JP3466507B2 - Audio coding method, audio coding device, and data recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To produce the coded data of high sound quality in real time and with no sound break by switching the using bit allocation means to execute the bit allocation and then the coding so as to perform the due processing by means of the prescribed one of plural bit allocation means. SOLUTION: A sub-band analyzing means 102 divides the inputted digital audio signals into 32 frequency components. A grouping means 111 divides the 32 frequency components into the number of groups that is designated by the throughput control information 121. An allocatable bit arithmetic means 113 decides each allocatable bit number to an aural psychological model bit allocation means 110 and a band output adaptive bit allocation means 109 based on the total allocatable bit number. Each of sub-band signals is coded by a quantizing/coding means 106 and an auxiliary information coding means 107, and a bit stream forming means 108 forms and outputs a bit stream.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a speech coding system,
The present invention relates to a speech coding apparatus and a data storage medium, and in particular, a speech coding method and a speech coding apparatus using a subband coding method used in the MPEG (Motion Picture Experts Group) method, and the speech coding apparatus. The present invention relates to a data storage medium that stores a program for executing the coding method.

[0002]

2. Description of the Related Art In recent years, with the advancement of multimedia in personal computers and the spread of the Internet, an environment has been established in which moving images and audio such as MPEG can be reproduced by software on personal computers (hereinafter also referred to as PCs). MPEG
The range of use of encoded data such as is expanding. However, with regard to an encoder that creates encoded data, it is still mainstream to use expensive hardware.
In addition, there are some that make encoded data by software, but it takes a lot of time because it takes a lot of processing time that is many times the real time of the playback time of the moving image or audio to be encoded. It takes time and labor, and has not been widely spread.

For this reason, in particular, in order for general user of a personal computer to be able to easily and inexpensively create coded data, there is a demand for real-time creation of coded data by software processing.

An example of a conventional speech coding system will be described below. FIG. 11 shows M, which is standardized by ISO / IEC11172-3, as a coded data format for voice.
It is a block diagram of a PEG audio encoder. Figure 1
In 1, the input digital audio signal is divided into 32 frequency components by the subband analysis unit 202, and the scale factor is calculated by the scale factor extraction unit 203 for each subband signal to make the dynamic range uniform. Further, the input digital audio signal is subjected to the fast Fourier transform in the FFT means 204.
(FFT: Fast Fourier Transform), and using this result, the psychoacoustic analysis unit 205 uses a signal-to-mask ratio (SM
R) using the relational model of values, the bit allocation means 20
The number of bit allocation for each subband signal is determined by 6. Quantization / encoding means 207 quantizes / encodes each subband signal in accordance with the number of bits assigned to each subband signal. Then, the bit stream forming unit 209 forms the bit stream by combining the header information and the auxiliary information encoded by the auxiliary information encoding unit 208, and outputs the bit stream.

Since this conventional speech coding system is a coding system which performs coding for each band (sub-band) by utilizing uneven distribution of power of each band, each sub-cycle using the psychoacoustic model. The bit allocation to the band signal affects the sound quality. In addition, since it was standardized for the purpose of using storage media, it is suitable for improving the sound quality of encoded data, but is not suitable for encoding in real time, and the psychoacoustic model that affects sound quality is extremely It has a large amount of calculation.

[0006]

The conventional speech coding system and speech coding apparatus are configured as described above, and are suitable for creating high-quality coded data for a storage medium. However, since the use of the psychoacoustic model requires a lot of processing capacity, the current CPU capacity for processing by software is not suitable for real-time processing on a PC. Also, a high-performance CPU capable of real-time processing
Even when it is operated on a PC equipped with, there is a possibility that real-time processing becomes impossible when the CPU occupancy of other applications becomes large, and as a result, sound interruption occurs. There is a potential problem.

The present invention has been made to solve the above problems and has a CPU processing capability of a personal computer.
And a speech coding method, a speech coding apparatus, and a speech coding apparatus capable of realizing, by software processing, creation of high-quality coded data that does not cause sound interruption in real time regardless of the CPU occupancy of other applications. An object of the present invention is to provide a data storage medium that stores a program for executing encoding.

[0008]

A speech coding system according to claim 1 of the present invention is a speech coding system in which a digital audio signal is divided into a plurality of frequency bands and coding is performed for each band. Generating bit allocation information for each of the divided bands, having a plurality of bit allocation means each having a different processing amount, and selecting a predetermined bit from among the plurality of bit allocation means based on external control information. The bit allocation means to be used is switched so that the processing is performed by using the allocation means, the bit allocation is executed, and the data is encoded to obtain the control information from the outside.
The central processing unit that can be occupied at the time of encoding.
Based on the above load value, using the load value that represents the
Using each bit allocation means on the central processing unit,
Data table that stores each processing amount when encoding
Processing of the occupable central processing unit
The bit allocation means is selected so as not to exceed the amount .

The speech coding method according to claim 2 of the present invention is the speech coding method according to claim 1, wherein the load value can be occupied to perform coding processing.
Monitoring means for monitoring the throughput of the above central processing unit
The processing amount control information from is used .

A speech coding method according to claim 3 of the present invention is the speech coding method according to claim 1, wherein the bit allocation processing is performed by the bit allocation means.
As high efficiency that can realize high sound quality of encoded data
Uses a highly efficient bit allocation method for bit allocation
Processing and processing using the above high-efficiency bit allocation method
Bit allocation is performed with low load and less processing compared to
The process using the low load bit allocation method is performed .

According to a fourth aspect of the present invention, in the voice encoding method according to the first aspect, the bit allocation means used during the encoding is switched off.
It is the smallest unit that can be replaced into an audio signal.
This is done in frame units .

A speech coding system according to claim 5 of the present invention is the speech coding system according to claim 1 in which the subbands of each band divided into a plurality of frequency bands.
Signal is a predetermined number of subband signals
Group into groups consisting of
Independent bit allocation processing is performed for each group
The bit allocation information for the band is generated .

A speech coding method according to claim 6 of the present invention is the speech coding method according to claim 5, wherein the grouping is performed by the number of groups or groups.
The number of consecutive subband signals in the
The number specified by the control information from the outside, or the above
To the number specified based on the throughput control information from the visual means
It is performed variably so that

According to a seventh aspect of the present invention, there is provided a voice encoding method according to the sixth aspect, wherein the number of subband signals is changed by changing the audio signal.
It is performed in units of frames, which is the minimum unit that can be decoded into .

According to an eighth aspect of the present invention, in the voice encoding method according to the seventh aspect , bit allocation is not performed at the time of grouping.
At least one group is provided .

A speech coding method according to claim 9 of the present invention is the speech coding method according to claim 5, wherein the subband signals are grouped into a low band.
Subband signals grouped into groups belonging to
In contrast, processing using the above high-efficiency bit allocation method
Bit allocation for subband signals within a group
On the other hand, grouped into high band groups
Low-load bit allocation for the selected subband signal
Subband signals within a group by processing using the method
Bit allocation to. Moreover, the contract of the present invention
The speech coding method according to claim 10 is the above claim 5.
In the speech coding method of
Determine the number of assignable bits for the bit allocation means
Assignable bit operation means for
Bands belonging to each group as a percentage of the entire group
Using what added weighting based on each characteristic,
The number of assignable bits for the entire group is set to each group.
In which it takes swing component to separate bit allocation means for each flop.

A speech coding method according to claim 11 of the present invention is the speech coding method according to claim 10 , wherein the weight is based on the characteristics of each band belonging to each group.
Based on a specified minimum audible limit for each band
Weighting is used. Further, claim 12 of the present invention
The voice encoding system according to claim 1 is the voice according to claim 10.
In the encoding system, the characteristics of each band belonging to each group
Based on the weighting of the input digital audio signal
Belong to each group obtained by subband analysis on
Weighting based on the subband signal level of each frequency band
It's just something.

According to a thirteenth aspect of the present invention, in the voice encoding method according to the tenth aspect , weighting based on the characteristics of each band belonging to each group is applied to the input digital audio signal. Linear transformation to
The weighting is performed based on the spectrum signal levels belonging to each group obtained as described above.

A speech coding system according to a fourteenth aspect of the present invention is the speech coding system according to the fifth aspect , in which the signal level belonging to each group is a predetermined threshold value.
For the above high level signals, the high efficiency bit allocation
Bit allocation is performed by processing using the allocation method.
The signal level belonging to each group is a predetermined threshold
For the following low level signals, the above low load bit allocation
Bit allocation is performed by processing using the allocation method .

Further, the speech encoding method according to claim 15 of the present invention, in the audio coding method of the claim 14, the signal level belonging to each group, the input digital
Obtained by subband analysis of digital audio signal
It is a subband signal level .

Further, the speech encoding method according to claim 16 of the present invention, in the audio coding method of the claim 14, the signal levels belonging to said each group is subjected to a linear transformation to the input digital audio signal Obtained Spec
Torr is a signal level der shall.

Further, the speech encoding method according to claim 17 of the present invention, in the audio coding method of the claim 14, the signal level belonging to each group, each given
This is the minimum audible limit value for each band .

The speech coding system according to claim 18 of the present invention is the speech coding system according to any one of claims 3, 9 and 14, wherein the high efficiency bit allocation is performed.
The processing using the method is based on a predetermined psychoacoustic model.
Signal-to-mask ratio value relationship.
Processing using the network allocation method is applied to multiple frequency bands.
Predetermined minimum audible limit for each band for divided signal levels
The value is taken into consideration .

According to a nineteenth aspect of the present invention, in the speech encoding system according to the eighteenth aspect , the psychoacoustic model is MPEG (Motion Picture Exp).
ertsGroup) is a psychoacoustic model specified by the ertsGroup .

The speech coding method according to claim 20 of the present invention is the minimum speech signal which can be decoded into the audio signal in the speech coding method according to claim 4 or 7.
The frame that is the unit is MPEG (Motion Picture Exp
erts Group) is the frame specified by.

The speech coding system according to claim 21 of the present invention is the speech coding system according to claim 1, wherein the bit allocation means is provided for each divided band.
On the other hand, based on the information output from the specified psychoacoustic model,
To generate bit allocation information, and N (N =
1, 2, 3 ...) Once per frame, the specified auditory psychological mode
Bit allocation information based on information output by Dell
And the bit allocation information above was not generated.
For Laem, it was output from the above psychoacoustic model.
Based on the information and the signal information of each divided band above.
It generates the allocation information and encodes it .

A speech coding method according to a twenty-second aspect of the present invention is the speech coding method according to the first aspect, which has a psychoacoustic model capable of controlling the processing amount stepwise.
Then, based on external control information, the above psychoacoustic model is used.
Control the processing volume of the
Bit for each band so that processing is performed using
To generate allocation information .

A speech coding method according to a twenty-third aspect of the present invention is the speech coding method according to the first aspect, wherein a plurality of psychoacoustic models each having a different processing amount are provided.
Then, based on external control information,
Process using a specified psychoacoustic model from the physical models
Switch the psychoacoustic model to be used so that
Then, the bit allocation information for each band is generated .

According to a twenty-fourth aspect of the present invention, there is provided a speech coding system according to the first aspect, which is equivalent to the performance of the central processing unit in which the coding process is executed.
Accordingly, multiple bit allocations are required during initialization before the encoding processing operation.
Allocation means, or each processing load of multiple psychoacoustic models
The value information is output to the outside .

The speech coding system according to claim 25 of the present invention is the speech coding system according to claim 24.
Multiple bit allocations that are output as external information.
Reliable measure or each processing load value of multiple psychoacoustic models
Information is output in descending or ascending order .

A speech coding apparatus according to claim 26 of the present invention is the speech coding device according to any one of claims 1 to 25.
Speech coding is performed using the described speech coding method.

A recording medium according to claim 27 of the present invention has recorded therein the steps of the audio encoding method according to any one of claims 1 to 25 . It is a thing.

[0033]

[0034]

[0035]

[0036]

[0037]

[0038]

[0039]

[0040]

[0041]

BEST MODE FOR CARRYING OUT THE INVENTION A speech coding system and a speech coding apparatus according to embodiments of the present invention will be described below with reference to the drawings. (Embodiment 1) Here, for example, in the case of a coding method in which an input signal is divided into a plurality of frequency components and coding is performed for each band (subband) by utilizing uneven distribution of power of each band. Will be described. FIG. 1 is, for example, a conceptual diagram of the entire system when a personal computer (hereinafter, also referred to as a PC) is used as a voice encoding device according to the above-described encoding method. In the figure, 1 indicates a camera 17 and a microphone 19. A hard disk drive (HDD) 11, which is a so-called multimedia type PC capable of inputting data from an external device, has a large storage capacity for storing various data and programs, and is a fixed type recording medium, and an HD.
PD, which is a removable storage medium with a relatively small storage capacity for inputting and outputting programs and data to and from D11
Drive 12a, FD drive 12b, the above HD
The program stored in D is the central processing unit (CP
U) is configured to be read and executed by the instruction from the U) 14 as appropriate on the memory 13 configured by a random access memory (RAM) or the like. In addition, a video capture card 16 and a sound card 18 are respectively incorporated in order to capture video and audio from the camera 17 and the microphone 19, which are external devices. Each element of the PC 1 having the above configuration is connected by the internal data bus 15.

FIG. 2 is a block diagram of the encoder 20 of the speech coding apparatus for realizing the speech coding process executed by the PC 1 shown in FIG.
Is realized by the program read from the memory 13 to the memory 13. In FIG. 2, 21 is a CPU 1
CPU load monitoring information for monitoring the load state of No. 4, and 22 is coding means for controlling the operation of the low band coding processing means 23 and the high band coding processing means 24 based on the CPU load monitoring information 21. It is a control means. Further, reference numeral 25 is a bit stream forming processing means for converting the outputs of the two encoding processing means 23, 24 into stream signals. Further, reference numeral 26 is a coding mode designating signal which is inputted to the coding means control means 22 when designated by the user.

As the configuration of the low band encoding processing means 23 in FIG. 2, for example, the configuration shown in the conventional example of FIG. 11 is used. As the configuration of the high band encoding processing means 24, for example, as shown in FIG. 3, as in the example of FIG. 11, the uneven distribution of power of each band is used to encode each band (subband). However, the bit allocation to each sub-band signal using the psychoacoustic model is not performed, and as an alternative means, a band output adaptive bit allocation means 304 is provided to The scale factor is weighted based on human auditory characteristics, and the first purpose is low load processing rather than high sound quality, and the amount of calculation is small. Further, in order to eliminate excessive concentration of bit allocation to a specific band, weighting adjustment is performed according to each band for each bit allocation.

FIG. 4 is a block diagram showing a detailed structure of the encoder 20 shown in FIG. 2. Reference numeral 101 denotes an encoder, which is a subband analyzing means 102, a scale factor extracting means 103, and an FFT means which will be described later. 104, psychoacoustic analysis means 105, quantization / encoding means 106, auxiliary information encoding means 107, bit stream forming means 108, band output adaptive bit allocation means 109, psychoacoustic model bit allocation means 110, grouping means 111, The bit allocation processing control unit 112 and the allocatable bit calculation unit 113 are included.

The subband analysis means 102 divides the input digital audio signal into 32 frequency components. The scale factor extraction means 103 calculates a scale factor for each subband signal and aligns each subband dynamic range. The grouping unit 111 divides the 32 divided frequency components into the number of groups designated by the processing amount control information 121 which is external control information. In the first embodiment, as shown in FIG. 5, the number of groups is three, and each group is a subband signal continuous in the frequency axis direction, and a low band group A of 0 to 15 subbands and 16 to 29 subbands. High bandwidth group B, and bit allocation is not performed 30 to
The invalid group C of 31 subbands is divided into groups. The processing amount control information 121 includes the CPU load monitoring information 21 and the encoding mode setting signal 26. Further, in the first embodiment, as a bit allocating means for allocating bits to each of the subband groups, a low band that is sensitive to the human ear,
The psychoacoustic model bit allocating means 110 for highly efficient bit allocation using the relationship with the signal-to-mask ratio value based on the psychoacoustic model specified by MPEG.
Using a high-frequency band that is relatively insensitive to the human ear, the scale factor information from the scale factor extraction means 103 plus a minimum audible limit value for each band. The band output adaptive bit allocating means 109 for allocating bits with a lower load than that of the psychoacoustic model bit allocating method is used.

Also, the bit allocation processing control means 112.
FFT (Fast Fourier Transform) means for performing the required psychoacoustic analysis on the input digital audio signal with respect to the low band group A of 0 to 15 subbands for which the psychoacoustic model bit allocation is performed. The FFT means 104 is controlled so that the fast Fourier transform is performed at 104. Then, using this conversion result, the psychoacoustic analysis means 1
05 derives a relation model of signal-to-mask ratio (SMR) values based on a psychoacoustic model that utilizes the characteristics of human hearing.

The assignable bit computing means 113
Is based on the sampling frequency and the coding bit rate value, the number of assignable bits for the entire group to be determined, the ratio of each group to be bit-allocated to the entire group, and the characteristics for each band belonging to each group. The number of assignable bits to the independent bit assigning means is calculated for each group by using the weighted values. In the first embodiment, the psychoacoustic model bit allocating means 110 and the band output adaptive bit allocating means 113 consider the scale factor index value and the ratio of the low band / high band region, and allocate the entire band. Based on the number of possible bits, both means 11
The number of bits that can be assigned to 0 and 113 is determined. That is, in reality, each scale factor index value obtained by the scale factor extraction means 103
From scf_index [i], the addition value Vpsy, V of scf_index [i] in both areas is calculated as shown in the following formulas 1 and 2.
Calculate non.

[Equation 1]

[Equation 2] Here, psy_end = 16: number of subbands for which psychoacoustic model bit allocation is performed subband_end = 30: total number of bit allocation subbands.

Next, in the low band which is sensitive to the human ear,
In order to allocate more bits, Vpsy is weighted, and Vpsy = Vpsy * 0.75 psychoacoustic model bit allocatable number psy_num and band output adaptive bit allocatable number non_num are calculated based on the following equations. Vnon = Vnon * psy-ratio psy _num = all _alloc _num * Vnon / (Vpsy + Vnon) non _num = all _alloc _num-psy _num where all _alloc _num: total allocatable bit number psy _ratio: psy _end / ( subband_end-psy_end).

Within the range of the number of assignable bits for each group (psy_num, non_num), the psychoacoustic model bit allocating means 110 uses the SM from the psychoacoustic analyzing means 105.
Using the relational model of the R value, 0 which is the low band group A
Bit allocation is performed for ~ 15 subbands. On the other hand, the band output adaptive bit allocating means 109 allocates bits to the 16 to 29 subbands of the high band group B. In addition, since the 30 to 31 subbands of the invalid group C are invalid subbands, bit allocation is not performed.

According to the number of bit allocation to each subband signal determined by these bit allocation means,
Each subband signal is quantized / encoded by the quantization / encoding means 106, and the header information and the auxiliary information encoded by the auxiliary information encoding means 107 are combined together to form a bitstream in the bitstream forming means 108. Output.

If the information from the processing amount control information 121 is, for example, information for reducing the encoding processing amount, as shown in FIG. 6, the psychoacoustic model bit allocating means 110 having a large processing amount is provided. The bandwidth of the 0 to 15 subband low band group A that is the target of bit allocation of
0 to 7 sub-bands A low band group A ′ is reduced, and conversely, a band output adaptive bit allocating means 10 with a small processing amount is provided.
The bandwidth targeted for the bit allocation of 9 is 8 to 29.
The subbands are increased to become the lowband group B ′. Further, in the final form when the reduction of the encoding processing amount is considered, the processing amount is controlled by setting the bit allocation target of the band output adaptive bit allocation unit 109 to the 0 to 29 subband groups. In this case, the psychoacoustic model bit allocating means 110 does not operate substantially, so that the FFT means 104 and the psychoacoustic analyzing means 10 do not operate.
5 will not work either.

On the other hand, when the information from the processing amount control information 121 is, for example, the information for improving the sound quality of the encoded data, the psychoacoustic method capable of highly efficient (high sound quality) bit allocation. The bandwidth targeted for bit allocation by the model bit allocation unit 110 is increased. further,
In the final form when considering high sound quality, the bit allocation target of the psychoacoustic model bit allocation unit 110 is 0 to 0.
29 sub-band groups. In the present embodiment, as described above, the increase / decrease of subband groups and the switching of bit allocation means are performed in frame units, which are the minimum units that can be decoded into audio signals, so that the encoding processing amount can be controlled in real time. .

Next, the flow of the overall operation of the speech coding apparatus according to the first embodiment will be described with reference to FIG. First, in order to recognize the processing load of each encoder (23, 24) using the configuration as shown in FIG. 7A, in each mode of each encoder (change in bandwidth to be bit-allocated) , By encoding the dummy data for a predetermined time, the CPU load monitoring unit 700
Stores the CPU load value in each mode in the data table 701. Then, when the sample (data) is input, in step S70 of FIG. 7B, subband analysis is performed and divided into 32 frequency components, and subsequently in step S71, the scale factor of each subband signal is calculated. Calculated.

Then, in step S72, the CPU
It is determined whether or not there is load detection data. Here, since there is no CPU load detection data immediately after the start of operation, the process proceeds to step S74, and normal grouping capable of performing voice reproduction with the highest sound quality is performed. Then, the process proceeds to step S75, and the psychoacoustic model bit allocation process is performed. Then, the process proceeds to step S76 and quantization /
The encoding process is performed, the bit stream is further formed in step S79, and the series of processes is completed. At the end of the process, the CPU load monitoring unit 700 determines the time required to encode a predetermined number of input samples. And the current CPU load is detected.

Then, in the processing from the next time, it is determined in step S72 that the CPU load detection is "present", and in step S73, the detected CPU load cannot be encoded in real time. If it is determined that it is, the process proceeds to step S77, the optimum mode (grouping) is selected with reference to the data table 701, and the band bit allocation process in step S78 is used together with the psychoacoustic process in step S75. The model bit allocation process is performed at a predetermined rate, the process proceeds to step S76, and the quantization / encoding process is performed. In step S79, a bit stream using these encoded data is formed. Become.

The calculation of the allocatable bit calculating means 113 according to the first embodiment is performed by allocating an independent bit to each group in consideration of the scale factor index value and the ratio of the low band region and the high band region. Although the number of assignable bits for the means is calculated, the spectrum signal level belonging to each group from the FFT means 104 may be used instead of the scale factor index value, and the minimum audible limit value preset for each band is used. May be

Further, the encoder 10 according to the first embodiment
The CPU load monitoring means 700 for monitoring the processing amount of the CPU for the information for controlling the processing amount of 1 is provided inside the encoder 101, and the encoder 101 is operated so as not to exceed the processing capacity of the CPU. However, the control information may be external control information such as user input. By performing the user input, it is possible to perform the encoding process in which the sound quality and the image quality are prioritized according to the user's preference.

Further, the encoder 10 according to the first embodiment
As the bit allocating means of 1, the psychoacoustic model bit allocating means 110 for efficiently allocating bits to the low band which is sensitive to the human ear is used, and
The band output adaptive bit allocating means 109 for allocating bits with a low load is fixedly used, but the signal from the scale factor extracting means 103 causes the sub-band signal level belonging to each group to be set in each preset band. If it is less than or equal to the threshold value, that is, as shown in FIG. 8, that is, if the number of signals meaningful as encoded data in the low band is smaller than that in the high band, the bit allocation means is fixed according to the band. However, the psychoacoustic model bit allocating means 110 may be used in a high band.

Further, as shown in FIG. 9, the signal from the scale factor extracting means 103 is not used to judge the subband signal level belonging to each group and the threshold value, but the signal from the scale factor extracting means 103 is used. Also inputs a signal from the FFT means 104 having a high resolution (of frequency) to the bit allocation processing control means 112,
It may be configured such that the level of the subband signal belonging to each group is compared with the preset threshold value of each band.

(Second Embodiment) Next, a data recording medium according to a second embodiment of the present invention will be described. By recording the coding program for realizing the configuration of the speech coding apparatus or the coding method shown in the first embodiment on a data storage medium such as a floppy disk, each embodiment of the present invention can be realized. The processing shown in the form can be easily implemented in an independent computer system.

That is, FIG. 10 is a diagram for explaining a case where the encoding process of the first embodiment is executed by a computer system using a floppy disk storing the above encoding program. Figure 10 (a)
Fig. 10 (b) shows the external appearance of the floppy disk, the cross-sectional structure, and the main body of the floppy disk.
Shows an example of the physical format of the floppy disk body.

In the floppy disk FD, the floppy disk body D is a floppy disk case FC.
A plurality of tracks Tr are formed concentrically from the outer circumference toward the inner circumference on the surface of the floppy disk body D, and each track Tr has 16 sectors Se in the angular direction. Is divided into Therefore,
In the floppy disk FD storing the program, the floppy disk body D has data (program) recorded in an area (sector) Se allocated thereon.

Further, FIG. 10C shows a configuration for recording the above-mentioned program in the floppy disk FD and performing voice encoding processing using the program stored in the floppy disk FD. When the above program is recorded on the floppy disk FD, the data as the above program from the computer system Cs,
Writing to the floppy disk FD via the floppy disk drive FDD. In addition, floppy disk FD
When the above audio coding apparatus is constructed in the computer system Cs by the program recorded in the computer system Cs, the program is read from the floppy disk FD by the floppy disk drive FDD, and the computer system C is read.
load into s.

In the above description, a floppy disk is used as the data recording medium, but an audio disk can be used to perform voice encoding processing by software as in the case of the floppy disk. Further, the recording medium is not limited to the above-mentioned optical disk or floppy disk, but may be an IC card, a ROM cassette or the like, as long as it can record a program. Similarly, a voice encoding process by software can be performed.

(Third Embodiment) Next, a speech coding system and a speech coding apparatus according to a third embodiment of the present invention will be described with reference to the drawings. As the configuration of the high band encoding processing means 24 shown in FIG. 2, for example, the configuration shown in FIG. 3 is used. Further, as the configuration of the low band encoding processing means 23, for example, as shown in FIG. 12, as in the case shown in FIG. 11, the uneven distribution of the power of each band is used to encode each band (subband). Although a coding method for performing encoding is used, bit allocation is not performed for each subband signal using only a predetermined psychoacoustic model analysis unit, and a simple psychoacoustic model unit 4062 having a small processing amount is newly provided.
Bit allocation can be performed based on the masking threshold of the psychoacoustic model unit 4061 output in the previous frame and the bit allocation information generated based on the band division signal of the frame.

That is, FIG. 12 is a block diagram showing a detailed configuration of the low band encoding processing means 23 shown in FIG. 2, and 401 is an encoder, which is a sub-pand analysis means 402 and scale factor extraction means which will be described later. 403, bit cutting processing control means 404, FFT processing means 405,
It is composed of a psychoacoustic analysis stage 406, a psychoacoustic model bit allocation means 407, a quantization / encoding means 408, an auxiliary information encoding means 409, and a bit stream forming means 410.

The operation will be described below. The subband analysis means 402 divides the input digital audio signal into 32 frequency components. The scale factor extraction means 403 calculates a scale factor for each subband signal and aligns each subband dynamic range. The FFT processing means 405 performs a fast Fourier transform on the input digital audio signal. The psychoacoustic analysis unit 406 includes, for example, a normal auditory psychological model unit 4061 designated by MPEG and the above-described simple auditory psychological model unit 4062 which has a smaller processing amount than the normal auditory psychological model unit 4061, and each model is a signal. Calculate the mask-to-mask ratio.

The normal auditory psychology model unit 4061 is used.
While the signal-to-mask ratio of each sub-band signal is calculated based on Equation 3 below, the simplified psychoacoustic model 4062 described above calculates each sub-hand of each sub-hand in the corresponding frame as shown in Equation 4 below. The minimum masking level is not calculated, but the minimum masking level in the most recent previous frame calculated by the normal psychoacoustic model unit 4061 is used, and the sound pressure is the scale extracted by the scale factor extraction unit 403 of the corresponding frame. -Calculate the signal-to-mask ratio using the factor value.

[Equation 3] Here, Lsb (n): pressure of each subband LT _min (n): minimum masking level of each subband.

[Equation 4] Here, Lsb (n) = 20.log (scf _max (n) .32768-10) db scf _max (n): Scale factor value for each subband of the frame LT _min (n): Recent normal Auditory psychology model unit 4061
Minimum masking level for each sub-band calculated in

In the third embodiment, the bit allocation processing control means 404 sets N to 3 based on the information of the processing amount control information 121, as shown in FIG. The psychoacoustic model unit 4062 and the normal psychoacoustic model unit 4061 capable of outputting optimum bit allocation information capable of realizing high quality,
It controls how many frames are performed once and whether or not the fast Fourier transform is performed in the FFT processing means 405. For example, in the state of FIG. 13, when the bit allocation processing control means 404 is informed as the information of the processing amount control information 121 that the CPU occupancy rate for the encoding process is reduced, the above-mentioned simple operation with a small processing amount is performed. In order to increase the usage of the psychoacoustic model 4062, the value of N is increased. On the contrary, when the bit allocation processing control means 404 is notified of the information that the CPU occupancy rate for the encoding processing may be used more, the normal auditory psychology model unit 4061 of the normal auditory psychology model unit 4061 capable of realizing high sound quality. Lower the value of N for greater use. Thereby, it is possible to control the processing amount.

Auditory psychological model bit allocation means 407
Assigns bits to each sub-pand signal divided by the sub-pand analysis means 402 based on the signal-to-mask ratio relationship which is information from the bit allocation processing control means 404. The quantizing / encoding means 408 quantizes and encodes each subband signal,
A bit stream is formed by the bit stream forming means 410 together with the auxiliary data from the auxiliary information encoding means 409 and is output.

As described above, according to the third embodiment, bit allocation is performed once every N frames, so that the CPU load in the time axis direction can be reduced.

Although the encoder 401 is assumed to be the low band encoding processing means 23 shown in FIG. 2 here, it is not applied only to the low band signal but to the full band signal. You may make it apply.

(Embodiment 4) Next, a speech coding system and speech coding apparatus according to a fourth embodiment of the present invention will be described with reference to the drawings. Similar to the case shown in FIG. 11, FIG. 14 uses an encoding method for performing encoding for each band (subband) by utilizing uneven distribution of power of each band. It differs in that it has a function to add external data other than audio data. Image data, text data, and the like are assumed as the external data.

That is, the encoder 501 shown in FIG.
Are sub-band analysis means 502, scale factor extraction means 503, FFT processing means 504, psychoacoustic analysis means 505, bit allocation means 506, quantization /
The encoding unit 507, the auxiliary information encoding unit 508, the bit stream forming unit 509, the bit allocation processing control unit 510, and the additional data encoding unit 511 are included.

The operation will be described below. The subband analysis means 502 divides the input digital audio signal into 32 frequency components. The scale factor extraction means 503 calculates a scale factor for each subband signal and aligns each subband dynamic range. The FFT processing means 504 performs a fast Fourier transform on the input digital audio signal. The psychoacoustic analysis unit 505 calculates the signal-to-mask ratio using, for example, a psychoacoustic model specified by MPEG.

The bit allocation processing control means 510 monitors the additional data buffer 512 for temporarily storing the data to be added in the output bit stream, and judges whether or not there is additional data, or Based on the allocation range control information 513 generated from the judgment of whether or not to overflow, the bit allocation means 506
Specify the range for bit allocation.

For example, when there is no data in the additional data buffer 512, bits are assigned to the subbands 0 to 29 as shown in FIG. In this case, assuming that the total number of assignable bits is 100, 80 bits are assigned to subbands 0 to 15 and 20 bits are assigned to subbands 16 to 29.

When the external data is written in the additional data buffer 512 and the additional data exists, that is, the allocation range control information 513.
As a result, the instruction to insert the additional data is notified to the bit allocation processing control means 510, and in the fourth embodiment, for example, 80 bits are allocated to the subbands 0 to 15, and the subband 16 to be originally allocated is assigned. ~
Bit allocation is not performed for 29, and the remaining 20
Bits are assigned as the number of additional bits of data.
Further, in order to reduce the amount of processing, the FFT processing and the psychoacoustic analysis in the corresponding range may not be performed on the subbands 16 and after in which the bit allocation is not performed.

Quantization / encoding means 507 quantizes and encodes the sub-band to which the above bit allocation has been performed, and the auxiliary data from auxiliary information encoding means 508 and, for example, MPEG. With the additional data encoded as the ancillary data of 1, the bit stream is formed by the bit stream forming means 509 and output.

As described above, according to the fourth embodiment, when transmission is performed at a constant bit rate, the bit allocation range at the time of encoding is controlled according to the amount of additional data other than audio data and the encoding is performed. Since the amount of audio data to be converted is variable and the additional data is inserted into the encoded data stream, various data can be superimposed on the surplus band and the band can be effectively used. The control of the bit allocation range executed by the bit allocation processing control stage 510 is performed on a frame-by-frame basis, and the bit allocation range is also variable according to the data amount of the additional data buffer 512.

By these processes, even when the additional data is inserted, the data insertion amount can be controlled in real time without impairing the sound quality within the pit allocation range.

(Embodiment 5) Next, a speech coding system and speech coding apparatus according to a fifth embodiment of the present invention will be described with reference to the drawings. FIG. 16 is a block diagram showing the configuration of the encoder of the speech coding apparatus using the speech coding system according to the fifth embodiment. In the figure, the same reference numerals as those in FIG. 2 indicate the same or corresponding portions, and 160 ˜162 are encoding processing means A to C and 1 capable of operating independently.
Reference numeral 63 is a processing load value storage buffer for storing the processing load value information of each of the encoding processing means A to C, and 164 is a sample data buffer for supplying sample data to each of the encoding processing means A to C. Is.

Next, the operation will be described. At initialization before encoding processing, first, the sample data buffer 1
The predetermined sample data stored in 64 is supplied to the respective encoding processing means A to C, and the processing load values of the encoding processing means A to C or the psychoacoustic model generated thereby are stored in the processing load value storage buffer 163. To store.

Then, by outputting the processing load values in ascending or descending order, the C used in the apparatus can be obtained.
Quickly select the encoding processing means that matches the performance of the PU,
Encoding processing is performed by the encoding processing means. The content of the encoding process is the same as that shown in the first embodiment, and therefore will not be repeated here.

As described above, according to this embodiment, at the time of initialization before encoding processing, each encoding processing means is operated using sample data, the load value at that time is acquired, and the CPU to be used Since the encoding processing means suitable for the processing capacity of is selected and used, the load on the CPU is reduced,
It becomes possible to perform an optimum encoding process.

In each of the above-described embodiments, a configuration in which a PC is used as an audio encoding device has been described as an example, but it is used by being incorporated in a device such as a VTR camera or a DVD encoder. It can also be applied in cases.

In each of the above embodiments, only the audio is handled. However, when processing the video together with the audio, as shown in FIG. 17, in the configuration of FIG. A signal is input, and video coding processing means 170 and audio coding processing means 171 are provided in place of the low band coding processing means and the high band coding processing means, and further,
This can be dealt with by providing the system stream processing means 172 instead of the bit stream forming processing means. Using the configuration as described above, based on the control information from the outside, the operation amount of voice encoding can be changed or a plurality of voice codes having different operation amounts can be changed by the method described in each of the above embodiments. It is possible to control the total amount of calculation as the CPU by switching the conversion method. Alternatively, the processing amount of the video signal to be encoded may be changed according to the amount of the audio signal to be encoded.

Further, in addition to MPEG1 which performs subband coding, MPE which performs time / frequency conversion
G2, AAC, Dolby AC-3, ATRAC (M
Also in the case of performing the coding method such as D), as shown in FIG. 18, the respective units related to the encoding process are the first quantum information calculating unit 181 and the second quantum information calculating unit 182 which have different calculation amounts. It is possible to deal with the same problem by substituting the same with the above and selecting and using them by the quantizing means controlling means 180 to handle the quantized information instead of the encoded information.

[0089]

As described above, according to the voice encoding system according to the first aspect of the present invention, the voice encoding system that divides the digital audio signal into a plurality of frequency bands and encodes each frequency band. The bit allocation information for each of the divided bands is generated, each of the plurality of bit allocation means having a different processing amount is provided, and based on control information from the outside, among the plurality of bit allocation means,
As processing using a predetermined bit allocation means is performed, by executing a bit allocation switching the bit allocation means to be used, have rows coding control from the outside
As information, central processing that can be occupied when encoding
Based on the above load value, using the load value that represents the processing amount of the device
Then, on the above central processing unit, each bit allocation means
Data that has been stored in advance for each processing amount when encoding using
The central processing unit that can be occupied by
Since the bit allocation means is selected so as not to exceed the maximum processing amount, the bit allocation means having the optimum processing amount can always be selected and used, and the CPU allocation that can be occupied in the operating state is limited. It is possible to perform encoding that does not exceed the processing amount, and it is possible to perform encoding in which the processing does not catch up with the input signal at the time of real-time encoding, that is, the reproduced sound can be encoded without sound interruption. Furthermore, the central processing unit always exceeds the operating capacity.
The entire system is no longer required
There is also an effect that can be smoothly controlled
It

According to a second aspect of the present invention, there is provided a speech encoding method according to the first aspect.
In the formula, to perform encoding processing as the above load value
Monitor the processing amount of the central processing unit that can be occupied by
Since the processing amount control information from the monitoring means is used,
Range of highest possible central processor performance
Select the bit allocation method with the optimum
And can process the input signal during real-time encoding
There is nothing that can not be done in time, that is, the playback sound
There is an effect that it is possible to perform coding without sound interruption .

[0091] Further, according to the speech encoding method according to claim 3 of the present invention, the speech coding method according to the first aspect, the bit allocation by the bit allocation means
As processing, high efficiency that can realize high quality of encoded data
A high-efficiency bit allocation method that allocates bits to rates
The processing used and the processing using the high-efficiency bit allocation method
Bit allocation with low load and less processing compared to
Perform processing using the low-load bit allocation method
Since it is set, the encoder gives priority to improving the quality of encoded data.
Or prioritize low load of encoding processing over sound quality
Code that can be executed by appropriately switching Ruka processing
There is an effect that can be realized .

According to the speech coding method of claim 4 of the present invention, in the speech coding method of claim 1, the bit allocating means used at the time of the coding is used.
Switching with the smallest unit that can be decoded into an audio signal.
Since it was done in a certain frame unit, real time
Share the CPU on the operating CPU when encoding
CPU occupancy of applications such as
Even when there is a problem, the encoding process is performed in frame unit time.
It becomes possible to follow the processing amount of the CPU that the logic can occupy.
Moreover, there is an effect that the sound quality and the processing amount can be controlled in real time .

According to the speech coding system of claim 5 of the present invention, in the speech coding system of claim 1 , subbands of each band divided into a plurality of frequency bands are used.
Band signal to a predetermined number of subbands
Group into groups of signals
Independent bit allocation processing for each group
And generate bit allocation information for each band.
Since this is done, bit allocation according to the characteristics of each band
There is an effect that encoding can be performed by selecting a process .

According to a sixth aspect of the present invention, in the voice encoding method according to the fifth aspect , the grouping is performed by the number of groups or the group.
Number of consecutive subband signals along the frequency axis in the loop
Is the number specified by the control information from the outside, or
Specified based on the throughput control information from the monitoring means
Since it is designed to be performed variably so that
There is an effect that grouping can be dynamically performed according to the usage status of PUs .

According to a seventh aspect of the present invention, in the speech encoding method according to the sixth aspect , the change in the number of the sub-band signals is changed to the audio.
It is performed in frame units, which is the smallest unit that can be decoded into a signal.
Since it has been done so, change the bit allocation method carefully
There is an effect that it can be performed and a highly accurate encoder can be realized .

[0096] Further, according to the speech encoding method according to claim 8 of the present invention, in the audio coding method of 請 Motomeko 7, wherein when in said grouping is not performed bit allocation
I have set up at least one group, so
Frame unit, which is the smallest unit that can be decoded into a Dio signal
In the direction of the number of groups or the frequency axis within the group.
The number of consecutive sub-band signals is used as control information from the outside.
More specified number, or throughput control information from monitoring means
Bit allocation by changing to the number specified by
Process the signals in the band that belongs to the group
And there is no need for bit allocation.
Bits that should be allocated to the band that belongs to the group
To the bandwidth of other groups where the bit allocation is done.
C, which can be allocated as a result of which the encoding process occupies
It is possible to control the amount of PU processing and allocate bits.
There is an effect that it is possible to improve the sound quality of the bands of other groups in which

According to a ninth aspect of the present invention, in the voice encoding method according to the fifth aspect , the sub-band signals are grouped into a low-level signal.
Subbands grouped into groups that belong to the band
Signals are processed using the above-mentioned high-efficiency bit allocation method.
By reason, bit allocation is performed on the subband signals within the group.
On the other hand, on the other hand,
The above-mentioned low-load bit allocation
Subbands within a group can be
Since the bits are assigned to the signals, the human ear
For low band, which is sensitive to
The sound quality can be improved, while the sensitivity of the human ear is reduced.
Low load bit allocation with priority on processing amount for bad high bandwidth
Can be performed, and the processing amount as a whole
There is an effect that reduced coding can be performed .

According to a tenth aspect of the speech encoding system of the present invention, in the speech encoding method according to the fifth aspect , independent bit allocation means is provided for each group.
Assignable to determine the number of assignable bits for
A bit operation means is provided so that each group
Based on the characteristics of each band belonging to each group
The weight of the entire group,
The number of assignable bits to be set is independent for each group.
Since it is distributed to the bit allocation means,
Consideration of auditory characteristics for input signal or characteristics of each band
Each group that is most suitable for achieving high sound quality of encoded data
There is an effect that it is possible to perform coding that enables bit allocation to the bit allocation means of the loop.

According to the eleventh aspect of the speech coding system of the present invention, the speech coding system according to the tenth aspect is based on the characteristics of each band belonging to each group.
Based on the specified minimum audible limit for each band.
By weighting based on
There is an effect that a meaningful and effective bit allocation process can be performed.

According to a twelfth aspect of the speech coding system of the present invention, in the speech coding system according to the tenth aspect , weighting based on the characteristics of each band belonging to each group is applied to the input digital signal. Sub signal to audio signal
Frequency that belongs to each group, which is obtained by performing a band analysis.
By weighting based on subband signal levels of several bands, there is an effect that it is possible to perform effective bit allocation process.

[0102] Also, according to claim 13 in such a speech coding scheme of the present invention, in the audio coding method of the claim 10, wherein, the weighting based on characteristics of respective bands belonging to each group, the input digital audio There is an effect that effective bit allocation processing can be performed by performing weighting based on the spectrum signal level belonging to each group, which is obtained by performing linear conversion on the signal.

According to the speech coding system of claim 14 of the present invention, in the speech coding system of claim 5 , the signal levels belonging to each group have a predetermined level.
For high level signals above the threshold value, the high efficiency
Bit allocation by processing using the bit allocation method
The signal level belonging to each group is
For low level signals below
Bit allocation by processing using the
Since it was done, it is treated as encoded data compared to other bands.
Processing load on less important signals
Code that can improve the sound quality of encoded data
There is an effect that it is possible to perform the reduction.

[0103] Further, according to the speech encoding method according to claim 15 of the present invention, in the audio coding method of the claim 14, the signal level belonging to each group, enter
Obtained by subjecting the digital audio signal to subband analysis
Effective by setting the sub-band signal level
There is an effect that various bit allocation processing can be performed.

According to a sixteenth aspect of the present invention, in the voice encoding method of the fourteenth aspect , the signal levels belonging to each group are linearly converted into an input digital audio signal. Obtained by applying
Since the spectral signal level, there is an effect that it is possible to perform effective bit allocation process.

[0105] Further, according to the speech encoding method according to claim 17 of the present invention, in the audio coding method of the claim 14, the signal level belonging to each group, where
Since the minimum audible limit value for each fixed band is set,
There is an effect that it is possible to perform an effective bit allocation process that makes sense when listening .

Further, according to the speech coding system of claim 18 of the present invention , any one of claims 3, 9 and 14 above .
In the speech coding method described in Section 1,
The processing using the assignment method is based on a specified psychoacoustic model.
The signal-to-mask ratio value relationship
Processing using the load bit allocation method is
The signal level divided into bands has a predetermined minimum value for each band.
Since it was supposed to be done with consideration of the listening limit value,
As long as it is heard by the human ear,
There is an effect that the processing amount of the system can be reduced .

According to a nineteenth aspect of the present invention, in the speech encoding system according to the eighteenth aspect , the auditory psychological model is designated by MPEG.
It is possible to use MPEG (Motion
For audio coding processing using Picture Experts Group)
However, there is an effect that the same effect as described above can be obtained .

The speech coding system according to claim 20 of the present invention is the speech coding system according to claim 4 or claim 7, wherein the audio signal can be decoded into the audio signal.
A frame that is a small unit is MPEG (Motion Picture E).
xperts Group)
As a result, the same effect as described above can be obtained even in the audio encoding process using MPEG .

[0109] According to the speech coding system of claim 21 of the present invention, in the speech coding system of claim 1 , the bit allocation means is divided into each of the divided parts.
Information output from the specified psychoacoustic model for the band
To generate bit allocation information based on
Once in N (N = 1, 2, 3 ...) frames, the specified hearing
Bit allocation based on the information output from the psychological model
To generate the information and not the above bit allocation information.
Output from the above psychoacoustic model
Based on the divided information and the signal information of each divided band above
To generate bit allocation information and encode it.
Therefore, there is an effect that the CPU load in the time axis direction can be reduced .

According to the speech coding system of claim 22 of the present invention, in the speech coding system of claim 1, the psychoacoustic model in which the processing amount can be controlled step by step.
Based on the control information from the outside,
By controlling the processing amount of the model, the auditory psychology of a predetermined processing amount
For each band so that it can be processed using the model
Since the bit allocation information is generated,
There is an effect that CPU load control can be performed in consideration of various effects .

Further, according to the speech coding method of claim 23 of the present invention, in the speech coding method of claim 1, a plurality of psychoacoustic models having different processing amounts are combined.
The number of the above-mentioned multiple listenings based on control information from the outside.
From the psychoacoustic model, using a predetermined auditory psychological model
Cut the psychoacoustic model used so that it can be processed.
Instead, since the bit allocation information for each band is generated, it is easier to add the auditory effect to C.
There is an effect that PU load control can be performed.

According to a twenty-fourth aspect of the speech coding system of the present invention, in the speech coding system according to the first aspect, the central processing unit for performing the coding process is
Depending on the performance, multiple bytes may be used during initialization before the encoding processing operation.
Allocation means, or each location of multiple psychoacoustic models.
Since the physical load value information is output to the outside, actually
Before encoding, information about the performance of the CPU used
Information can be obtained and the processing load of the CPU can be efficiently
There is an effect that it can be reduced .

According to a twenty-fifth aspect of the present invention, there is provided a voice encoding method according to the twenty-fourth aspect.
In an expression, multiple bits output as information to the outside
Allocation means or each processing of multiple psychoacoustic models
Load value information is now output in descending or ascending order
Since the effect is Ru Oh that it is possible to select the encoding means quickly.

According to a twenty-sixth aspect of the speech coding apparatus of the present invention, the speech coding apparatus according to any one of the first to twenty-fifth aspects can be used.
Line speech encoding using a speech coding method of Zurekani wherein
Therefore, the VT incorporating the voice encoding system
The same effect as described above can be obtained in a device such as an R camera .

The data according to claim 27 of the present invention
The storage medium is any one of claims 1 to 25.
The steps of the speech encoding method described is recorded in the
Then, by using the storage medium, the above audio coding method is set in the device.
By incorporating, there is an effect that the same effect as described above can be obtained .

[0116]

[0117]

[0118]

[0119]

[0120]

[0121]

[0122]

[0123]

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of an entire system using a personal computer in order to realize a voice encoding device using a voice encoding method according to a first embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of an encoder that constitutes the speech encoding apparatus according to the first embodiment.

FIG. 3 is a block diagram showing a detailed configuration of high-band encoding processing means that constitutes the encoder.

FIG. 4 is a block diagram showing a more detailed configuration of an encoder that constitutes the speech encoding apparatus according to the first embodiment.

FIG. 5 is a schematic diagram showing an example of a bit allocation process for each group, which is used in the speech coding method according to the first embodiment.

FIG. 6 is a schematic diagram showing another example of the bit allocation process for each group used in the speech coding system according to the first embodiment.

FIG. 7 is a diagram showing a flow for explaining an encoding operation of an encoder forming the speech encoding apparatus according to the first embodiment.

FIG. 8 is a schematic diagram showing an example of performing a bit allocation process for each group using a threshold, which is used in the speech coding method according to the first embodiment.

[Fig. 9] Fig. 9 is a block diagram illustrating a detailed configuration of a modified example of the encoder forming the speech encoding apparatus according to the first embodiment of the present invention.

FIG. 10 is a block diagram showing a data storage medium according to a second embodiment of the present invention and a configuration when a speech encoding apparatus is configured using the storage medium.

[Fig. 11] Fig. 11 is a block diagram illustrating a configuration of an encoder that configures a conventional speech encoding device.

[Fig. 12] Fig. 12 is a diagram illustrating a detailed configuration of low-band encoding processing means that constitutes a speech encoding apparatus according to Embodiment 3 of the present invention.

FIG. 13 is a diagram for explaining a state of a psychoacoustic model in each frame at the time of low-band encoding by the speech encoding apparatus according to the third embodiment.

[Fig. 14] Fig. 14 is a diagram illustrating a detailed configuration of low-band encoding processing means that constitutes a speech encoding apparatus according to Embodiment 4 of the present invention.

FIG. 15 is a diagram showing an example of a bit allocation process using the speech coding apparatus according to the fourth embodiment.

[Fig. 16] Fig. 16 is a block diagram showing the structure of an encoder forming the speech coding apparatus according to Embodiment 5 of the present invention.

[Fig. 17] Fig. 17 is a block diagram illustrating a configuration of an encoder when a video signal is handled together with an audio signal.

[Fig. 18] Fig. 18 is a block diagram illustrating a configuration in the case where the present invention is applied to an encoding process in an encoding device that performs time / frequency conversion system coding.

[Explanation of symbols]

1 Personal computer (speech encoder) 11 HDD 12a PDD 12b FDD 13 memory 14 CPU (Central processing unit) 15 data bus 16 video capture cards 17 cameras 18 sound card 19 microphone 20 encoder 21 CPU load monitoring information 22 Encoding means control means 23 Low Band Coding Processing Means 24 High Band Coding Processing Means 25 bit stream forming processing means 26 Encoding mode designation signal 101 encoder 102 Subband analysis means 103 scale factor extraction means 104 FFT means 105 Auditory psychological analysis means 106 quantization / encoding means 107 auxiliary information coding means 108 bit stream forming means 109 band output adaptive bit allocation means 110 Auditory Psychological Model Bit Allocation Means 111 Grouping means 112-bit allocation processing control means 121 throughput control information 160-162 Encoding processing means A-C 163 Processing load value storage buffer 164 sample data buffer 170 Video coding processing means 171 Speech coding processing means 172 system stream forming processing means 180 Quantization means control means 181 First Quantization Information Calculation Means 182 Second quantized information calculation means FC floppy disk case FD floppy disk D floppy disk body Se sector Tr truck Cs computer system FDD floppy disk drive

─────────────────────────────────────────────────── ─── Continuation of front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) H03M 7/30 G10L 19/00

Claims (27)

(57) [Claims] 1. A voice coding method in which a digital audio signal is divided into a plurality of frequency bands, and coding is performed for each band, wherein bit allocation information for each of the divided bands is generated and a processing amount is respectively set. A plurality of different bit allocating means, and based on external control information, the bit allocating means to be used is selected from among the plurality of bit allocating means to be used. run the bit allocation switching, performs encoding, as control information from the external, occupied when performing coding
The load value used to represent the amount of processing that can be a central processing unit, based on the load value, the bi in the central processing unit on
Processing amount when encoding is performed using the bit allocation means
Can be occupied by referring to the data table that stores in advance
The above-mentioned bit should not exceed the processing capacity of the central processing unit.
A voice coding method characterized in that the allocation means is selected . 2. The speech encoding system according to claim 1, wherein the load value can be occupied for performing encoding processing.
From the monitoring means for monitoring the throughput of the central processing unit
A speech coding method characterized by using the processing amount control information of . 3. The voice coding system according to claim 1, wherein the bit allocation processing is performed by the bit allocation means.
The bit rate with high efficiency, which can realize high-quality encoded data.
Processing using a highly efficient bit allocation method that allocates
If, compared to the treatment with the high efficiency bit allocation method
A low-load task that allocates bits with a low load and low processing load.
The audio coding method is characterized in that processing is performed using a bit allocation method . 4. The voice coding method according to claim 1, wherein the bit allocation means used in the coding is switched.
Frame, which is the smallest unit that can be decoded into an audio signal.
A voice coding method characterized by being performed in units of frames . 5. The speech coding method according to claim 1 , wherein the sub-band signal of each band is divided into a plurality of frequency bands.
From a predetermined number of subband signals
Groups are divided into groups and each group is
Independent bit allocation processing is performed on each
A voice coding method characterized by generating bit allocation information for a band . 6. The voice coding method according to claim 5, wherein the grouping is performed continuously in the number of groups or in the frequency axis direction within the group.
The number of sub-band signals
Specified, or the processing amount control information from the above monitoring means.
Variable so that the specified number is based on the report
Speech coding, characterized in that it is. 7. The voice encoding system according to claim 6, wherein the change in the number of subband signals is restored to an audio signal.
A voice coding method that is performed in frame units, which is the smallest unit that can be signaled . 8. The voice coding method according to claim 7 , wherein a bit is not assigned at the time of grouping.
A speech coding method characterized in that at least one group is provided . 9. The voice coding system according to claim 5, wherein the subband signals are grouped into a low band by grouping.
Sub-band signals grouped into
However, the processing using the above high-efficiency bit allocation method
Bit assignment to the subband signals in the group.
On the other hand, it is divided into groups that belong to the high band.
Low load bit allocation method for sub-band signals
To the subband signals in the group by
Speech coding method characterized by bit allocation
Expression . 10. The voice coding system according to claim 5 , wherein a bit allocation unit independent for each group is allocated.
Assignable bit operation that determines the number of assignable bits
Providing means and ratio of each group to the whole group
, Weights based on the characteristics of each band belonging to each group
Using the one with the addition,
Bit of a hit can be the number of bits, was independent for each of the groups Ri
Speech coding method characterized by assigning to allocation means . 11. The speech coding method according to claim 10 , wherein weighting is performed based on a characteristic of each band belonging to each group.
Is weighted based on a predetermined minimum audible limit for each band.
Speech coding characterized by Ketosu Rukoto. 12. The audio encoding method according to claim 10 , wherein weighting based on characteristics of each band belonging to each group is performed on the input digital audio signal by subband analysis.
Subs of each frequency band belonging to each group
A speech coding method characterized by weighting based on band signal levels . 13. The audio encoding method according to claim 10 , wherein weighting based on characteristics of each band belonging to each group is obtained by linearly converting an input digital audio signal.
It is, speech coding, characterized in that a weighting based on spectrum signal levels belonging to each group. 14. The speech coding method according to claim 5 , wherein the signal level belonging to each group is equal to or higher than a predetermined threshold value.
The high-efficiency bit allocation for
The bit level is assigned by the processing using the method described above, and the signal level belonging to each group is below a predetermined threshold.
The low load bit allocation above is applied to the low level signal of
A voice coding method characterized in that bit allocation is performed by processing using this method. 15. The speech encoding system of claim 14, the signal level belonging to each group, the input Dejitaruo
Sub-band obtained by sub-band analysis of audio signal
A voice coding method characterized by setting the audio signal level . 16. The speech encoding system of claim 14, wherein the signal levels that belong to each group, a spectrum signal <br/> No. levels obtained by applying a linear transformation to the input digital audio signal And audio coding method. 17. The voice encoding system according to claim 14 , wherein the signal levels belonging to each of the groups are set for each predetermined band.
A speech coding method characterized by setting the minimum audible limit value of . 18. The speech coding method according to claim 3, wherein the processing using the high-efficiency bit allocation method is a predetermined method.
Use signal-to-mask ratio value relationship based on psychoacoustic model
The process using the above low-load bit allocation method
The signal level divided into frequency bands has a predetermined value for each band.
A speech coding method that is performed with the minimum audible limit taken into consideration . 19. The audio encoding system according to claim 18, wherein the psychoacoustic model is MPEG (Motion Picture Exp).
erts group) is a psychoacoustic model specified by the audio coding method. 20. The voice encoding system according to claim 4 or 7, wherein the audio signal is a minimum unit that can be decoded into a frame.
The audio coding method is characterized in that the frame is a frame specified by MPEG (Motion Picture Experts Group). 21. The voice encoding system according to claim 1, wherein the bit allocation means is provided for each divided band.
Based on the information output from the specified psychoacoustic model
Bit allocation information is generated by using the above-mentioned predetermined auditory information once in every N (N = 1, 2, 3 ...)
Bit allocation based on the information output from the psychological model
Information for each frame that did not generate the above bit allocation information.
The information output from the above psychoacoustic model and the above
Bit allocation based on the signal information of each divided band
A speech coding method characterized by generating information and coding it. 22. The audio encoding method according to claim 1 , further comprising a psychoacoustic model whose processing amount can be controlled in stages, and the psychoacoustic model of the psychoacoustic model is generated based on control information from the outside.
By controlling the amount of processing, a psychoacoustic model with a predetermined amount of processing
Bit allocation for each band so that
A speech coding method characterized by generating allocation information . 23. The speech coding method according to claim 1, wherein a plurality of psychoacoustic models each having a different processing amount are provided , and a predetermined psychoacoustic model is selected from the plurality of psychoacoustic models based on control information from the outside . as treatment with hearing psychological model is formed <br/>, switch the psychoacoustic model used,
A speech coding method characterized by generating bit allocation information for each band. 24. The speech coding system according to claim 1, wherein the coding processing is executed according to the performance of a central processing unit.
Multiple bit allocations during initialization before encoding operation.
Method or each processing load value information of multiple psychoacoustic models
A voice coding method characterized by outputting information to the outside . 25. The voice encoding system according to claim 24,
And multiple bit assigners that are output as external information.
Step or multiple processing load value information of the psychoacoustic model
Are output in descending order or in ascending order . 26. claims 1 to any one of claims 25
Speech coding apparatus and performing speech encoding using a speech coding method according to. 27. Any one of claims 1 to 25.
Data recording medium which step of the speech coding method is characterized and this <br/> recorded according to.