JPH04302531A - High-efficiency encoding device for digital data - Google Patents

Abstract

PURPOSE:To obtain an optimum encoding output by selecting one of the outputs of orthogonal transforming means according to the outputs of the orthogonal transforming means. CONSTITUTION:Input digital audio data are divided into plural bands which are wider and wider in band width as the frequency range is higher and higher and blocks consisting of plural samples are formed by the divided bands; and coefficient data are obtained by orthogonal transformation by, for example, fast Fourier transformation, block by block, by the bands and then encoded by the adaptive allocated number of bits. In this case, encoders 2-5 which perform the orthogonal transformation of the input digital data with different block lengths by the blocks are provided and a selecting circuit 6 selects one of the outputs of the respective encoders 2-5 according to the respective outputs of the orthogonal transforming means of the encoders 2-5. Namely, only one of the outputs is sent out through the switching operation of a changeover switch 7 according to the select signal from the selecting circuit 6.

Description

[Detailed description of the invention]

0001

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a high-efficiency digital data encoding apparatus for encoding input digital data by so-called high-efficiency encoding.

0002

2. Description of the Related Art There are various methods for highly efficient encoding of audio or voice signals. For example, band division encoding (which divides an audio signal, etc. on the time axis into multiple frequency bands and encodes them) is available. Sub band coding: SBC
), and so-called transform coding in which a signal on the time axis is converted into a signal on the frequency axis (orthogonal transformation), divided into a plurality of frequency bands, and encoded for each band. Also,
A high-efficiency encoding method that combines the above-mentioned band division coding and transform coding has also been considered. In this case, for example, after performing band division using the band division coding described above, The signal is orthogonally transformed into a signal on the frequency axis, and encoding is performed for each orthogonally transformed band. Here, the above-mentioned orthogonal transformation includes, for example, converting the input audio signal into blocks of predetermined unit time (frames) and performing fast Fourier transform (FFT) on each block to transform the time axis into the frequency axis. There is an orthogonal transformation. Furthermore, as the above-mentioned band division, for example, band division may be performed in consideration of human auditory characteristics. That is, an audio signal may be divided into a plurality of bands (for example, 25 bands) using a bandwidth generally called a critical band, in which the higher the band, the wider the band. Further, when encoding data for each band at this time, encoding is performed by predetermined bit allocation for each band or adaptive bit allocation for each band. For example, when encoding the FFT coefficient data obtained by the above FFT processing using the above bit allocation, the FFT coefficient data for each band obtained by the above FFT processing for each block is adaptively Encoding will be performed with the appropriate number of allocated bits.

0003

[Problems to be Solved by the Invention] In the above-mentioned encoding, the input audio signal is divided into, for example, a plurality of bands, and each band is further subjected to fast Fourier transform (FFT).
When performing orthogonal transformation such as FT) (that is, when performing frequency analysis in each band), the signals for each band are usually divided into blocks in predetermined time units (frame units), and the block unit for each band is I am trying to perform an orthogonal transformation.

[0004] Also, the coefficient data (FFT coefficient data) obtained by this orthogonal transformation is encoded, and the number of bits allocated during this encoding is allocated to each block of the frame unit.

By the way, the input audio signal is not always a steady signal with little fluctuation in level etc.
This level etc. changes variously. For example, there are cases where the signal has a large temporal variation in peak level within the frame (a signal that changes transiently). That is, for example, in the case of an audio signal such as the sound of a percussion instrument,
The signal of this hitting sound portion becomes the signal that changes transiently.

[0006] This audio signal whose signal characteristics (properties) change, for example, stationary or transient, is uniformly orthogonally transformed in blocks of frames of a predetermined time as described above, and this orthogonally transformed data is It cannot be said that encoding is a good encoding that is adapted to the characteristics of the signal, and the sound quality after subsequent decoding cannot necessarily be said to be good in terms of audibility.

The present invention has been proposed in view of the above-mentioned circumstances, and is capable of highly efficient compression encoding that is more suited to the nature (characteristics) of input audio signals, and also enables decoding. It is an object of the present invention to provide a highly efficient encoding device for digital data, which provides an audibly good signal after encoding.

[0008]

[Means for Solving the Problems] A high-efficiency encoding device for digital data of the present invention has been proposed to achieve the above-mentioned object, and blocks input digital data into a plurality of samples. The present invention is a highly efficient digital data encoding device that performs orthogonal transformation for each block to obtain coefficient data, and encodes this coefficient data with an adaptive number of assigned bits. It has a plurality of orthogonal transformation means for orthogonally transforming the input digital data with mutually different block lengths, and selects only one output from each orthogonal transformation means based on each output from the plurality of orthogonal transformation means. This is how it was done. Here, the input digital data supplied to each of the orthogonal transform means can be, for example, data for each band divided by a so-called critical bandwidth. Further, this orthogonal transformation may include, for example, so-called fast Fourier transform (FFT) processing. In this case, the FFT coefficient data is encoded as described above. Furthermore, when making the above selection, for example, the number of allocated bits (the number of bits necessary to achieve a specified sound quality) required when encoding each output from a plurality of orthogonal transform means is , only the output of the orthogonal transform means that minimizes the number of outputs is selected. In other words, the number of bits per frame output from the device after each of the orthogonally transformed coefficient data is encoded is a certain predetermined number of bits. Since adaptive bit allocation is performed during encoding processing, the number of bits actually required for encoding the frame may differ from the fixed number of bits for each frame. Therefore, for example, if the number of bits actually required for the encoding is less than the predetermined number of bits for each frame, better encoding can be performed using the remaining surplus bits. Therefore, the best encoded output can be obtained by selecting the one that requires the least number of required bits among the outputs of each orthogonal transform means. Furthermore, for example, in each encoding process, even if the actually required number of bits is greater than the predetermined number of bits per frame, the most necessary bits among the outputs of each orthogonal transform means may be By selecting one with a small number, it becomes possible to obtain an encoded output with the least deterioration due to encoding.

[0009]

[Operation] According to the present invention, input digital data is orthogonally transformed with a plurality of block lengths, and only one output from the orthogonal transform means is selected based on the output of each orthogonal transform means. When making this selection, for example, by selecting the one with the least number of bits required for encoding, it becomes possible to obtain the optimum encoded output.

0010

Embodiments Hereinafter, embodiments to which the present invention is applied will be described with reference to the drawings. As shown in FIG. 1, the high-efficiency encoding device for digital data of this embodiment divides input digital data, such as audio, into a plurality of bands such that the higher the frequency band, the wider the bandwidth. For each band, a block consisting of a plurality of samples is formed, and each block of each band is subjected to orthogonal transformation using, for example, fast Fourier transform (FFT) to obtain coefficient data (FFT coefficient data), and this coefficient data is adaptively allocated bits. It is encoded by numbers. In the high-efficiency encoding device of this embodiment, there are a plurality of encoders (for example, four encoders 2 to 5) each having orthogonal transform means for orthogonally transforming the input digital data with different block lengths for each band. Based on each output from each orthogonal transformation means of each of these encoders 2 to 5, only one output from each of the encoders 2 to 5 (i.e., one output from each orthogonal transformation means) is provided. I try to choose. In other words, the FF for each band in the encoders 2 to 5 at this time
The block length of the block to be processed is determined by each encoder 2~
Based on the outputs from these encoders 2-5, the selection circuit 4 selects only one output from the outputs of the encoders 2-5. That is, by performing the switching operation of the changeover switch 7 based on the selection signal from the selection circuit 6, only one of the outputs of the encoders 2 to 5 is outputted.

[0011] Here, during the selection process in the selection circuit 6, for example, a process of selecting only the output of the encoder that requires the least number of bits in each encoding process in each of the encoders 2 to 5 is performed. I try to do it. That is, the number of bits in each frame output from the encoders 2 to 5 is a predetermined certain number of bits, but during each of these encoding processes, masking effects etc., which will be described later, are applied. By performing adaptive bit allocation in consideration of the above, the number of bits actually required for encoding the frame can be determined. In this selection, only the output of the encoder that requires the least number of actually required bits is selected. Therefore, if the number of bits required for this encoding is smaller than the predetermined number of bits for each frame, the remaining surplus bits can be used to perform better encoding, and each The best encoded output can be obtained by selecting the encoder output that requires the least number of bits. Also, for example, in each encoder 2 to 5, even if the actually required number of bits is greater than the predetermined number of bits per frame, the most necessary number of bits among the outputs of each encoder 2 to 5 may be By selecting one with a small number of bits, it is possible to obtain an encoded output with the least deterioration due to encoding.

That is, each of the encoders 2 to 5 of the high-efficiency encoding device for digital data of this embodiment receives an input of audio or voice supplied via the input terminal 1, for example, as shown in FIG. Digital data is processed by QMF (quadrature), a so-called mirror filter.
With mirror filters 41 and 42, the signal is divided (for example, roughly divided into three) so that the higher the frequency range, the wider the bandwidth, taking into consideration the division in the so-called critical band. Fast Fourier transform (FF
T) In the circuits 43, 44, and 45, a block consisting of a plurality of samples is formed for each divided band, and orthogonal transform (converting the time axis to the frequency axis) by fast Fourier transform is performed for each block. Therefore, the coefficient data (FFT
coefficient data). The output of the fast Fourier transform circuit 43 corresponds to, for example, two bands in the high range of the critical band, and the output of the fast Fourier transform circuit 44 corresponds to, for example, three bands in the middle range of the critical band. The output of the conversion circuit 45 is, for example, 2 in the low range of the critical band.
It is made to correspond to 0 bands. Furthermore, the FFT coefficient data of each band from these fast Fourier transform circuits 43, 44, and 45 is encoded by encoding circuits 46, 47, and 48 with an adaptive number of assigned bits. That is, the encoding circuits 46, 47, 4 of this embodiment
When encoding the FFT coefficient data of the three bands mentioned above in No. 8, the encoding is performed using an adaptive number of assigned bits based on human auditory characteristics.

In order to perform the band division as described above, the input terminal (input terminal of each encoder 2 to 5) 1 in FIG.
Digital data (0 to 22.1k) obtained from sample)
Hz) is supplied, and the digital data is
The MF41 and 42 roughly divide the frequency range into three bands (0 to 5.5kHz, 5.5kHz) so that the higher the frequency range, the wider the bandwidth.
Hz~11.0kHz, 11.0kHz~22.1kHz
z). In the above QMF41, the above 0 to 22
．． 1kHz digital data is divided into two and becomes 11.0
Two outputs of kHz to 22.1 kHz and 0 to 11.0 kHz are obtained, the output of 11.0 kHz to 22.1 kHz is sent to the fast Fourier transform circuit 43, and the output of 0 to 11.0 kHz is sent to the QMF 42. 0 sent to QMF42
The output of ~11.0kHz is further divided into two by the QMF42 to 5.5kHz~11.0kHz and 0~5.5kHz.
Two outputs of z are obtained. Above 5.5kHz~11.
The 0 kHz output is sent to the fast Fourier transform circuit 44, and the 0 to 5.5 kHz output is sent to the fast Fourier transform circuit 45.

[0014] Each fast Fourier transform circuit 43, 44, 45
Here, one frame B is constructed from a plurality of samples (for example, 1024 samples) of the supplied data of each band, and the blocks of each frame B are subjected to Fourier transform processing to obtain FFT coefficient data. However, at this time, F in each fast Fourier transform circuit 43, 44, 45
As mentioned above, the block length of the FT processing is different for each of the encoders 2 to 5.

For example, in the case of the encoder 2 described above, as shown in FIG. 3, the block lengths of the FFT processing for each band are, for example, the same length. That is, in this encoder 2, the FFT processing block length bH in the fast Fourier transform circuit 43 corresponding to the high frequency range of 11.0 kHz to 22.1 kHz, and the above 5.5 kHz to 11.0 kHz.
The block length bM in the fast Fourier transform circuit 44 corresponding to the middle range of z and the block length bL of the FFT processing in the fast Fourier transform circuit 45 corresponding to the low range from 0 to 5.5 kHz are determined by the above predetermined unit time. The block length is the same as that of frame B.

In the case of the encoder 3, as shown in FIG. 4, the block length subjected to FFT processing in each band is shortened in the high frequency band. That is, in this encoder 3,
The block length bL in the fast Fourier transform circuit 45 in the low range and the fast Fourier transform circuit 24 in the middle range
For example, the block length in the fast Fourier transform circuit 43 in the high frequency range is the block length bM in the low frequency range (
1/2 of the block length bL (or bM ) of
The block length is set to . In the illustrated example, the high frequency block is divided into block lengths bH1 and bH2.

In the case of the encoder 4, as shown in FIG. 5, the block lengths subjected to FFT processing in each band are shortened in the middle and high bands. That is, each F of this encoder 4
In the block length of FT processing, the block length b is used for the low frequency range.
L, the middle range has block lengths bM1, bM2 that are 1/2 of the low range, and the high range has block lengths bH1, bH2, bH3, and 1/4 of the low range (1/2 of the middle range), for example. b
It is assumed to be H4.

Furthermore, in the case of the encoder 5, as shown in FIG. 6, the block length is short in the high and middle ranges and long in the low range. That is, in the block length of each FFT process of the encoder 5, if the low range is the block length bL, the high and middle ranges are, for example, block lengths bH1, bH2, bH3, bH4, and bM1, which are 1/4 of the low range.
bM2, bM3, bM4.

Here, as mentioned above, the reason why the block length of the high range (and middle range) is made shorter than the block length of the low range and the block length of the low range is made longer in FIGS. 4 to 6 is as shown below. Due to reasons. In other words, the frequency analysis ability (frequency resolution) of human hearing is generally not very high in the high range, but high in the low range.Therefore, it is necessary to ensure frequency resolution in the low range. As mentioned above, the block length of FFT processing cannot be made very short in this low frequency range. For this reason, the block length is increased in the low range. In addition, in general, the steady interval is long for low frequency signals, and conversely short for high frequency signals, so the block length in the high frequency range (and midrange) is shortened (improving time resolution).
This is valid. In view of the above, in this embodiment, the block length of the high range (and middle range) is shortened, and the block length of the low range is lengthened.

In this way, this embodiment has a configuration that simultaneously satisfies the resolution on the frequency axis and the resolution on the time axis required for hearing, and the above-mentioned low frequency (
0 to 5.5 kHz), the number of processing samples is increased to increase the frequency resolution, and the high frequency range (11.0 kHz to 22.1 kHz) is
Hz), the time resolution is increased. In some cases, the temporal resolution is also increased in the middle range (5.5 kHz to 11.0 kHz).

[0021] Also, the characteristics (properties) of the input audio signal
Therefore, as described above, it is effective to shorten the FFT processing block length in the high frequency range (and the middle frequency range). That is, it is effective to change the block length of the FFT processing, depending on whether the input audio signal is, for example, a transient signal or a stationary signal. For example, in the case of a stationary signal, it is effective to set the block length of each band to the same length as shown in FIG. As shown,
It is effective to shorten the FFT processing block lengths in the high and middle ranges. In this way, by shortening the FFT processing block length in the case of a transient signal, more bits are allocated to the block (transient signal part) with a large peak level in frame B during encoding. You will be able to do this. On the other hand, reduce the number of bits for other blocks. This makes it possible to follow temporal changes in the spectrum and provide bits only to blocks that truly require bits in each band of frame B. Furthermore, for example, in the case of a stationary signal, there is no need to redundantly encode signals with similar spectra in each block within frame B.

[0022] The FFT processing block lengths for each of the above bands are not limited to the examples shown in Figs. 3 to 6 above. Various block length patterns can be considered.

The encoders 2 to 5 perform FFT processing using different block lengths for each band as described above in the fast Fourier transform circuits 43, 44, and 45 of FIG. Encoding circuits 46, 47
, 48.

In each of the encoders 2 to 5 of this embodiment, the following configuration is used to perform encoding by adaptive bit allocation in each of the encoding circuits 46, 47, and 48. I have to.

That is, each encoder 2 to 5 of this embodiment
The number of bits actually required to encode the FFT coefficient data in the frame B from the fast Fourier transform circuits 43, 44, and 45 (the number of bits required to achieve the specified sound quality) ), and this primary bit allocation number determining circuit 60.
A bit number correction circuit 61 performs bit allocation or bit reduction in order to match the primary bit number determined in frame B with the final bit number predetermined for the frame B. Therefore, the encoding circuits 46, 47,
Each encoding of the FFT coefficient data in 48 is performed using the number of bits (ie, the final number of bits) obtained by correcting the number of primary bits in the number of bits correction circuit 61.

Here, the primary bit allocation number determining circuit 6
The number of primary bits determined for 0 is determined, for example, in consideration of a so-called masking effect as described later.

Note that the above-mentioned masking relates to human hearing characteristics. That is, in general, human auditory characteristics with respect to sound include something called a masking effect, and the masking effects include a temporal masking effect, a simultaneous masking effect, and the like. The above-mentioned temporal masking effect is an effect in which a small sound (or noise) that occurs at the same time as a loud sound is masked by the loud sound and becomes inaudible. The effect is such that the small sounds (noise) before and after the sound are masked by the loud sound and become inaudible. In this temporal masking effect, masking temporally behind the loud sound is called forward masking, and masking temporally forward is called backward masking. In addition, in temporal masking, due to the characteristics of human hearing, the effect of forward masking is effective for a long period of time (for example, about 100 msec), whereas the effect of backward masking is effective for a short period of time (for example, about 100 msec).
For example, about 5 msec). Furthermore, the level of the masking effect (masking amount) is about 20 dB for forward masking and about 30 dB for backward masking.

Therefore, if this masking effect is taken into consideration when allocating bits within the frame B, more optimal bit allocation becomes possible. In other words, even if the number of bits is reduced for the signal in the masked part, there is no adverse effect on the auditory sense, so the number of bits in the masked part can be reduced, and effective encoding can be achieved with a small number of bits. becomes possible. Note that the amount of masking in the masking effect is obtained, for example, by calculating the sum of the energies for each of the critical bands, and based on the energy for each of the critical bands. It is also possible to obtain the amount of masking of another critical band (or the certain critical band itself) at other times by a signal in a certain critical band. The allowable noise level for each band is determined based on this masking amount, and the number of allocated bits for the encoding can be determined based on the allowable noise level for each band error.

The primary bit number determined by the primary bit allocation number determination circuit 60 as described above is sent to the bit number correction circuit 61. In the bit number correction circuit 61,
Bit allocation or bit reduction is performed to match the primary bit number determined by the primary bit allocation number determining circuit 60 to the final bit number predetermined for the frame B.

Here, the primary bit allocation number determining circuit 6
FIG. 7 shows a flowchart of the determination of the primary bit number and the bit allocation or bit reduction process performed by the bit number correction circuit 61.

That is, in this flowchart,
In step S1, the primary bit allocation number determining circuit 60
The number of primary bit allocations determined in , that is, the encoding circuits 46 and 4 determined based on the masking calculation, etc.
The above-mentioned number of primary bits (number of used bits) actually required for encoding with 7.48 is substituted into the variable nsum0. In step S2, this variable nsum0 is sent to the bit number correction circuit 61, and the bit number correction circuit 61 further substitutes this variable nsum0 into the variable nsum.

In step S3, it is determined whether this variable nsum is smaller than (less than) a number nlimit indicating the final number of bits predetermined in the frame B, or whether it is greater than or equal to this number nlimit. variable nsu
If m is smaller than the above number nlimit, step S
The process proceeds to step S4, and when the variable nsum is equal to or greater than the above number nlimit, the process proceeds to step S5.

Here, if the variable nsum is less than the number nlimit indicating the final number of bits, there will be a surplus of bits. Therefore, in step S4 above,
This remaining number of bits (the number of bits corresponding to the difference between the variable nsum and the number nlimit indicating the final number of bits) is further distributed within the frame B. When allocating bits, the bits are allocated to each band or block so as to improve the sound quality. Thereafter, the number of bits for which this bit allocation has been made is returned to step S3.

In step S5, the variable nsum
It is determined whether the number nlimit is larger than the number nlimit indicating the final number of bits, or is less than or equal to the number nlimit. When the variable nsum is less than or equal to the number nlimit indicating the final number of bits, the variable nsum becomes equal to the number nlimit in relation to step S3, and the process ends. Further, when the variable nsum is larger than the number nlimit indicating the final number of bits, the process advances to step S6.

Here, if the variable nsum is greater than the number nlimit indicating the final number of bits, the number of bits is insufficient. For this reason, the above step S6
Now, a process is performed to reduce this missing bit from the number of bits according to the variable nsum. When reducing bits, the bits are reduced starting from the band or block that has the least effect on sound quality deterioration. Thereafter, the number of bits resulting from this bit reduction is returned to step S5.

According to the above flowchart, the number of bits is corrected, and the encoding circuits 46, 47, and 48 perform encoding using the corrected number of bits.

The encoded data of each of the aforementioned encoding circuits 46, 47, and 48 is sent to a combining circuit 50. In addition, above 1
The information on the number of primary bits determined by the next bit allocation number determination circuit 60 (bit number information indicating the number of bits actually required to achieve a predetermined sound quality) is also transmitted to the synthesis circuit 50.
It is now sent to The data of each band is synthesized in the synthesis circuit 50, the encoded data of the synthesized data is outputted from an output terminal 52, and the primary bit number information is outputted from an output terminal 53. .

The outputs from the output terminals 52 and 53 in FIG. 2 are output from each encoder 2 to 5 in FIG. 1, respectively. The encoded data from each encoder 2 to 5 is sent to the changeover switch 7 in FIG. 1, and the primary bit number information is sent to the selection circuit 6.

Here, as described above, the selection circuit 6 selects a code from among the outputs of the encoders 2 to 5 based on the primary bit number information sent from each of the encoders 2 to 5. A selection process is performed to select only the output of the encoder that minimizes the number of bits required for encoding (the number of bits required to achieve a specified sound quality), and this selection signal is sent to the changeover switch 7. It will be done. As a result, the changeover switch 7 performs a switching process to output only one encoded output based on the selection signal from among the supplied outputs of the encoders 2 to 5. Therefore, only the selected encoded output is output from the output terminal 8 of the device of this embodiment.

The above encoded output of the apparatus of this embodiment is
By performing decoding processing using a decoding device (not shown) based on the next bit allocation information, the sound quality of the obtained audio becomes optimal.

As described above, in the high-efficiency encoding device for digital data of this embodiment, only the output of the encoder that has been optimally encoded is selected from among the outputs of the plurality of encoders 2 to 5, and encoded. Since the encoded data is obtained as encoded data, the best sound quality can be obtained by decoding this encoded data and converting it to voice. Also,
In the device of this embodiment, since the output after actual encoding is selected, the block length of the block in the above FFT processing is selected to be the optimal block length.
Moreover, selection of this block length is also easy. Furthermore, the device of this embodiment can handle so-called compact discs (
The present invention can be applied to an encoding device for data recorded on package media such as CDs. In this case, the general user only needs a decoder (player) and does not need an encoding device, so the scale of the encoding device is not a problem and the device of this embodiment is particularly effective.

Furthermore, in the above-described embodiment, the encoders 2 to 5 are each provided with fast Fourier transform circuits 43 to 45 having different FFT processing block lengths. 1
In this encoder, a fast Fourier transform circuit that performs FFT processing with different FFT processing block lengths as described above is arranged for each band, and from among the outputs of the fast Fourier transform circuit for each band, It is also possible to adopt a configuration in which only the output of the fast Fourier transform circuit that minimizes the number of bits actually required for encoding as described above is selected.

That is, in the case of the high-efficiency encoding device of this example, only one encoder is provided instead of a plurality of encoders as shown in FIG. The internal configuration of the encoder in this example will be explained by comparing it with the configuration in Figure 2.
As shown in FIG. 2, there is one fast Fourier transform circuit corresponding to each band (each Fourier transform circuit 43, 44, 45
), a plurality of fast Fourier transform circuits corresponding to the number of types of block lengths in each band portion in FIGS. 3 to 6 are arranged. For example, three fast Fourier transform circuits with different block lengths are arranged in the high band, three fast Fourier transform circuits with different block lengths are arranged in the middle band, and one fast Fourier transform circuit is arranged in the low band. In other words, in the high frequency range, the block length bH in FIG. 3, the block lengths bH1, bH2 in FIG. 4, and the block length bH1 in FIGS. 5 and 6.
, bH2, bH3, and bH4. Three fast Fourier transform circuits are arranged to perform FFT processing on high-frequency data, respectively, with three types of block lengths corresponding to block lengths bH2, bH3, and bH4. The block lengths bM1, bM2 and Figure 6
The mid-range data is subjected to FFT with three types of block lengths corresponding to the block lengths bM1, bM2, bM3, and bM4.
Three fast Fourier transform circuits are arranged to process the data, and in the low range, a fast Fourier transform circuit is arranged to perform FFT processing on the low range data with one type of block length corresponding to the block length bL in FIGS. 3 to 6. . The outputs of these fast Fourier transform circuits are sent to the selection circuit 6, and among the outputs of these fast Fourier transform circuits, only the output of one fast Fourier transform circuit for each band is selected and converted. If this is done, the same processing as in FIG. 1 described above becomes possible. In this example as well, it is possible to obtain the same effects as described above, and it is also possible to simplify the configuration.

Note that the orthogonal transformation described above is not limited to the above-mentioned fast Fourier transformation, but may also be applied to, for example, discrete cosine transformation.

[0045]

Effects of the Invention In the high-efficiency encoding device for digital data of the present invention, only one output from each orthogonal transform means is selected based on the outputs of a plurality of orthogonal transform means, For example, by selecting only the output of the orthogonal transform means that minimizes the number of allocated bits during encoding, the optimal encoded output can be achieved according to the characteristics of the input digital data and the human auditory characteristics (in the case of audio data). Therefore, if this encoded output is decoded and converted into speech, the best sound quality can be obtained. Furthermore, the device of the present invention is particularly effective when applied to an encoding device for data recorded on package media such as so-called compact discs (CDs).

[Brief explanation of drawings]

FIG. 1 is a block circuit diagram showing a schematic configuration of a highly efficient digital data encoding device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a schematic configuration of a specific example of an encoder.

FIG. 3 is a diagram showing FFT processing blocks when the block length is the same in each band.

FIG. 4 is a diagram showing an FFT processing block when the high-frequency block length is 1/2 of a frame.

FIG. 5 is a diagram showing FFT processing blocks when the block length of the high band is 1/4 of the frame and the block length of the middle band is 1/2 of the frame.

FIG. 6 is a diagram showing an FFT processing block when the block length of the high band is 1/4 of the frame and the block length of the middle band is 1/4 of the frame.

FIG. 7 is a flowchart for explaining primary bit allocation number determination and bit number correction.

[Explanation of symbols]

2 to 5... Encoder 6... Selection circuit 7... Changeover switch 41, 42... QMF 43, 44, 45... Fast Fourier Conversion circuit 46, 4
7, 48... encoding circuit

Claims (1)

[Claims] Claim 1: A digital data processing method that blocks input digital data with a plurality of samples, performs orthogonal transformation for each block to obtain coefficient data, and encodes this coefficient data with an adaptive number of allocated bits. The efficiency encoding device has a plurality of orthogonal transform means for orthogonally transforming the input digital data with mutually different block lengths, and based on each output from the plurality of orthogonal transform means, one of the outputs of each orthogonal transform means is A high-efficiency encoding device for digital data, characterized in that only one output is selected.