixed length.
Header information including a Fill Element identifier indicating that it is a Fill Element and bit number data indicating the bit length of the entire Fill Element 603 is provided. For example, the area 603 is the Fill Element
If the integrated information 500 is stored in the area 603, the integrated information 500 is recorded in an area where “0” is originally recorded below the header information. As described above, when the integrated information 500 is stored in the Fill Element, the conventional decoding device does not recognize the encoded signal to be decoded and ignores it.

The DSE is an MPE for future expansion.
This is an area in which only the physical structure such as bit length is defined by the G-2 AAC standard.
Similar to Element, header information including a DSE identifier indicating that the following data is a DSE and bit number data indicating the bit length of the entire DSE is provided. For example, when the area 603 is the DSE, when the integrated information 500 is stored in this area 603, the integrated information 500 is added to the portion which becomes the data area of the DSE following the header information.
To store. When the integrated information 500 is stored in this DSE, even if the integrated information 500 is read by the conventional decoding device, the operation of the decoding device for the read integrated information 500 is not specified, so the decoding device No processing corresponding to this is performed.

Therefore, the integrated information 50 is stored in the above area.
If 0 is stored, the integrated information 500 will not be decoded as an audio coded signal even if the audio coded bit stream by the coding apparatus 111 of the present invention is input to the conventional decoding apparatus. It is possible to prevent noise and the like due to the fact that the information 500 cannot be correctly decoded. However, in the high frequency band, as a result of the integration, the integrated value spectrum data is shifted to the low frequency side scale factor band by the amount of the spectral data that is thinned out to every two continuous data, so that the high frequency band is reproduced. It is unavoidable that the band becomes narrower and the sound quality of the high frequency part of the restored sound becomes different from the originally sampled acoustic signal. Although the DSE is described here as the area 603 located at the end of the coded audio bitstream, the DSE may be located between the header 601 and the coded signal 602, or even in the coded signal 602. Good.

Further, in the above, the integrated information 500 is set to MP
The acoustic coded bitstream according to the EG-2 AAC standard is stored in an area that is ignored by a conventional decoding device, but the acoustic coded bitstream 610 is stored.
When only the decoding device 122 of the present invention is output, the integrated information 500 may be incorporated in a predetermined area 611 in the header information 601, or the encoded signal 60 may be included.
2 in a predetermined position other than the area 603 (for example, the area 61
2) or may be incorporated over both the header information 601 and the encoded signal 602.
(For example, it may extend over the area 611 and the area 612.) In order to store the integrated information 500 in the audio coded bitstream 610, a continuous area is secured in the header 601 and the coded signal 602. You don't have to. For example, the coded signal 602 may be incorporated over a predetermined area, that is, an area 612 and an area 613.

The decoding device 122, which has input the above acoustically encoded bit stream via satellite broadcast radio waves, extracts an encoded signal from the acoustically encoded bit stream and decodes it. However, since the extracted coded signal has the spectrum data of the high band part integrated, the decoding device 122 decodes and dequantizes the extracted coded signal, and then samples 256 high band parts. Processing for expanding the spectral data of the above into 512 samples. FIG. 8A is a diagram showing an example of the spectrum data before expansion which is the output of the inverse quantization unit 125 shown in FIG. FIG. 8B is a diagram showing an example of the spectrum data after the expansion processing by the spectrum data expansion unit 126 shown in FIG. This expansion method corresponds to the spectral data integration method shown in FIG. 4, and one frame of 768 samples, which is the result of inverse quantization by the inverse quantization unit 125, is converted into one.
This is a method of expanding the spectrum data of 024 samples. The spectrum data expanding unit 126 keeps the low band portion 512 samples of the spectrum data of 768 samples of one frame shown in FIG.
Each of the 6 samples is expanded into the spectral data of 2 samples continuous on the frequency axis to generate the spectral data of 512 samples in the high frequency region. Comparing the expanded spectrum data shown in FIG. 8B with the spectrum data of the original sound shown in FIG. 4A shows that the spectrum data similar to the original sound is roughly restored. .

The above-mentioned expansion processing is executed by the spectrum data expansion unit 126 by specifically performing data processing in the following procedure. FIG. 9 is a flowchart showing the operation of the spectrum data expanding section 126 in the spectrum data expanding process shown in FIG. In the figure, i and j are ordinal numbers indicating the numbers of the respective spectrum data. Spectral data expansion unit 126
From the inverse quantization unit 125, inv_spectra of spectrum data that is the result of inverse encoding and inverse quantization.
Input l [j], j = 0, 1, ..., 767 (S
1001). These spectral data are respectively j
Storage area inv_spectral [j] represented by a one-dimensional array specified by = 0, 1, ..., 767
It is stored in. Next, the spectrum data integration unit 1
14 assigns “512” to the register i and the register j, and performs the subsequent processing for the 512th (the top of the high frequency band) and subsequent spectrum data (S1002). Further, it is determined whether j is less than "768" (S1).
003). If j is less than "768", the expansion process is not yet completed, so i = 512, 513, ...
, Inv_ in the temporary storage areas tmp [i] and tmp [i + 1] represented by the one-dimensional array corresponding to 1023.
Substitute the value of spectral [j] (S100
4). First, here, tmp [512] and tmp [5]
13] and inv_spectral [512] are stored.

Further, the spectral data expansion unit 126
Increments i by "2" and j by "1" (S1005), and then returns to step S1003. As a result, i = 514 and j = 513.

In this way, inv_spectral [513] is changed to tmp [5] by incrementing i by "2" and incrementing j by "1" and repeating the processing of the following steps S1003 to S1005.
14] and tmp [515], and inv_sp
electrical [514] is tmp [516] and tmp
The inverse quantizer 1 is expanded to [517].
Each of the continuous spectrum data input from 25 is sequentially expanded into two continuous temporary storage areas, two by two. As a result, when j becomes “768” or more in step S1003, the temporary storage area tmp [i], j = 512, 513, ..., 10
In 23, the spectrum data of 512 samples in the high frequency part having the same value for each two consecutive samples are held. Therefore, the spectrum data expansion unit 126
When j becomes “768” or more in step S1003, the temporary storage area tmp [i], i = 5
The spectrum data held in 12, 513, ..., 1023 are respectively stored in the storage area inv_spect.
ral [i], i = 512, 513, ..., 1023
(S1006), and the low-frequency spectrum data inv_spectral that has not been subjected to expansion processing
Inv in combination with [i], i = 0, 1, ..., 511
_Spectral [i], i = 0, 1, ..., 10
23 is output to the inverse conversion unit 127 (S1007), and the expansion processing in the frame ends.

As described above, according to the coding apparatus 111 of the present embodiment, the spectral data of 1024 samples in one frame is integrated into 768 samples, so that the coding apparatus 111 can perform quantization and coding processing. The load can be reduced, and the acoustically encoded bit stream can be transmitted without difficulty on the transmission path. Further, since the decoding device 122 side can restore the spectrum data of 1024 samples in the entire reproduction band from the integrated spectrum data of 768 samples of one frame, it is possible to restore high-quality sound data. Further, in the broadcasting system 100 according to the present embodiment, the number of samples per frame to be transmitted is smaller than that in the conventional case, and therefore, if the acoustic coded bit stream having the same data amount is transmitted, the number is more than that in the past. The amount of information per sample can be increased and transmitted. Therefore, each sample data in the audio coded bit stream can be represented with higher accuracy, and the decoding device 122
On the side, it is possible to reproduce a sound that is more faithful to the original sound.

Further, the coding apparatus 1 according to the present embodiment
11 and the decoding device 122 are only the addition of the spectrum data integration unit 114 and the spectrum data expansion unit 126 to the conventional encoding device and decoding device, so that the configuration of the existing encoding device and decoding device is greatly improved. It can be realized without changing to.

The broadcasting system 100 of the present embodiment
Has been described as a satellite digital broadcasting system using the broadcasting satellite 130, but the broadcasting system of the present invention is not limited to such a BS digital broadcasting system, and may be a CS digital broadcasting system using a communication satellite. Needless to say, it may be a digital broadcasting system using terrestrial waves. Further, the encoding device and the decoding device of the present invention can be applied not only to the transmission side device and the reception side device of the broadcasting system, but also to a content distribution system using a bidirectional communication network such as the Internet or a telephone. It can also be applied to the transmitting side device and the receiving side device of the system. Furthermore, the encoding device of the present invention can be applied to a recording device that records an acoustic signal on a recording medium such as a CD (Compact Disc), and the decoding device can be applied to a reproducing device that reproduces the acoustic signal from the recording medium. it can.
The processing by the encoding device 111 and the decoding device 122 can be realized by software as well as hardware, or can be realized by a configuration in which one part hardware and the rest are software.

In this embodiment, MPEG is used.
-2 Although AAC has been described as an example of the prior art, the present invention can be applied to other existing acoustic coding schemes and can also be applied to other existing acoustic coding schemes. it can.

Further, although the spectrum data integration unit 114 of the present embodiment sets the spectrum data of the high frequency side half (512 samples) as the integration processing application range while keeping the low frequency band half (512 samples) as it is, Of course, the method of defining the processing application range is not limited to this example. Figure 10 (a)
FIG. 9 is a diagram showing another example of the integration processing application range in one frame. For example, as shown in FIG. 10A, the low-frequency part 256 samples from the beginning of the frame are directly used by the quantizer 11
It is also possible to apply the integration process from the lower half side than the half, such as outputting to 5, and integrating 768 consecutive high-frequency parts.
Further, FIG. 10B is a diagram showing still another example of the integration processing application range in one frame. For example, in FIG.
As shown in (b), 768 samples in the low frequency band are directly output to the quantizing unit 115 from the beginning of the frame, 256 continuous samples in the high frequency band are integrated, and so on. May be applied. Alternatively, the integration process may be applied to all 1024 samples in one frame. Furthermore, FIG.
It is a figure which shows the further another example of the integration process application range in 1 frame. For example, as shown in FIG. 10C, the continuous spectrum data of 256th to 319th and the continuous spectrum data of 768th to 1023th in one frame may be the integrated application range of the frame. . That is, the integration process may be applied to a plurality of regions that are not continuous in the frequency axis direction.

Further, as the applicable range of the above integration processing, different ranges may be set for each frame. FIG. 11 (a)
FIG. 8 is a diagram showing an example of an integration processing application range of the spectrum data integration unit 114 when different integration processing application ranges are set for each frame. As shown in FIG. 11A, the integration processing may be applied to all 1024 samples in one frame in a certain frame, and the integration processing may not be applied in a certain frame at all. FIG. 11B is a diagram showing another example of the integration processing application range of the spectrum data integration unit 114 when different integration processing application ranges are set for each frame. Further, as shown in FIG. 11B, the spectrum data integration unit 114 outputs the low frequency band 512 samples to the quantization unit 115 as they are in a certain frame, and the high frequency region 512 samples as the application range of the integration process,
In the next frame, 768 samples in the low-frequency part may be output as they are, and 256 samples in the high-frequency part may be set as the applicable range of the integration process, so that the applicable range may be determined for each frame.

Further, in the present embodiment, the spectrum data integration unit 114 has been described as generating the integration information 500 of FIG. 6 and designating the integration processing application range in each frame, but the integration information 500 is not always generated. It is not necessary to perform, for example, “in the high-frequency part 51 in the odd-numbered frame
2 samples are set as the application range of the integration process, and in the even-numbered frame, the 768th and subsequent consecutive 256 samples are set as the application range. " When the integrated processing application range is defined, the application range need not be specified in the integrated information 500.

Furthermore, in this embodiment, the integrated information 5
00 includes one or more blocks 520, the specific contents of the integration process are shown in the block 520, and each time the integration method is switched in one frame, the integration method in that portion is recorded. The integrated information 500 is not limited to this example. For example, if it is determined in advance how to perform the integration processing within one frame, the integration information 5
As 00, only a 1-bit flag indicating whether or not to perform integration processing for each frame is sufficient. Also, when performing the same processing as the integration processing for the immediately preceding frame,
The generation of the integrated information 500 for the frame may be omitted.

In the present embodiment, the spectrum data integration unit 114 integrates the continuous spectrum data of two samples into one sample, but the present invention is not limited to this example. FIG. 12A is a diagram showing another example of a combination of integrated spectrum data. For example, as shown in FIG. 12A, the spectral data of three consecutive samples may be integrated into one sample, or a larger number of consecutive samples may be integrated into one sample.

In addition, discontinuous samples may be combined. FIG. 12B is a diagram showing still another example of a combination of integrated spectrum data. For example, in FIG.
In 2 (b), the spectral data of every two consecutive samples is integrated into one sample. In a similar manner, every third consecutive sample or more may be combined, or every second, third, or more consecutive samples may be combined. Further, the sample data to be integrated may be overlapped with at least one sample data to be integrated. FIG. 12C is a diagram showing still another example of a combination of integrated spectrum data. Figure 1
2 (c) is an example in which the spectral data of three consecutive samples is integrated into one sample, and the spectral data at the beginning of the consecutive three samples overlaps with the last sample of the immediately preceding set.

The combination of samples to which the above-mentioned integration processing is applied may be different for each frame or for each band. For example, in one frame, two consecutive samples may be integrated into one sample, but in another frame, three consecutive samples may be integrated, or in a low-range 512 sample, two consecutive samples may be combined. Although they are integrated, it may be determined for each band such that four consecutive samples are integrated in the high frequency 512 samples. Also, the combination of integrated samples may be determined for each scale factor band.
In this, the number of integrated spectrum data may be changed according to the frequency of spectrum data. For example, a larger number of samples may be integrated in a higher scale factor band.

Furthermore, the number of samples to be integrated may depend on the actual spectral data of the frame. For example, when the spectrum data of 10 samples in the high frequency region is continuously “0”, the spectrum data of these 10 consecutive samples should be integrated into one sample of which the spectrum value is “0”. You may In addition to the number of samples to be integrated, the method of calculating the integrated value, the range of integration processing, the combination of integrated samples, the presence / absence of weighting, and their numerical values should be determined according to the spectrum data of the frame. You may When performing the integration process in this way, the spectrum data integration unit 114 holds the integration method in each case in advance in association with the pattern of the spectrum assumed in each frame, and the spectrum of each frame. The pattern of the spectrum is identified by converting the data into a function, and if a corresponding pattern appears as a result of the identification, the corresponding integration method is applied. Further, some of the above items regarding the integration method are determined in advance by the encoding device and the decoding device, and the integration information 50 is set for these items.
It is also possible to generate integrated information 500 and describe it in the information defined according to actual spectrum data without generating 0.

In this embodiment, the sample data having the maximum absolute value among the sample data to be integrated has been described as the integrated value, but the integrating method of the present invention is not limited to this example. FIG. 13A is a diagram showing another example of the case where the integrated value is calculated from the spectrum data of two consecutive samples. For example, as described above, the spectral data S (A) and S (B) to be integrated are weighted by multiplying the coefficients α and β, respectively, and then, the value of the spectral data having the maximum absolute value is calculated as Spectral data S
It may be an integrated value of (A) and S (B). Further, as described above, the average value of the integration target sample data S (A) and S (B) may be obtained as an integrated value, and the average value may be the respective spectrum data S (A) and S (B). It may be calculated from the absolute value of. Further, this average value may be calculated from each sample data S (A), S (B) to be integrated with a weighting coefficient. Furthermore, as described above, the order of the sample data to be the integrated value may be determined in advance. That is, for example, the value of the lower spectrum data of the two consecutive spectrum data may be set as the integrated value. Of course, the value of the spectrum data with the lower frequency may be set as the integrated value.

FIG. 13B is a diagram showing still another example of the case where the integrated value is calculated from the spectrum data of two consecutive samples. As shown in FIG. 13B, the spectral data integration unit 114 further refers to the spectral data adjacent to both sides of the sample data to be integrated with respect to the integrated value calculated as described above to obtain the integrated value. It may be adjusted. For example, in the example of FIG.
Spectral data S (C), S (D), S (E), and S (F) of two samples respectively adjacent to the low-frequency side and the high-frequency side of each spectrum data S (A), S (B) to be integrated. ) With reference to the adjacent spectrum data S (C), S
If any of (D), S (E) and S (F) exceeds a predetermined threshold value, a predetermined weighting coefficient “1.5” is multiplied. In this,
The number of adjacent spectra to be referred to is not limited to 2 samples, and may be 1 sample or 3 or more samples. Further, the spectrum data adjacent to only the high band side may be referred to, or only the low band side may be referred to. Further, the predetermined weighting coefficient is not limited to "1.5", and it is not necessary that it be "1" or more. For example, if the values of adjacent spectrum data are very large, the spectrum data indicating the integrated value may be masked. In such a case, it may be "0", for example.

The method of calculating the integrated value is not limited to these. For example, it is possible to obtain a spectrum value as an integrated value by performing a function conversion different from the above, which is predetermined, on each sample data to be integrated. The method of calculating the integrated value may be different for each frame, for each band, or for each scale factor band.

The method of calculating the integrated value may be determined in advance between the encoding device and the decoding device,
It may be described in the integrated information 500. In particular, it is preferable to describe, in the integrated information 500, a method of expanding the spectrum data in the decoding device by using these integrated values.

In the present embodiment, the structure of the scale factor band is the same as that before the integration of the spectrum data, but of course it may be changed before and after the integration. FIG. 14A is a diagram showing an example of high-frequency band spectrum data and scale factor bands before integration. FIG. 14B is a diagram showing an example of the relationship between the integrated high band spectrum data and the scale factor band. FIG. 14C is a diagram showing the relationship between the integrated high frequency band spectrum data and the scale factor band in the present embodiment. In addition,
In FIG. 14A to FIG. 14C, the low frequency band spectrum data is out of the integration processing application range, and therefore the spectrum data and the scale factor band do not change, and therefore are omitted. Further, for the sake of explanation, the scale factor band having the scale factor band number “40” is set as the leading scale factor band of the high frequency band 512 spectrum data for the sake of description. In this embodiment,
The spectrum data integrated by the spectrum data integration unit 114 according to the flowchart shown in FIG. 5 is directly distributed to the scale factor band set by the conversion unit 113. Therefore, the spectrum data integrated by the spectrum data integration unit 114 shifts to the previous scale factor band by the amount of the spectrum data reduced by the integration, as shown in FIG. 14C, and the scale factor band in the high frequency band. Will be reduced. In the present embodiment, by integrating the spectrum data in this way, by reducing not only the number of quantized values transmitted as an encoded signal, but also the number of scale factors transmitted as an encoded signal, The data amount of the encoded signal can be greatly reduced.

However, the integration method of the present invention is not limited to the configuration of the scale factor band described above, and FIG.
As shown in, the configuration of the scale factor band within the integrated processing application range may be changed. The number of spectrum data included in each scale factor band is MPEG.
-2 AAC, but this is changed in the present invention. For example, when two spectral data are integrated into one spectral data, the number of spectral data included in each scale factor band is also 2 respectively. It may be changed to one-half. By doing this, more accurate quantization can be performed in each scale factor band within the integrated processing application range, so the number of scale factors cannot be reduced, but the number of quantized values can be reduced. By reducing the amount, it is possible to reduce the data amount of the encoded signal and to transmit more accurate acoustic data. In this case, the change of the scale factor band may be determined in advance by the encoding device and the decoding device, or may be encoded as integrated information.

In the present embodiment, as shown in FIG. 9, one spectrum data indicating the integrated value is expanded into two spectrum data. However, for example, even if only one of the two data is copied, Good. That is, the spectrum data expansion unit 126 copies each sample of the high frequency side 256 samples including the spectrum data indicating the integrated value to one of the two continuous samples (integrated value) on the frequency axis to the other and outputs 512 samples. May be generated. Alternatively, the integrated value spectrum data may be multiplied by a weighting coefficient and then copied to the other. Further, each of the two spectrum data obtained by expanding or copying may be multiplied by a weighting coefficient.

The spectral data expansion unit of the present invention is
If there is integrated information, the spectrum data may be expanded according to it, or may be expanded according to a unique expansion method regardless of the presence or absence of integrated information. Further, it is not limited to these.

[0066]

The coding device of the present invention is a coding device for coding an input acoustic signal, and the input acoustic signal is cut into frames at regular time intervals and converted into a frequency spectrum data group. A conversion means and an integration for integrating and outputting two or more spectrum data into a smaller number of spectrum data representing the spectrum data according to a predetermined function in a specific frequency band included in the frequency spectrum data group. And a coding means for quantizing and coding the integrated spectrum data and outputting the coded data.

According to the encoding device of the present invention, the integrating means reduces the data amount of the encoded audio signal to be transmitted by integrating the spectrum data based on a predetermined function,
Since it is possible to comfortably transmit the coded acoustic signal even in a transmission path of a low transfer rate, and integrates two or more spectrum data within a specific frequency band included in the spectrum data group, for example, hearing By integrating the high frequency band which is hard to be recognized,
It is possible to minimize the recognition of deterioration of sound quality due to integration. Further, even in the specific frequency band in which the integration is performed, the spectrum data representative of them is encoded with respect to two or more spectrum data. The sound quality is improved as much as the data is encoded as the representative spectrum. As a result, it is possible to achieve both reduction of the data amount of the coded audio signal and transmission of the coded audio signal of high quality.

Further, according to the encoding device of the present invention, the spectral data of 2 or more becomes a smaller number of spectral data than that even in the specific frequency band in which the integration is performed. Even when the encoded audio signal by the encoding device is input to the conventional decoding device, it can be decoded. In this case, it is undeniable that there is a slight difference in sound quality between the original sound and the sound reproduced from the encoded sound signal due to the decrease in the number of spectrum data due to the integration. If the band is selected as a high frequency band that is hard to be perceptually perceived, it is possible to minimize the perceived change in sound quality due to the integration of the spectrum data.

According to another encoding device of the present invention,
Since it is integrated for each of two or more spectral data that are continuous or discontinuous on the frequency axis, unlike the case of cutting data in a certain band, it is possible to evenly thin out and transmit any band, and the decoding device can output the original sound. However, it is possible to reproduce a high-quality sound that is more faithful to the original sound while significantly reducing the data amount of the encoded audio bit stream transmitted through the transmission path.

According to still another encoding device of the present invention,
Since the integrating means integrates the two or more spectrum data for each band by using an integrating method determined according to the spectrum in each frequency band, the spectrum data is integrated by using an integrating method more suitable for the original sound. be able to. As a result, it is possible to selectively reduce the data that is expected to be unnecessary in order to restore the original sound, reduce the data amount of the coded acoustic bit stream transmitted through the transmission path, and reduce the data amount. There is an effect that it is possible to minimize the deterioration of the sound quality due to.

The decoding device of the present invention is a decoding device for decoding input coded data to restore an acoustic signal, and decoding and dequantizing the input coded data to obtain a frequency. Dequantizing means for converting into spectral data group, and expanding means for expanding the spectral data contained in the frequency spectral data group into two or more spectral data respectively by using a predetermined inverse function in a specific band. , And inverse conversion means for converting the expanded spectrum data into an acoustic signal on the time axis and outputting the acoustic signal.

According to the decoding device of the present invention, since the encoded data generated by the encoding device can be restored to the original number of spectrum data, it is possible to restore a specific frequency band which has not been transmitted conventionally. As a result, a spectrum close to the original spectrum is restored, and an acoustic signal having a wide frequency band can be restored from less encoded data.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a broadcasting system of the present embodiment.

2 (a) is a diagram showing an example of a simplified waveform of acoustic data on a time axis cut out by the acoustic signal input unit shown in FIG. 1. FIG. FIG. 3B is a diagram showing an example of spectrum data on the frequency axis when the acoustic data on the time axis is MDCT-transformed by the conversion unit shown in FIG. 1.

FIG. 3 is a diagram showing an example of a scale factor band in which spectrum data is classified by the conversion unit shown in FIG.

4A is a diagram showing an example of spectrum data before integration which is an output of the conversion unit shown in FIG. FIG. 2B is a diagram showing an example of spectrum data after integration processing by the spectrum data integration unit shown in FIG. 1.

5 is a flowchart showing an operation of a spectrum data integration unit in the spectrum data integration process shown in FIG.

6 is a diagram showing an example of integrated information generated when the spectral data integration process of FIG. 4 is performed.

FIG. 7 (a) MPEG-2 with integrated information incorporated
It is a figure which shows an example of the data structure of an AAC audio encoding bit stream. (B) Other MPEG-2 AA with integrated information
It is a figure which shows the example of the data structure of C audio encoding bit stream.

8 (a) is a diagram showing an example of spectrum data before expansion which is an output of the inverse quantization unit shown in FIG. 1. FIG. FIG. 2B is a diagram showing an example of spectrum data after the expansion processing by the spectrum data expanding unit shown in FIG. 1.

9 is a flowchart showing an operation of a spectrum data expanding section in the spectrum data expanding process shown in FIG.

FIG. 10A is a diagram showing another example of the integration processing application range in one frame. (B) It is a figure which shows the further another example of the integration process application range in 1 frame. (C) It is a figure which shows the further another example of the integration process application range in 1 frame.

FIG. 11A is a diagram showing an example of an integration processing application range of a spectrum data integration unit when a different integration processing application range is determined for each frame. (B) It is a figure which shows the other example of the integration process application range of the spectrum data integration part in the case of defining different integration process application range for every frame unit.

FIG. 12A is a diagram showing another example of a combination of integrated spectrum data. (B) It is a figure which shows the further another example of the combination of the integrated spectrum data. (C) It is a figure which shows other example of the combination of the integrated spectrum data.

FIG. 13A is a diagram showing another example of a case where an integrated value is calculated from continuous two-sample spectrum data. (B) It is a figure which shows another example at the time of calculating an integrated value from the spectrum data of two continuous samples.

FIG. 14A is a diagram showing an example of high-frequency band spectrum data and scale factor bands before integration. (B) It is a figure which shows an example of the relationship between the high frequency part spectrum data after integration, and a scale factor band. (C) It is a figure which shows the relationship between the high frequency part spectrum data after integration and a scale factor band in this Embodiment.

[Explanation of symbols]

100 broadcasting system 110 broadcasting stations 112 Acoustic signal input section 113 converter 114 spectral data integration unit 115 Quantizer 116 encoder 117 Stream output unit 118 transmitter 120 households 121 Receiver 122 Decoding device 123 Stream input section 124 Decoding unit 125 inverse quantizer 126 Spectral data expansion unit 127 inverse converter 128 acoustic signal output section 129 speaker 130 broadcasting satellites

Claims (22)

[Claims] 1. A coding device for coding an input acoustic signal, wherein the input acoustic signal is cut out into frames at regular time intervals and converted into a frequency spectrum data group, and the frequency spectrum data. Integrating means for integrating and outputting two or more spectrum data into a smaller number of spectrum data representative of the spectrum data within a specific frequency band included in the group, and quantizing and encoding the integrated spectrum data. And a coding unit that outputs the coded data. 2. The encoding device according to claim 1, wherein the spectrum data integrated by the integrating means is obtained by a function that evaluates two or more spectrum data and integrates them into less spectrum data. . 3. The code according to claim 2, wherein the function is a function for determining a smaller number of spectrum data representing each of the two or more spectrum data continuous on the frequency axis. Device. 4. The function according to claim 2, wherein for each of the two or more spectral data that is discontinuous on the frequency axis, a smaller number of spectral data representative of these is determined. Encoding device. 5. The function is a function for selecting, from the two or more spectrum data, the spectrum data having the maximum absolute value as a result of the integration of the two or more spectrum data. 2. The encoding device according to 2. 6. The function according to claim 2, wherein the function is a function that sets spectrum data having an average value of the two or more spectrum data as a value to an integration result of the two or more spectrum data. Encoding device. 7. The function is a function that, for each of the two or more spectrum data, sets the spectrum data at a predetermined position in a row in which the spectrum data are arranged on a frequency axis as an integration result. The encoding device according to claim 2. 8. The encoding apparatus according to claim 2, wherein the function is a function that determines the integration result after weighting the two or more spectrum data. 9. The encoding apparatus according to claim 2, wherein the integrating unit selects the function for each frame and integrates the two or more spectrum data. 10. The code according to claim 2, wherein the integrating means integrates the two or more spectrum data for each band by using the function determined according to the spectrum in each frequency band. Device. 11. The function determines the integration result of the two or more spectrum data as a value that depends on both the two or more spectrum data and the spectrum data in the vicinity thereof. Encoding device described. 12. The conversion means divides the frequency spectrum data group after conversion into a plurality of groups each having a predetermined number of spectrum data included therein, and the integration means determines a target of integration. Arranged spectrum data and a plurality of spectrum data after integration in the order of frequency, the plurality of arranged spectrum data is divided into each group by the predetermined number, the encoding means, The coding apparatus according to claim 1, wherein the spectrum data divided into each group is quantized using a parameter having each group as a unit. 13. The integrating means further comprises integrated information generating means for generating integrated information indicating an integrating method of the two or more spectrum data, and the encoding means encodes the integrated information and the code. The data is embedded in the digitized data and is output.
Encoding device described. 14. The integrated information generating means does not generate the integrated information when the integrating method by the integrating means is the same as the immediately preceding frame.
Encoding device described. 15. A decoding device for decoding input coded data for each frame of an acoustic signal cut out at regular time intervals to restore an acoustic signal, wherein the input coded data is decoded. And dequantizing means for dequantizing and converting to frequency spectrum data group, and spectrum data included in the frequency spectrum data group into two or more spectrum data in a specific band using a predetermined function. A decoding device comprising: a expanding means for expanding and an inverse converting means for converting the expanded spectrum data into an acoustic signal on a time axis and outputting the acoustic signal. 16. The decoding apparatus according to claim 15, wherein the function is a function for copying each spectrum data included in the frequency spectrum data group and expanding the spectrum data into two or more spectrum data. 17. The decoding device according to claim 15, wherein the function is expanded after weighting the spectrum data included in the frequency spectrum data group. 18. The dequantizing means further extracts integrated information indicating an integrating method of two or more spectrum data from the encoded data, and the expanding means based on the extracted integrated information. ,
The decoding device according to claim 15, wherein the spectral data is expanded. 19. The decoding apparatus according to claim 18, wherein the expanding means expands the spectrum data based on the integrated information of the immediately preceding frame when the integrated information is not extracted. . 20. A broadcasting system comprising a coding device for coding an input acoustic signal, and a decoding device for receiving coded data from the coding device and decoding the data to restore the acoustic signal. The encoding device cuts the input acoustic signal into frames at constant time intervals and converts the acoustic signal into a frequency spectrum data group, and a predetermined frequency band included in the frequency spectrum data group. An integrating means for integrating and outputting two or more spectrum data into a smaller number of spectrum data representing those spectrum data according to one function, and quantizing and encoding the integrated spectrum data to obtain encoded data. And a decoding unit that decodes and dequantizes the input coded data. Dequantization means for converting into frequency spectrum data group, and expansion for expanding the spectrum data included in the frequency spectrum data group into two or more spectrum data in a specific band by using a predetermined second function A broadcasting system comprising: means and an inverse conversion means for converting the expanded spectrum data into an acoustic signal on a time axis and outputting the acoustic signal. 21. A program for an encoding device that encodes an input acoustic signal, the converting means extracting the input acoustic signal into frames at regular time intervals and converting the frames into frequency spectrum data groups. Integrating means for integrating and outputting two or more spectrum data into a smaller number of spectrum data representing the spectrum data according to a predetermined function within a specific frequency band included in the frequency spectrum data group; A program for causing a computer to function as an encoding unit that quantizes and encodes the generated spectrum data and outputs the encoded data. 22. A program for a decoding device for decoding input coded data to restore an acoustic signal, wherein the input coded data is decoded and dequantized into a frequency spectrum data group. Dequantizing means for transforming, expanding means for expanding the spectrum data included in the frequency spectrum data group into two or more spectrum data in a specific band by using a predetermined function, and the expanded spectrum A program for causing a computer to function as an inverse conversion unit that converts data into an acoustic signal on a time axis and outputs the acoustic signal.