JP2001267928A - Audio data compressor and storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an audio data compressor that can enhance the bodily sensation sound quality of audio data encoded at a low bit rate and to provide a storage medium. SOLUTION: In the audio data compressor that divides audio data of each frame into bands, assigns a proper bit number to each of the divided bands and quantizes them, an encoding sub-band decision section 23 limits the bands to an encoding object band depending on a bit rate and a bit assignment section 300 assigns a usable bit number depending on the bit rate to the limited band.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an audio data compression device and a storage medium, and more particularly, to an audio data compression device and a storage medium for improving the perceived sound quality.

[0002]

2. Description of the Related Art In recent years, various data compression methods have been provided to increase the efficiency of data communication channels and data storage media. For example, as an audio data compression method,
MPEG (Moving Picture Experts Group) / Audi
o (ISO / IEC 11172-3).

[0003] The above-mentioned MPEG / Audio compresses and encodes a two-channel stereo audio signal while maintaining the sound quality equivalent to that of a CD, and has three layers (layer 1 and layer 2) according to the complexity of the hardware and the sound quality. , Layer 3) are defined. As the layers increase, the hardware becomes more complex but the sound quality improves. Layer 1 and layer 2 use a subband coding scheme, and layer 3
A method is used in which the layer 1 and layer 2 subband coding methods and the transform coding method are combined.

[0004] Compression methods include 32-band division coding, M
Using DCT (Modified Discrete Cosine Transform), highly efficient compression is realized using psychoacoustic characteristics.

FIG. 11 is a block diagram showing a configuration of a conventional MPEG / Audio encoder. In FIG. 11, the MPEG / Audio encoder includes a filter bank 100, an psychoacoustic model unit 200, a bit allocation unit 300, and a bit stream forming unit 400.

The filter bank 100 uses a polyphase filter bank, an MDCT filter bank, or the like, and is a band division method capable of removing noise by processing the entire audible frequency band at a time by a mathematical method and performing efficient calculations. Yes, it splits the input signal (original sound) into groups on several frequencies. The divided bands are called subbands.

The psychoacoustic model unit 200 generates a data set for controlling quantization and encoding using a mathematical model of a masking characteristic in auditory sense, and generates a signal-to-mask ratio (SMR) for a divided band. Output as The SMR is a value calculated by dividing an input signal (original sound) by a masking threshold described later, and represents an energy value of the signal.

FIG. 12 is a diagram showing the concept of psychoacoustic encoding processing performed in the psychoacoustic model unit 200. In FIG. 12, the solid line represents the frequency distribution of the original sound,
The dotted line represents the minimum audible limit in silence, and the shaded rectangle represents the sample range.

The psychoacoustic encoding process will be described in detail with reference to FIG. In the psychoacoustic encoding process, first, a threshold (not shown) called a masking threshold, which is a limit that can be sensed while listening to the original sound, is obtained. Next, the scale factor indicating the magnification at the time of waveform reproduction is adjusted according to the maximum amplitude included in the original sound of each subband. And
A quantization step for each sub-band is assigned so that the quantization noise generated by quantization is lower than a threshold for each sub-band. In a portion where the original sound is smaller than the threshold value, the original sound is not audible and can be omitted from the sample. The encoding target can be determined by an encoding target flag (not shown). The encoding target flag sets “1” to a band to be encoded and sets “0” to a band not to be encoded. In the figure, 0 bits are assigned to the sub-bands 10 to 15 and the sub-bands 18 to 31 because the original sound is at a level lower than the threshold value, and “0” is set to the encoding target flag. Further, since subbands 0 to 9 and subbands 16 to 17 are audible bands, “1” is set to the encoding target flag. It is necessary to reduce the quantization noise level by making the quantization step fine in portions where the sensitivity of hearing is high and the masking is not sufficiently working.

The bit allocation section 300 adjusts the number of bits to be allocated so as to satisfy both the bit rate and the SMR requirements while considering the output samples of the filter bank 100, and generates an appropriate bit allocation table.
Then, the signal of each sub-band is quantized by the allocated number of bits, and transmitted to the bit stream forming unit 400 as a quantized sample.

[0011] The bit stream forming section 400 codes the quantized sample using a PCM (pulse code modulation) code or a Huffman code to form a bit stream.

FIG. 13 is a flowchart of an encoding process in a conventional MPEG / Audio encoder. Hereinafter, the encoding process of audio data by the conventional MPEG / Audio encoder will be described. For the input signal input to the encoder, an FFT (Fast Fourier Transfor
m: fast Fourier transform), while performing sub-band analysis in the filter bank 100.

First, the input signal is converted into data on the frequency axis by FFT, spectrum analysis is performed, and data for calculating SMR relating to PCM samples is obtained (step S101).

On the other hand, in the sub-band analysis, first, a sub-band encoding method (layer 1, layer 2, and layer 3 in MPEG) is specified and a sampling frequency for performing sampling is specified. Next, the entire audio frequency band of one frame is divided into 32 bands (sub-bands) by the filter bank 100 according to the specified sub-band coding scheme (step S102).

Next, the data obtained by the sub-band analysis is separated into a waveform and a magnification, and the waveform has a maximum amplitude of 1.
It is normalized to be 0 (step S103). The magnification is obtained as a scale factor.

Then, in the psychoacoustic model unit 200, the above-mentioned masking threshold is obtained, and the SMR of each subband is calculated (step S104). Thereafter, the SMR and the designated bit rate (step S10
A bit allocation process of allocating an appropriate number of bits to each subband based on 5) is performed (step S10).
6). The conventional bit allocation process executed here will be described based on the flowchart of FIG.

In FIG. 14, first, the CPU 2 causes the bit allocating section 300 to set the number of available bits N according to a designated bit rate (for example, 128 kbps).
Is calculated (step S111), and first, the number of bits of each subband is initialized to zero (step S11).
2).

Next, the CPU 2 calculates a signal-to-noise ratio (SNR) of each sub-band (step S113), and subtracts the SMR from the calculated SNR to obtain a mask-to-noise ratio (MN).
R) is calculated (step S114), and the encoding target subband having the smallest MNR is determined (step S115).
Then, a predetermined number of bits C preset as a default in the encoding target sub-band having the smallest MNR is set.
(For example, 1 bit) is assigned, the assigned bit number is added (step S116), and the total bit number M is calculated (step S117). The calculated total bit number M is compared with the available bit number N (step S118). If the total bit number M is smaller than the available bit number N (step S118; NO), step S118 is performed.
Returning to step 113, the process is repeated.

As described above, the predetermined number of bits C is repeatedly added to the sub-band having the smallest MNR, and the total number of bits M of the target sub-band is calculated each time.
Is repeated until the number of available bits exceeds N. If the total bit number M exceeds the usable bit number N (step S118; YES), the bit number added immediately before is canceled (step S119), a bit allocation table is acquired (step S120), and the bit allocation is performed. The process ends.

When the above bit allocation processing is completed,
The CPU 2 proceeds to step S107 in FIG. That is, since some bands contain only signals of a level that is not recognized by the auditory system depending on the SMR, these are set as sub-bands that do not transmit (step S107). Then, the encoding target flag of the sub-band not to be transmitted is set to “0”, and the scale factor of the sub-band to be transmitted is encoded. (Step S108). Further, the encoding target subband, the scale factor, and the encoding target flag are quantized and encoded (step S109),
The bit stream forming unit 400 forms a bit stream and transmits it to a communication path or the like (step S110).

[0021]

However, in the conventional method, bit allocation is performed at a specified bit rate for all subbands. As a result, in the case of a low bit rate, the number of bits that can be used is originally insufficient, so that a sufficient number of bits cannot be allocated to each subband, quantization noise is generated, and annoying noise is generated. I was

SUMMARY OF THE INVENTION It is an object of the present invention to provide an audio data compression device and a storage medium capable of improving the perceived sound quality of audio data encoded at a low bit rate.

[0023]

The present invention has the following features in order to achieve the above object. In the following description of the means, a configuration corresponding to the embodiment will be exemplified by parentheses as an example. Reference numerals and the like are reference numerals and the like in the drawings described later.

According to the first aspect of the present invention, there is provided an audio data compression apparatus which divides audio data of each frame into a plurality of bands, allocates an appropriate number of bits to each of the divided bands, and performs quantization. And a coding band limiting unit (for example, a coding sub band determining unit shown in FIG. 3) for limiting a band to be coded among the plurality of bands according to a bit rate. Unit 23 and steps S5 to S6 shown in FIG. 5) and the band limited by the coding band limiting unit.
Bit allocation means (for example, a bit allocation unit 300 shown in FIG. 3) for allocating the number of bits according to the specified bit rate.

According to the audio data compression apparatus 1 of the present invention, the audio data of each frame is divided into a plurality of bands, and an appropriate number of bits is assigned to each of the divided bands and quantized. In the audio data compression device, according to a bit rate, an encoding band limiting unit limits a band to be encoded among the plurality of bands, and a bit allocating unit is limited by the encoding band limiting unit. Only for the specified band, the number of bits according to the specified bit rate is allocated.

Therefore, according to the first aspect of the present invention, the band to be encoded can be limited according to the bit rate, and bits can be allocated to the limited band, so that it can be used according to the bit rate. A large number of bits can be efficiently allocated to the band to be encoded, so that harsh noise can be limited and the perceived sound quality can be improved.

As in the invention according to claim 2, the audio data compression apparatus according to claim 1, further comprising designation means (for example, a bit rate designation unit 22 shown in FIG. 3) for designating the bit rate. 2. The coding band limiting unit, when the bit rate specified by the specifying unit is lower than a predetermined bit rate, limits a band to be coded among the plurality of bands. An audio data compression device according to claim 1.

According to the second aspect of the present invention, when the designated bit rate is lower than the predetermined bit rate, the band to be coded among the plurality of bands can be limited. Bits can be sufficiently allocated to a limited band even at a low bit rate, and the sound quality is improved.

In the audio data compression apparatus according to the first aspect of the present invention, the encoding band limiting means allocates a predetermined number of bits to each band to be encoded, and assigns a predetermined number of bits to each band. A determination unit (for example, a coding subband determination unit 300 shown in FIG. 3 and a step shown in FIG. 6) that determines whether or not the total sum of the allocated bit numbers falls within the number of usable bits according to the bit rate. S21 to S24) and an encoding target band reducing unit (for example, FIG. 3) that removes the highest band from the encoding target when the sum does not fall within the number of usable bits according to the bit rate. And the steps S25 to S26 shown in FIG.
And the following.

According to the third aspect of the present invention, predetermined bits are assigned to each band to be encoded, and the sum of the number of bits assigned to each band is determined by the number of usable bits according to the bit rate. , The highest band can be excluded from the coding target and the coding band can be limited, so that bits can be allocated to the low band and the middle band, and the sound quality of the low band and the middle band is improved.

In the audio data compression apparatus according to the first aspect of the present invention, the encoding band limiter assigns a predetermined number of bits to each band to be encoded, and assigns a predetermined number of bits to each band. A determination unit (for example, a coding subband determination unit 300 shown in FIG. 3 and a step shown in FIG. 6) that determines whether or not the total sum of the allocated bit numbers falls within the number of usable bits according to the bit rate. S21 to S24), and if the sum does not fall within the number of usable bits in accordance with the bit rate, the band for which the energy value of the signal to be encoded is minimum is excluded from the encoding target. Target band reducing means (for example, the encoding sub-band determining unit 300 shown in FIG. 3);
It is characterized by having.

According to the fourth aspect of the present invention, predetermined bits are allocated to each band to be coded, and the sum of the number of bits allocated to each band is set to the number of usable bits according to the bit rate. Remove the band with the minimum energy value of the encoding target signal from the encoding target so as to fit,
Since the encoding band can be limited, it is possible to efficiently allocate the number of bits available according to the bit rate to the limited band to the band to be encoded,
, And the sound quality of the entire band to be encoded is improved.

According to a fifth aspect of the present invention, in the audio data compressing apparatus according to the first aspect, the encoding band limiting means determines a minimum number of bits required for each band to be encoded according to a degree of quantization distortion. And a required minimum bit number allocating means (for example, a required minimum bit number setting unit 31 shown in FIG. 7 and step S36 shown in FIG. 8), and a required minimum bit number allocated for each band by the required minimum bit number allocating means. Determination means for determining whether or not the sum of the number of bits falls within the number of usable bits according to the bit rate (for example, the coding subband determination unit 33 shown in FIG. 7 and steps S61 to S64 shown in FIG. 10) ), And when the sum does not fall within the number of usable bits according to the bit rate, the highest band is excluded from the encoding target. Encoding target bandwidth reduction means (e.g.,
And the coding sub-band determining unit 33 shown in FIG. 3 and steps S65 to S66 shown in FIG.

According to the fifth aspect of the present invention, the required minimum number of bits for each band to be encoded is calculated from the degree of quantization distortion and assigned, so that the bits can be assigned efficiently. Further, the minimum necessary number of bits is allocated to each band to be coded, and the highest band is subjected to coding so that the total number of bits allocated to each band falls within the number of bits available according to the bit rate. , And the bandwidth can be limited, so that the required minimum number of bits can be assigned to the limited bandwidth,
The perceived sound quality is improved.

According to a sixth aspect of the present invention, in the audio data compression apparatus according to the first aspect, the coding band limiting means includes a minimum bit number necessary for each band to be coded according to a degree of quantization distortion. A minimum bit number allocating unit (for example, the minimum bit number setting unit 31 shown in FIG. 7 and step S36 shown in FIG. 8), and each required minimum bit allocated to each band by the minimum bit number allocating unit. Determining means for determining whether or not the sum of the numbers falls within the number of usable bits according to the bit rate (for example, the coding subband determining unit 3 shown in FIG. 7)
3 and steps S61 to S64 shown in FIG. 10), and when the sum does not fall within the number of usable bits in accordance with the bit rate, the band having the minimum energy value of the encoding target signal is determined. 2. The audio data compression apparatus according to claim 1, further comprising: an encoding target band reducing unit (for example, an encoding subband determining unit 33 illustrated in FIG. 7) to be excluded from an encoding target.

According to the sixth aspect of the present invention, the required minimum number of bits for each band to be coded is calculated from the degree of quantization distortion and allocated, so that the bits can be allocated efficiently. Further, the necessary minimum number of bits is assigned to each band to be encoded, and the energy of the signal to be encoded is set such that the total number of bits assigned to each band falls within the number of usable bits according to the bit rate. Since the band with the smallest value can be excluded from the encoding target and the band can be limited, the required minimum number of bits can be allocated to the limited band, and the perceived sound quality is improved.

In the audio data compression apparatus according to the fifth or sixth aspect of the present invention, the necessary minimum number of bits allocating means assigns each of the bands so that the quantization noise does not exceed a predetermined value. It is characterized in that the required minimum number of bits is calculated and assigned.

According to the seventh aspect of the present invention, the necessary minimum number of bits to be allocated to each band to be encoded is allocated so that the quantization noise does not exceed a predetermined value. Is improved.

[0039]

DESCRIPTION OF THE PREFERRED EMBODIMENTS (First Embodiment) FIG.
An embodiment of the audio data compression device 1 according to the present invention will be described in detail with reference to FIG.

First, the configuration will be described. FIG. 1 is a diagram showing an audio data compression / playback system configured using an audio data compression device 1 according to the present embodiment. In FIG. 1, in an audio data compression device 1 having an audio codec unit 7 described later, a sound source 110 and a portable audio player 120 are connected via an interface.

The sound source 110 is a CD (Compact Dis
k), a disk 111 storing an audio signal such as an MD (Mini Disk) or the like, and a DAT (Digital Audio Taperecorder) 1 which is a tape device capable of recording sound in a digital system.
And MIDI (Musical) which is a standard protocol for controlling electronic musical instruments such as synthesizers and samplers.
Instrument Digital Interface) 113.
The sound source 110 is a disk, DAT, MID
Not limited to I, any type may be used as long as audio data is stored in a digital system. For example, audio data that can be downloaded via the Internet may be used as a sound source.

The portable audio player 120 is fixedly or detachably provided with a small and portable semiconductor storage medium having a built-in storage medium such as a nonvolatile semiconductor storage element such as a flash memory. A reproduction function for reproducing a sound based on the audio data stored in the storage device.

In the audio data compression / reproduction system shown in FIG. 1, audio data from the sound source 110 is compressed and encoded by the audio data compression device 1. The compression-encoded audio data is transferred to the portable audio player 120 according to a user's instruction.
Output to the side. And portable audio player 1
The audio data stored in the semiconductor storage medium 20 and stored according to a user's instruction is expanded and reproduced.

FIG. 2 is a block diagram showing the internal configuration of the audio data compression device 1 according to the first embodiment. In FIG. 2, an audio data compression device 1
Is a CPU (Central Processing Unit) 2, an input device 3, a display device 4, a RAM (Random Access Memory)
5, communication unit 6, acoustic codec unit 7, external I / O interface 8, storage device 9, storage medium 10
Each part except the storage medium 10 is connected by a bus 11.

The CPU 2 sends an application program designated from among various application programs corresponding to the audio data compression device 1 stored in the storage device 9 and various instructions input from the input device 3 to a work memory in the RAM 5. Various processing is executed in accordance with the application program stored in the RAM 5 in accordance with the input instruction and the input data, and the processing results are stored in the work memory area in the RAM 5 and displayed on the display device 4. Then, the processing result stored in the work memory area is stored in a storage destination in the storage device 9 designated by the input device 3.

The CPU 2 encodes the sound source input through the communication unit 6 and the audio data transmitted from the Internet by passing the audio data to the audio codec unit 7 in accordance with an encoding instruction from the input device 3 and forms the encoded data. The stored bit stream is stored in the storage device 9. Then, the CPU 2 stores the bit stream formed in the audio codec unit 7 in the semiconductor storage medium of the portable audio player 120 via the external I / O interface 8 in accordance with an instruction from the input device 3, Via an external personal computer.

The CPU 2 encodes the audio data by the audio codec unit 7 in accordance with the bit rate specified by the input device 3. And CPU2
Determines whether the designated bit rate is lower than a predetermined bit rate (for example, 128 bps), and if it is determined that the bit rate is lower than the predetermined bit rate (for example, 128 bps), a subband determination process to be described later (FIG. 6) See).

Further, the CPU 2 allocates the set predetermined number of bits C to each encoding sub-band and calculates the bit sum S in the sub-band determining process (see FIG. 6) in the audio codec unit 7. . If the calculated bit sum S does not fall within the number of usable bits N according to the specified bit rate, the CPU 2 removes the highest-band sub-band from the sub-bands to be encoded, and will be described later. An encoding target sub-band table is obtained and stored in the RAM 5.

The input device 3 includes a mouse or a keyboard having cursor keys, numeric input keys, various function keys, and the like. The input device 3 outputs to the CPU 2 a pressing signal pressed by the keyboard and a mouse position signal.

The display device 4 is a CRT (Cathode Ray Tu)
be) or a liquid crystal display
A signal based on the display data input from U2 is generated to perform various displays.

The RAM 5 has a work memory area for storing designated application programs, input instructions, input data, processing results, and the like.

The communication unit 6 has a modem (MODEM:
Modulator / DEModulator) or a terminal adapter (TA), and controls communication with an external device via a communication line such as a telephone line or an ISDN line. The modem modulates the digital data processed by the CPU 2 into an analog signal matching the frequency band of the telephone line, and communicates with the external device such as a personal computer via the telephone line. A device for demodulating an input analog signal into a digital signal. The terminal adapter converts an existing interface into an interface corresponding to ISDN in order to communicate with an external device such as a personal computer via an ISDN line. Device.

The audio codec unit 7 has an external I / O
Encoding processing is performed on audio data input from the sound source 110 via the interface 8 or audio data transmitted from an external device via the communication unit 6. The encoder includes a filter bank 100 that is a band division method, an acoustic psychological model unit 200 that generates a set of data for controlling quantization and encoding, and outputs the generated data as an SMR, and the number of bits to be allocated according to the bit rate and the SMR. And a bit allocator 300 that quantizes the signal of each subband with the allocated number of bits,
The quantized sample is basically configured by a bit stream forming unit 400 or the like that forms a bit stream by coding the PCM code or the Huffman code.

FIG. 3 is a block diagram showing the configuration of the encoder of the audio codec unit 7 in the first audio data compression device 1. In the present embodiment, the audio codec unit 7 of the audio data compression apparatus 1 has the same components (the filter bank 100, the psychoacoustic model unit 200, the bit allocation unit) as those of the conventional MPEG / Audio encoder shown in FIG. The detailed description of the unit 300 and the bit stream forming unit 400) is omitted, and the same reference numerals are given.

As shown in FIG. 3, the encoder of the acoustic codec unit 7 has the same components as the conventional MPEG / Audio encoder (filter bank 100, psychoacoustic model unit 200, bit allocation unit 300, bit stream Forming section 400), predetermined bit number setting section 21, bit rate specifying section 22, coding subband determining section 23
And

The predetermined bit number setting section 21 sets a predetermined bit number C which is the number of bits allocated to each subband to be encoded. The predetermined number of bits C is obtained by the following equation (1).

The predetermined number of bits C assigned to each band = (the number of usable bits N at the predetermined bit rate B) ÷ the number of subbands (1)

Here, the predetermined bit rate B is a bit rate preset in the audio codec unit 7 as a default. The number N of usable bits corresponding to the predetermined bit rate B is obtained by the following equation (2).

Available number of bits N = (number of samples per frame 当 た り sampling frequency) × bit rate (2)

Here, the number of samples per frame and the sampling frequency are determined by the filter bank 100 described above.
Depends on the subband coding method and sampling frequency specified in the subband analysis in. For example, when layer 1 is specified, the number of samples is 32 bands × 12
It becomes 384 in a sample. Further, a predetermined bit rate B is set to the “bit rate” in the equation (2).

FIG. 4 shows that the layer 1 (38) is obtained by using the equation (2).
4 samples), sampling frequency (44.1 kHz)
FIG. 11 is a diagram showing, as a table, a result of calculating the number of usable bits for each bit rate at the time of designation.

For example, as shown in FIG. 4, when the layer 1 and the sampling frequency (44.1 kHz) are designated and the designated bit rate A is designated as 128 kbps, the number N of usable bits becomes 1114 bits. Become.

In the bit rate specification section 22, the user specifies the bit rate. The bit rate that can be specified by the user is 32 kbp as shown in the table shown in FIG.
Each bit rate from s to 320 kbps. When the bit rate is specified from the table of FIG. 4, the number N of usable bits for the bit rate is determined. The designated bit rate is called designated bit rate A. Here, the number N of usable bits is calculated by setting the designated bit rate A to “bit rate” in the equation (2).

The encoding sub-band determining unit 23 includes a predetermined number of bits C set by the predetermined number of bits setting unit 21, a specified bit rate A specified by the bit rate specifying unit 22,
The subband to be encoded is determined according to. The bit sum S when the predetermined number of bits C is allocated to each encoding target subband is obtained by the following equation (3).

Bit sum S = predetermined number of bits C × number of subbands to be encoded (3)

If the bit sum S obtained here is equal to or more than the number N of bits that can be used according to the designated bit rate A, the highest sub-band is determined and the determined sub-band is excluded from the encoding target. . The encoding subband determination processing (see FIG. 6) is performed by this method.

The encoding sub-band determining section 23 determines a sub-band to be encoded and stores it in the RAM 5 as an encoding sub-band table. The encoding subband table is a table that stores information on the encoding target subband, and is encoded together with the encoding subband.

The external I / O interface 8 is connected to the RS2
It is configured by a bus such as 32C or GPIB, and controls input and output of data. In FIG. 1, a transfer operation of audio data transmitted and received between the audio data compression device 1 and the sound source 110 is controlled. Also, the audio data compression device 1 and the portable audio player 1
20 controls the transfer operation of the compressed audio data exchanged with the H.20.

The storage device 9 has a storage medium 10 in which programs, data and the like are stored in advance, and this storage medium 10 is constituted by a magnetic or optical storage medium or a semiconductor memory. This storage medium 10 is a storage device 9
The storage medium 10 stores a system program, various application programs corresponding to the system, data processed by each processing program, and the like.

The programs, data, and the like stored in the storage medium 10 are partially or wholly received from another unit such as a server or a client from the communication unit 6 via a transmission medium such as a network line and stored. The storage medium 10 may be a storage medium of a server built on a network. Further, the program may be configured to be transmitted to a server or a client via a transmission medium such as a network line and installed in these devices.

Next, the operation will be described. FIG. 5 is a flowchart showing an encoding process executed by the audio codec unit 7 of the audio data compression device 1 in the first embodiment.

In the encoding process shown in FIG. 5, steps S101 to S10 of the conventional encoding process shown in FIG.
Similarly to 4, the audio data input from the sound source 110 is divided into 32 bands (sub-bands), the scale factor of each sub-band is obtained, and the SMR of each sub-band is calculated (steps S1 to S1). S4).

Then, the designated bit rate A specified in advance by the bit rate specifying unit 22 is compared with a predetermined bit rate B preset in the audio codec unit 7 as a default (step S5).

When the designated bit rate A is equal to the predetermined bit rate B
If so (step S5; YES), step S7
When the designated bit rate A is smaller than the predetermined bit rate B (step S5; NO), the encoding sub-band determining unit 23 executes the encoding sub-band determining process shown in FIG. 6 (step S6). .

FIG. 6 is a flowchart showing the flow of the encoding subband determination process. In FIG. 6, first, the number of usable bits N is calculated according to the designated bit rate A (step S21).

Next, the predetermined number of bits C set by the predetermined number of bits setting section 21 is allocated to each encoded subband (step S22). Then, a bit sum S, which is the total number of bits allocated to each encoded subband, is calculated (step S23), and the bit sum S is compared with the available bit number N (step S24).

If the bit sum S is larger than the number N of usable bits (step S24; NO), the coding subband of the highest band among the coding subbands whose coding target flag is set to "1" is set. Determine the band (step S2
5) After removing the determined subband from the encoding target, and setting the encoding target flag of the subband to “0” (step S26), the process returns to step S23.

By the processing of steps S23 to S26,
Sub-bands to be coded are sequentially removed until the total bit number S becomes smaller than the number N of usable bits, and if the total bit S becomes smaller than the number N of usable bits (step S24; YES), the encoding is performed. Determine the target subband,
An encoding target subband table (not shown) for the encoding target subband is acquired (step S27). Then, the encoding sub-band determination process ends, and the process proceeds to step S7 in the flowchart shown in FIG.

Then, the necessary number of bits is allocated to each subband to be encoded by the encoding subband determination processing. This bit allocation is the same as in step S106 of the conventional encoding process shown in FIG. Then, step S10 of the conventional encoding process shown in FIG.
As in steps 7 to S110, a sub-band not to be transmitted is determined, and a scale factor of the sub-band to be transmitted is encoded. Further, the sub-band to be transmitted is quantized and encoded, and a bit stream is formed in the bit stream forming unit 400 from the quantized sample and the scale factor, and is transmitted to a communication path or the like (steps S7 to S7 in FIG. 5). S11).

An example of the progress of the process will be specifically described with reference to FIG. Layer 1, sampling frequency 44.1 kHz
z, the designated bit rate A is 32 kbps, and the predetermined bit rate B (default) is 128 kbps.

In FIG. 1, the audio data compression apparatus 1 fetches audio data from the sound source 110, specifies the layer 1 by the processing of steps S1 to S4, divides it into 32 sub-bands, Obtain the scale factor corresponding to the band and calculate the SMR for each subband. Then, by the processing of steps S5 to S6, the predetermined bit rate B (1
28 kbps) and designated bit rate A (32 kbps)
s), and since the designated bit rate A is smaller than the predetermined bit rate B, the sub-band determination processing for encoding (steps S21 to S27 in FIG. 6) is executed.

In the subband determination process to be encoded, the number of bits N that can be used according to the designated bit rate A
(Available bits corresponding to a bit rate of 32 kbps in FIG. 4: 278 bits) are calculated, and a predetermined number of bits C (34 bits) is allocated to each subband (32 bands). Then, the bit sum S of each encoded subband is
(34 bits × 32 bands: 1088 bits) is calculated, and the bit sum S is compared with the number N of usable bits. Since the bit sum S is larger than the number N of usable bits, the highest encoding subband (one band) is determined from the encoding subbands for which the encoding target flag is set to “1”. Remove subbands from encoding. Then, the bit sum S (34 bits × 31 bands:
1054 bits) again, and the processing is continued so that the bit sum S does not exceed the number N of usable bits. After allocating the number N of usable bits to each subband,
The encoding target subband is determined, and the encoding target subband table is obtained.

Then, bits are allocated up to the number of usable bits N (278 bits) for the subband to be encoded determined by the processing of steps S7 to S11 (see FIG. 14), and a subband not to be transmitted is determined. .
The scale factors of the determined non-transmission sub-band and the determined transmission sub-band are encoded. Further, the sub-band to be transmitted is quantized and coded by a coding method corresponding to a layer to be a quantized sample. A bit stream is formed in the bit stream forming unit 400 from the quantized sample and the scale factor and transmitted to a communication path or the like.

As described above, in the first embodiment, the audio data compression device 1 compares the specified bit rate A with the predetermined bit rate B, and determines that the specified bit rate A is a low bit rate. If it is determined that there is, a predetermined number of bits C is assigned to the coding subband, and the coding subband is determined so that the calculated bit sum S falls within the usable bit number N. , The high-frequency sub-bands are sequentially removed, and an encoding target sub-band table is obtained. Then, only for the limited encoding target subband, the number of usable bits is allocated according to the designated bit rate.

Therefore, even when a low bit rate is specified, high-frequency sub-bands that are not likely to affect sound quality are excluded from coding targets and bits are preferentially allocated from low-frequency and mid-frequency bands. Therefore, the sound quality in the low and middle ranges is improved. Thereby, harsh noise can be limited, and the perceived sound quality can be improved.

When a predetermined number of bits C is allocated to each encoding sub-band, and the bit sum S of the allocated predetermined number of bits C does not fall within the usable number of bits N, SM
A sub-band with a low R may be removed.

(Second Embodiment) Next, an audio data compression apparatus 1 to which the present invention is applied with reference to FIGS.
A second embodiment will be described. The configuration of the audio data compression apparatus 1 according to the second embodiment is the same as that of the audio data compression apparatus 1 according to the first embodiment shown in FIG. Elements will be described below with the same reference numerals.

In the second embodiment, the acoustic codec unit 7 executes a necessary minimum bit number calculation process (see FIG. 9) to calculate the minimum number of bits required for each subband according to the degree of quantization distortion. I do. That is, the quantization / dequantization is performed on all the sub-bands, and the required minimum number of bits is obtained using a square error energy Ee obtained by a square error between the quantized / dequantized signal and the original input signal. Is calculated. Further, the CPU 2 registers the calculated required minimum number of bits in a bit allocation table (not shown) and stores the number in the RAM 5.

The CPU 2 determines the subband to be encoded in the audio codec unit 7 (see FIG. 10).
), The required minimum number of bits calculated by the required minimum number of bits calculation process is allocated to each encoding subband, and the bit sum S is calculated. And CPU2
If the calculated bit sum S does not fall within the number N of usable bits, S
The sub-band with the smallest MR is removed, and the above-described encoding target sub-band table is acquired and stored in the RAM 5.

FIG. 7 is a block diagram showing the configuration of the encoder of the acoustic codec unit 7 in the second audio data compression device 1. In the present embodiment, the audio codec unit 7 of the audio data compression apparatus 1 has the same components (the filter bank 100, the psychoacoustic model unit 200, the bit allocation unit) as those of the conventional MPEG / Audio encoder shown in FIG. The detailed description of the unit 300 and the bit stream forming unit 400) is omitted, and the same reference numerals are given.

As shown in FIG. 7, the encoder of the audio codec unit 7 has the same components as the conventional MPEG / Audio encoder (filter bank 100, psychoacoustic model unit 200, bit allocation unit 300, bit stream Forming unit 400) and the minimum necessary bit number setting unit 31
, A bit rate designating unit 32, and a coding subband determining unit 33.

[0092] The required minimum bit number setting unit 31 calculates the required minimum bit number according to the degree of quantization distortion, and sets it in the bit allocation table. The necessary minimum number of bits is obtained by a later-described required minimum number of bits calculation process (see FIG. 9) using a square error energy or the like obtained by a square error between the quantized / dequantized signal and the original input signal. Is calculated. Here, the square error energy is obtained by the following equation (4).

The square error energy = ∫ (f ′ (t) −f (t)) ² dt (4)

Here, f ′ (t): a time function of the restored signal obtained by quantizing the input signal f (x) and then dequantizing the signal f (t): a time function of the input signal f (x).

The square error energy obtained by the necessary minimum number of bits setting unit 31 is equal to a certain value T (for example,
If it is larger than 100), it is determined that unpleasant quantization distortion will occur, and the necessary minimum number of bits is set in the bit allocation table so as not to exceed the fixed value T.

In the bit rate designating section 32, like the bit rate designating section 22 in the first embodiment,
The bit rate is specified by the user. The bit rates that can be specified by the user are, for example, the respective bit rates shown in the table shown in FIG.

[0097] The encoding sub-band determining section 33 determines a sub-band to be encoded according to the designated bit rate and the calculated required minimum number of bits. If the total bit number S when the necessary minimum number of bits is allocated to each subband to be encoded is equal to or more than the number N of bits that can be used according to the designated bit rate A, the subband with the smallest SMR is determined and determined. Remove subbands from encoding. Encoding subband determination processing by this method (see FIG. 10)
I do.

The encoding subband determining section 33 determines a subband to be encoded and stores it in the RAM 5 as an encoding subband table. The encoding subband table is a table that stores information on the encoding target subband, and is encoded together with the encoding subband.

Next, the operation will be described. FIG. 8 is a flowchart showing an encoding process performed by the audio codec unit 7 of the audio data compression device 1 in the second embodiment.

In the encoding process shown in FIG. 8, steps S101 to S10 of the conventional encoding process shown in FIG.
Similarly to 4, the audio data input from the sound source 110 is divided into 32 bands (sub-bands), the scale factor of each sub-band is obtained and encoded, and the SMR of each sub-band is calculated (step). S31 to step S34).

Then, the specified bit rate A specified by the bit rate specifying section 22 is compared with a predetermined bit rate B preset in the audio codec unit 7 as a default (step S35).

When the designated bit rate A is equal to the predetermined bit rate B
If so (step S35; YES), step 3
Then, as in step S106 of the conventional encoding process shown in FIG. 13, the necessary number of bits is allocated to each subband to be encoded.

If the designated bit rate A is smaller than the predetermined bit rate B (step S35; NO), the encoding subband determination unit 33 executes the processing for calculating the required minimum number of bits shown in FIG. 9 (step S35). S36).

FIG. 9 is a flowchart showing the flow of the necessary minimum bit number calculation process. In FIG. 9, first, a table number counter SBn of the encoding subband table is set to 0 (step S51). Next, the number of quantization bits of the 0th subband is set to 1 (step S
52). Then, quantization / inverse quantization is performed (step S
53), the square error energy Ee is calculated based on the above equation (4) (step S54), and the square error energy Ee is compared with a constant value T (step S5).
5).

If the square error energy Ee is larger than the fixed value T (step S55; NO), the number of quantization bits is increased by 1 to reduce quantization distortion (step S5).
6), and return to step S53.

When the square error energy Ee is constant T
If not (step S55; YES), the calculated number of quantization bits is registered in the bit allocation table as the required minimum number of bits (step S57), and 1 is added to the table number counter SBn of the encoding subband table. The process for the next sub-band is performed (step S58).

Then, the necessary minimum number of bits is sequentially allocated to each sub-band, and it is determined whether or not the calculation of the required minimum number of bits has been completed for all the sub-bands (step S5).
9) If the calculation of the required minimum number of bits has not been completed for all subbands (step S59; NO), the process returns to step S52 to continue the processing. When the calculation of the required minimum number of bits for all subbands is completed (step S59; YES), the required minimum number of bits calculation process is completed, and the process proceeds to step S37 in FIG.

Next, the encoding sub-band determining section 33 executes a sub-band determining process for encoding (step S37 in FIG. 8). The subband determination process will be described with reference to the flowchart shown in FIG.

In the subband determination process for encoding shown in FIG. 10, first, the number N of bits that can be used is calculated according to the specified bit rate A specified by the bit rate specifying unit 32 (step S61).

Next, based on the bit allocation table acquired by the necessary minimum bit number setting unit 31, the minimum number of bits is allocated to each encoded subband (step S6).
2). Then, the bit sum S of each encoded subband is calculated (step S63), and the bit sum S is compared with the number N of usable bits (step S64).

If the bit sum S is larger than the number N of usable bits (step S64; NO), a band having the lowest SMR is determined from the bands to be encoded (step S65), and the determined sub-band is determined. It is removed from the encoding target (step S66), and the process returns to step S63.

The processes of steps S63 to S66 are repeated until the total bit S becomes smaller than the number N of usable bits. When the total bit S becomes smaller than the number N of usable bits (step S64; YES), the encoding target The subband is determined, and an encoding target subband table is obtained (step S67). Then, the encoding sub-band determination processing ends, and the flow shifts to step S39 of the encoding processing in FIG.

Then, as in steps S107 to S110 of the conventional encoding process shown in FIG. 13, a sub-band not to be transmitted is determined, and a scale factor for the sub-band to be transmitted is encoded. Further, the sub-band to be transmitted is quantized and encoded, a bit stream is formed in the bit stream forming unit 400 from the quantized sample and the scale factor, and the bit stream is transmitted to a communication path or the like (steps S39 to S42).

As described above, according to the audio data compression apparatus 1 of the second embodiment, the necessary minimum number of bits is calculated in advance so that the square error energy becomes smaller than a certain value T for each subband. In advance, the calculated minimum required number of bits is assigned to each subband,
The subbands with the smallest SMR are sequentially removed from the encoding subbands so that the calculated bit sum S falls within the number N of usable bits, and the encoding target subband is acquired.
Then, only for the limited subbands to be encoded,
Allocate the number of available bits according to the specified bit rate.

Therefore, the required minimum number of bits is calculated according to the degree of quantization distortion and assigned to each subband, so that bits can be efficiently assigned and the perceived sound quality can be improved. Further, even when the number of bits is insufficient, a band having a low SMR is excluded from the encoding target, so that the muffled sound can be reduced and the perceived sound quality can be improved.

Note that the present invention is not limited to the contents of the above-described embodiment, but can be appropriately changed without departing from the gist of the present invention. You may make it reproduce | regenerate.

Although the required minimum number of bits is calculated based on the square error energy, SNR or the like may be used.

When the calculated required minimum number of bits is allocated to each sub-band, and the calculated total bit number S does not fall within the usable number of bits N, the sub-band with a low SMR is removed. As in the first embodiment, the high frequency sub-band may be removed.

[0119]

According to the first and eighth aspects of the present invention, the band to be encoded is limited according to the bit rate.
Since bits can be allocated to a limited band, the number of available bits can be efficiently allocated to the band to be encoded according to the bit rate, thereby limiting harsh noise and improving the perceived sound quality. Can be.

According to the second aspect of the present invention, when the designated bit rate is lower than the predetermined bit rate,
Since the band to be encoded among the plurality of bands can be limited, sufficient bits can be allocated to the limited band even at a low bit rate, and the sound quality is improved.

According to the third aspect of the present invention, predetermined bits are allocated to each band to be coded, and the sum of the number of bits allocated to each band is set to the number of usable bits according to the bit rate. As it fits, the highest band is excluded from the encoding target and the coding band can be limited, so that bits can be allocated to the low band and the middle band,
The sound quality in the low and middle ranges is improved.

According to the fourth aspect of the present invention, predetermined bits are allocated to each band to be coded, and the sum of the number of bits allocated to each band is set to the number of usable bits according to the bit rate. Remove the band with the minimum energy value of the encoding target signal from the encoding target so as to fit,
Since the encoding band can be limited, it is possible to efficiently allocate the number of bits available according to the bit rate to the limited band to the band to be encoded,
, And the sound quality of the entire band to be encoded is improved.

According to the fifth aspect of the present invention, the required minimum number of bits for each band to be coded is calculated from the degree of quantization distortion and allocated, so that the bits can be allocated efficiently. Further, the minimum necessary number of bits is allocated to each band to be coded, and the highest band is subjected to coding so that the total number of bits allocated to each band falls within the number of bits available according to the bit rate. , And the bandwidth can be limited, so that the required minimum number of bits can be assigned to the limited bandwidth,
The perceived sound quality is improved.

According to the sixth aspect of the present invention, the required minimum number of bits for each band to be coded is calculated from the degree of quantization distortion and allocated, so that the bits can be allocated efficiently. Further, the necessary minimum number of bits is assigned to each band to be encoded, and the energy of the signal to be encoded is set such that the total number of bits assigned to each band falls within the number of usable bits according to the bit rate. Since the band with the smallest value can be excluded from the encoding target and the band can be limited, the required minimum number of bits can be allocated to the limited band, and the perceived sound quality is improved.

According to the seventh aspect of the present invention, the minimum number of bits required to be allocated to each band to be coded is allocated so that the quantization noise does not exceed a predetermined value. Is improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing an audio data compression / playback system.

FIG. 2 is a block diagram illustrating an internal configuration of the audio data compression device 1.

FIG. 3 is a block diagram showing a configuration of an encoder of the audio codec unit 7;

FIG. 4 is a table showing a result of calculating the number of usable bits for each bit rate.

FIG. 5 is a flowchart illustrating a flow of an encoding process.

FIG. 6 is a flowchart illustrating a flow of an encoding subband determination process.

7 is a block diagram illustrating a configuration of an encoder of the audio codec unit 7. FIG.

FIG. 8 is a flowchart illustrating a flow of an encoding process.

FIG. 9 is a flowchart illustrating a flow of a process of calculating a required minimum number of bits.

FIG. 10 is a flowchart illustrating a flow of a subband determination process.

FIG. 11 is a block diagram showing a configuration of a conventional MPEG / Audio encoder.

FIG. 12 is a diagram showing the concept of psychoacoustic encoding processing that is processed in the psychoacoustic model unit 200.

FIG. 13 is a flowchart showing a flow of an encoding process in a conventional MPEG / Audio encoder.

FIG. 14 is a flowchart showing the flow of a conventional bit allocation process.

[Explanation of symbols]

 Reference Signs List 1 audio data compression device 2 CPU 3 input device 4 display device 5 RAM 6 communication unit 7 acoustic codec unit 8 external I / O interface 9 storage device 10 storage medium 11 bus 21 predetermined bit number setting unit 22 bit rate specification unit 23 encoding Subband determination unit 31 Required minimum number of bits setting unit 32 Bit rate specification unit 33 Encoding subband determination unit 100 Filter bank 110 Sound source 111 Disk 112 DAT 113 MIDI 120 Portable audio player 200 Psychoacoustic model unit 300 Bit allocation unit 400 Bit stream forming section A Specified bit rate B Predetermined bit rate C Predetermined bit number Ee Square error energy M Cumulative bit number N Available bit number S Bit summation T Constant value f Input signal

Claims (8)

[Claims] 1. An audio data compression apparatus for dividing audio data of each frame into a plurality of bands, allocating an appropriate number of bits to each of the divided bands, and quantizing the divided data. Coding band limiting means for limiting a band to be coded among a plurality of bands; and a bit allocating means for allocating the number of bits according to the bit rate to the band limited by the coding band limiting means. An audio data compression device, comprising: 2. The apparatus according to claim 1, further comprising: a designating unit for designating the bit rate, wherein the coding band limiting unit is configured to, when the bit rate designated by the designating unit is lower than a predetermined bit rate, set the plurality of bands. 2. The audio data compression apparatus according to claim 1, wherein a band to be encoded is limited. 3. The encoding band limiting means assigns a predetermined number of bits to each band to be encoded, and the sum of the number of bits assigned to each band is the number of bits available according to the bit rate. Discriminating means for discriminating whether or not the sum falls within a range of bits to be encoded; when the sum does not fall within the number of usable bits according to the bit rate, The audio data compression device according to claim 1, further comprising: a reduction unit. 4. The encoding band limiting means assigns a predetermined number of bits to each band to be encoded, and the sum of the number of bits allocated to each band is the number of bits available according to the bit rate. Determination means for determining whether or not the sum falls within the range.If the sum does not fall within the number of usable bits in accordance with the bit rate, the band having the minimum energy value of the signal to be encoded is determined by the determination means. 2. The audio data compression device according to claim 1, further comprising: an encoding target band reducing unit that removes the encoding target band from the encoding target. 5. The coding band limiting means includes: a minimum bit number allocating means for allocating a minimum bit number required for each band to be coded according to a degree of quantization distortion; Discriminating means for discriminating whether or not the sum of the necessary minimum number of bits allocated to the band is within the number of usable bits according to the bit rate; and 2. The audio data compression apparatus according to claim 1, further comprising: an encoding target band reducing unit that removes a highest band from the encoding target when the number of bits does not fall within the usable number of bits. 6. The coding band limiting means includes: a necessary minimum bit number allocating means for allocating a required minimum number of bits to each band to be coded according to a degree of quantization distortion; Discriminating means for discriminating whether or not the sum of the necessary minimum number of bits allocated to the band is within the number of usable bits according to the bit rate; and 2. The audio encoding apparatus according to claim 1, further comprising: encoding target band reducing means for excluding a band having the minimum energy value of the encoding target signal from the encoding target when the number of bits does not fall within the usable number of bits. Data compression device. 7. The method according to claim 5, wherein said minimum necessary bit number allocating means calculates and allocates a minimum bit number required for each band so that the quantization noise does not exceed a predetermined value. Audio data compression device. 8. A storage medium storing a computer-executable program, comprising: dividing audio data of each frame into a plurality of bands;
A computer-executable program code for allocating an appropriate number of bits to each of the divided bands and performing quantization, and for limiting a band to be encoded among the plurality of bands according to a bit rate. A computer-executable program code; and a computer-executable program code for allocating the number of bits according to the bit rate for a limited band. .