JP2992994B2 - How to encode digital audio data - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The description will be made in the following order.

A Industrial Field of Use B Summary of the Invention C Prior Art D Problems to be Solved by the Invention E Means for Solving the Problems (FIG. 1) F Function G Embodiment G ₁ First Embodiment (1st Embodiment) (FIGS. To 8) H Effect of the Invention A Industrial Field of the Invention The present invention relates to a method for encoding digital audio data.

B. Summary of the Invention In the present invention, when digital audio data is transmitted after compressing the data, more excellent data transmission can be performed by exchanging bits between two blocks forming a pair. .

C Conventional technology For example, in the case of 8 mm video, as an optional function, an audio signal is digitized into a PCM signal during recording, and this PCM signal is recorded in an overscan section of a tape, and the reverse processing is performed during reproduction. To obtain the original audio signal.

In this case, if the sampling frequency and the number of quantization bits of the PCM signal are increased, the audio signal can be recorded and reproduced with more excellent characteristics.However, the number of bits to be recorded and reproduced increases, and the recording and reproduction cannot be performed. Would.

Therefore, during recording, the number of bits of the PCM signal is compressed,
At the time of reproduction, it has been considered to expand the number of bits so that excellent recording and reproduction characteristics can be obtained even if the number of bits on the tape is small.

And as such a method of bit compression / expansion
There is a method called ADPCM.

FIG. 9 shows an example of a transmission system based on the ADPCM. In this example, every 64 consecutive samples of input data, the 64 samples are regarded as one block, and the prediction coefficient of the prediction filter is optimized for each block. This is the case when controlling to a value. At this time, the main data which is bit-compressed is output for each sample of the input data, and the auxiliary data relating to the bit compression is output for each block.

That is, in the figure, (10) is an encoder, (3)
0) indicates a signal transmission system, and (40) indicates a decoder. For example,
When applied to the PCM audio system for 8 mm video, the encoder (10) is provided in the recording system, the decoder (40) is provided in the reproduction system, and the transmission system (30) is used for error correction processing. It includes a circuit, a circuit magnetic head, and the like.

Then, in the encoder (10), the digital data
Xt is supplied from the input terminal (11) to the subtraction circuit (14) through the delay circuits (12) and (13) in parallel for each sample. In this case, the input data Xt is a PCM signal obtained by linearly A / D converting an analog audio signal.
The sampling frequency is 48 kHz and the number of quantization bits is 16 bits. Further, as shown in FIG. 10, the data Xt is -1.
It is assumed that the data is represented by a fixed point of Xt <1 and also represented by a two's complement (the same applies to other data).

Further, the delay circuits (12) and (13) are for adjusting the timing of the main data and the auxiliary data.
Each has a delay time of one block period. (Strictly speaking, if the input value of the terminal (11) is Xt, the output of the delay circuit (13) will be Xt _-128 . ,
Simply Xt).

Further, a prediction value t for the data Xt is taken out from the prediction filter (19), and this value t is supplied to the subtraction circuit (14), and the difference Dt Dt = Xt between the value Xt and t is output from the subtraction circuit (14). -T is retrieved. This value Dt is the error (prediction residual) of the predicted value with respect to the input value Xt. Therefore, the value Dt is ideally Dt = 0, and is generally a small value.
Even if the word length of the value Dt is, for example, 16 bits, for example,
As shown in FIG. 10, when Dt0, a considerable bit on the MSB side is all "0", and when Dt <0, it is all "1" and the remaining few bits on the LSB side have the value It becomes "0" or "1" according to the difference between Xt and t. When the value Dt becomes a large value, the lower bits can be ignored.

Then, this value Dt is supplied to a gain control circuit (15) and is multiplied by G (G1) to obtain a normalized value Dt ·
G, and this value G · Dt is supplied to the requantization circuit (16), and is requantized into, for example, a 4-bit value Et = · G.

Further, this value Et is supplied to the gain control circuit (17).
It is multiplied by 1 / G, so that it is an unnormalized value t in the same order as the value Dt, and this value t is supplied to the addition circuit (18), and the predicted value t from the filter (19) is Addition circuit (18) supplied to the addition circuit (18)
, The sum tt = t + t of the values t and t is extracted, and this value t is supplied to the filter (19).

In this case, the value t is a predicted value for the value Xt, and the value t is a value obtained by truncating or rounding the lower bits of the error Dt at the time of the prediction.
T, which is the sum of t and t, is approximately equal to the input value Xt.
Then, since this value t is supplied to the filter (19), the value t, which is the filter output, can be a value that predicts the input value _{Xt + 1} at the next sample time.

Then, the value Et from the requantization circuit (16) is supplied to the decoder (40) through the transmission system (30).

In the decoder (40), the value Et is multiplied by 1 / G by a gain control circuit (41) to obtain a value t, and the value t is supplied to an addition circuit (42), and the added output is output to an output terminal (44). ), Is supplied to a prediction filter (43) configured similarly to the filter (19), and the filter output is supplied to an addition circuit (42).

Therefore, the output of the filter (43) becomes the value t, and digital data substantially equal to the input data Xt is extracted from the terminal (44).

Further, the following circuit is provided to make the prediction coefficients in the filters (19) and (43) optimal values for each block.

That is, the prediction filters (19) and (43) are tertiary filters that use, for example, partial autocorrelation coefficients (PARCOR coefficients) as prediction coefficients, and the first to third coefficients a _1. ~a ₃ is to be changed to any value.

Further, the input data Xt from the terminal (11) is supplied to a time window circuit (21), and after a predetermined weighting, supplied to an autocorrelation circuit (22) to calculate a correlation coefficient. coefficient partial autocorrelation coefficients k ₁ to k ₃ is calculated as the third prediction coefficient to order for each block of the supplied the data Xt in prediction coefficient circuit (23), the coefficient k ₁ to k ₃ 1 The timing is adjusted by the delay circuit (27) during the block period, and then supplied to the filter (19) and supplied to the filter (43) through the latch (51).

Further, the data Xt from the delay circuit (12) is supplied to the prediction error filter (24), and the output of the filter is supplied to the intra-block maximum value detection circuit (25).

In this case, the filter (24) includes a tertiary prediction filter (241) configured in the same manner as the prediction filter (19), and a subtraction circuit (242), and the prediction coefficient k from the coefficient circuit (23). _{1 to} k ₃ are supplied to the filter (241), and the predicted value (prediction error) t of the error Dt with respect to the input data Xt is
It is generated for each sample. The detection circuit (25) detects, for each block of the input data Xt, the absolute value max of the prediction error having the maximum absolute value among the prediction errors t (there are 64) in the block. Things.

The maximum value max is calculated by the normalized gain calculation circuit (2
The gain GG = b / max b at the time of normalization supplied to 6) is a safety coefficient of 0 <b <1, for example, converted to b = 0.9, and the gain coefficient G for normalization is obtained by the gain control. The signal is supplied to the circuits (15) and (17), and is supplied to the gain control circuit (41) through the latch (52). In this case, the value
Since max is the maximum value of 64 values t, the value Et
Is normalized to −1Et <1.

Incidentally, the latch (51), (52), the coefficient k ₁ to k _3, G,
This is for holding the corresponding value t · G for one block period.

Considering the amount of data transmitted from the encoder (10) to the decoder (40) through the transmission system (30),
The prediction residual Et, which is the main data, is transmitted, for example, every 4 bits at each sample, and the prediction coefficients k _{1 to} k ₃ and the gain coefficient G, which are auxiliary data (prediction parameters), are, for example, 16 bits, 12 bits, 12 bits. Since data is transmitted for each block in bits and 8 bits, the data amount in one block period is as follows: 4 bits × 64 samples + 16 bits + 12 bits + 12 bits + 8 bits = 304 bits. Then, 1 when no data compression is performed
The data amount during the block period is 16 bits x 64 samples = 1024 bits. Therefore, the data amount was compressed and transmitted to 304 bits / 1024 bits9.729.7%.

When the transmission system (30) is an 8 mm video PCM audio system, the data amount in one field period is calculated, and the data amount of the prediction residual Et becomes 48000 samples / field frequency = 48000 / channel per channel. 59.94 ‥‥ ≒ 800.8 samples, but it cannot be rounded down, so 801 samples = 801 × 4 bits. Therefore, for the two left and right stereo channels, 801 × 4 bits × 2 channels = 801 words (1 word = 8 bits).

Also, the data amounts of the prediction coefficients k _{1 to} k ₃ and the gain coefficient G are
Since one set is obtained for every 64 samples (one block),
For one channel, 801 samples / 64 samples6412.51 sets, but it cannot be rounded down, so there are 13 sets, and there are 26 sets for the left and right channels. Then, when this is converted into the number of words, (16 + 12 + 12 + 8 bits) × 26 sets = 6 words × 26 sets = 156 words.

Therefore, the total data amount in one field period is 801 words + 156 words = 957 words.

On the other hand, when calculating the amount of PCM audio data in the current 8-mm video, the sampling frequency is
Since 2fh ≒ 31.468 kHz (fh is the horizontal frequency) and the number of quantization bits is 8, the data amount in one field period is 2 channels of signals L and R, 2fh / field frequency × 2 channels = 525 samples × 2 channels = 1050 words.

Therefore, since the data amount of the above-mentioned ADPCM audio is smaller than the data amount of the PCM audio in the current 8 mm video, each data by the above ADPCM is regarded as digital audio data in the current 8 mm video and recorded. Encoding processing and decoding processing for reproduction can be performed. In addition, audio signals can be recorded and reproduced with better sound quality without changing the current tape format.

Further, according to the ADPCM system described above, even if the coefficients and the word length of the operation are limited, the prediction filters (19), (4)
Since the prediction coefficient of 3) is controlled to an optimum value in accordance with the input data Xt, errors caused by compression of the decoded data t can be minimized.

When transmitting the prediction residual Dt, the residual Dt is requantized to reduce the number of bits, and normalization is performed before the requantization.
Is data having a small number of bits and a small error.

Literature: "Basic of speech information processing" Published by Ohmsha Patent and Japanese Patent Application No. 61-299285 D. Problems to be Solved by the Invention By the way, in the above ADPCM system, as shown in FIG. In any block period Ti, the number of bits of the transmitted prediction residual Et is constant at 4 bits.

In this case, if the level (amplitude) change of the audio signal At that is A / D-converted to the digital data Xt is small in any sample, the residual Et is small, and therefore the residual Et is small.
There is no problem expressing Et in 4 bits.

However, if the level of the audio signal At suddenly changes greatly during a certain block period Tm as shown in FIG. B, for example, the prediction of the level during this period Tm is incorrect, and the residual Et is large. And this is 4
When expressed in bits, the accuracy is low and the decoder (40)
In this case, the error of the decoded data t increases.

Therefore, it is preferable that the residual Et is represented by as many bits as possible. However, as can be seen from the above numerical examples, the number of bits of data Et and the number of bits per block are determined based on the transmission capacity of the transmission system (30). Since the number of samples is determined, it is impossible to increase the number of bits of the residual Et. Therefore, when the level change of the audio signal At is abrupt, an error occurs in the data t, and distortion and noise occur in the reproduced audio signal At.

The present invention is intended to solve such a problem.

E Means to solve the problem Now, considering the human hearing, this hearing has a masking effect, and when the sound level changes suddenly, distortion and noise will occur within about ± 10 ms from the time of the change. Even if there is, there is a characteristic that this cannot be sensed much.

The present invention pays attention to such a point, and makes two blocks having a predetermined interval into one pair, and interchanges the number of bits required for the residual Et between the two blocks forming the pair. In addition to the transmission, the data indicating the bit accommodation state is also transmitted. The interval between the two blocks forming a pair is within a period in which the masking effect is effective.

F action The average number of bits for the residual Et does not increase or decrease, but the number of bits increases for the residual Et requiring accuracy.

G Example G ₁ a first embodiment in this example, as shown in FIG. 8, the period Ti and Ti _-5 · · · period Ti ₊₁ and Ti ₊₆ · · · period Ti ₊₂ and Ti _{- 3} ... Period Ti _{+ 3} and Ti _{+ 8} ... Period Ti _{+ 4} and Ti- ₁ ... Period Ti _{+ 5} and Ti _{+ 10} ... Period Ti _{+ 6} and Ti _{+ 1} ... .. Are paired, that is, block periods separated by 5 block periods are paired, and for each block period, the block period of the partner to be paired is defined as the previous block period and the subsequent block period. Switch to.

Also, as shown in FIG. 5C, during the period Tm, the level of the audio signal At changes drastically and greatly, so that the number of bits of each of the residuals Et in the period Tm is increased by 2 bits to 6 bits.
In addition to the bits, the number of bits of the residual Et in the paired period Tm- ₅ is reduced by 2 bits to 2 bits. When the level change of the signal At is not so large as in the block period Tn, the number of bits of the residual Et at that time Tn is all increased by one bit to 5 bits, and the paired block period Tn Each bit number of the residual Et of ₊₅ is reduced by one bit to three bits.

Then, the j-th (j = 1 to 64) residual Et (6) of the period Tm
), And the j-th residual Et (2 bits) of the period Tm- ₅ are combined into 8-bit data WE, so that the corresponding residuals Et, Et of the two block periods forming a pair are combined. , 8
The data is reconstructed into bit data WE and transmitted. Further, at this time, for example, the number pair and the number pair have the same target block period, and
Since WE is the same for the turn and the turn, the data WE for the turn
Is not sent. In other words, the paired block period is 5
If the blocks are separated by a block period (generally, an odd block period), the data WE is the same between every other block period and every other block period. , But the data WE,
,..., Are transmitted to the transmission system (30) every other block period (period indicated by a circle in FIG. 8).

When the number of bits is changed from the standard 4 bits as in the periods Tm and Tn, this is performed based on the level change of the audio signal At. For this purpose, the gain coefficient G is used. That is, since the count G indicates the level of the signal At in units of blocks, the gain coefficient Gi of the period Ti and the period Ti
Determine the specific Ri Ri = Gi / Gi _-5 with Gi _-5 gain factor of _-5, the number of bits of the residual Et accordance with the ratio Ri, determined, for example, as follows.

i.Ri ≧ 11.3 or 0 ≦ Ri ≦ 1 / 11.3, period Ti is 6 bits, period Ti− ₅ is 2 bits (period Tm, Tm− ₅ ) or vice versa.

ii.11.3> Ri ≧ 2.8 or 1 / 11.3 <Ri ≦ 1 / 2.8 The period Ti is 5 bits and the period Ti- ₅ is 3 bits (period Tn, Tn
_-5 applies) or vice versa.

iii. Other than i and ii above, all 4 bits And when the number of bits of the residual Et is changed in this way, data indicating this number of bits also needs to be transmitted to the decoder (40). For example,

That is, the data indicating the number of bits of the residual Et
Assuming BR, this is 4 bits in size. And
As shown in FIG. 6, the lower 3 bits of the data NMBR indicate the number of bits to be increased / decreased by a binary value. When this is "1", the MSB indicates two block periods in which the data NMBR itself forms a pair.
It indicates that the data is the data of the bit number of the previous period Ti- ₅ of Ti and Ti- ₅ , and “0” indicates the data of the bit number of the subsequent period Ti.

Therefore, in the period Tm, the data NMBR is data of the period Tm after the pair, and the MSB is “0”,
Since the residual Et is 6 bits, the lower 3 bits are “010”, and NMBR = “0010” as a whole.

In the period Tn, the data NMBR is the data of the period Tn before the pair, so that the MSB becomes “1” and the residual Et
Is 3 bits, the lower 3 bits are “111”, and NMBR = “1111” as a whole.

Note that the data NMBR of the period Ti to be paired and the data NMBR of the period Ti- ₅ have a two's complement relationship.

Further, in the current 8-mm video, all the processing is performed in units of one word. Therefore, as shown in FIG. 7, two data NMBR, NMBR are combined to form data WNMB of one word length. However, in this case, the data NMBR of the upper 4 bits of the data WNMB and the data NMBR of the lower 4 bits are
BR refers to each data NMBR, N
MBR.

FIG. 1 shows an encoder (10) that performs encoding according to the above rules, and in order to facilitate understanding, in the following description and FIG.
The explanation focuses on Ti. In addition, each data in FIG. 1 is shown in the number () of the block period at the time of the number processing, and for the circuits (61) and (63), the function of the internal data is shown. Indicates the number of the block period to which the data belongs.

That is, in the delay circuit (13), the processing is started from the delay of six block periods, and when the input data Xt is the data of the period Ti _{+ 6} , the data Xt of the period Ti is extracted. Then, the data Xt is supplied to the subtraction circuit (14), and the residual Et of the period Ti is extracted from the requantization circuit (16), and the residual Et is supplied to the delay memory (61). Period Ti, T
The residuals Et, Et of i- ₅ are taken out, and the residuals Et, Et are supplied to the synthesis circuit (62), and the data NMBR of the period Ti
Is supplied to the synthesizing circuit (62), and the data NMBR of the period Ti is supplied to the synthesizing circuit (62), and the residual Et of the period Ti
And the residual Et in the period Ti- ₅ are combined with the data NMBR and thus the one-word-length data WE. Thus, the turn period T
The data WE of i, Ti- ₅ is formed, and thereafter, similarly, every other block period, data WE of,... are sequentially formed, and the data WE is transmitted to the transmission system (30).

At this time, the delay coefficient ( ₁ ) is calculated in the delay circuit (27).
To k ₃ is 6 block period is delayed coefficients k ₁ to k ₃ of periods Ti is retrieved, along with the coefficients _k1 to k ₃ are supplied to the prediction filter (19), is transmitted to the transmission system (30) . In addition,
Processing and transmission of the coefficients k ₁ to k ₃ is performed for each block period.

Further, the gain coefficient Gi _{+ 5} of the period Ti _{+ 5} is obtained from the calculation circuit (26).
Is extracted, and the coefficient Gi _{+ 5} is stored in the memory for delay (63).
To obtain a gain coefficient Gi for the period Ti, and this coefficient Gi is supplied to the gain control circuits (15) and (17) and sent out to the transmission system (30). The transmission of the coefficient Gi is also performed for each block period.

Also, the gain coefficient of the period Ti ₊₅ , Ti, Ti _-5 from the memory (63)
Gi ₊₅ , Gi, Gi _-5 are extracted and supplied to the bit length calculation circuit (64) of these coefficients. Then, as shown in FIG. 3A, during the numbering period Ti, the value Ri is detected from the coefficients Gi and Gi- ₅ , and is converted from the value Ri into data NMBR. Circuit (16) and synthesis circuit (6
In 2), a signal specifying the number of bits of the residual Et in the period Ti is supplied. Also, as shown in FIG. B, during the number period Ti _{+ 1} , the coefficient G in the memory (63) advances by one block period, so that the coefficient Gi _{+ 6} , G from the memory (63).
i _{+ 1} and Gi- ₄ are taken out and supplied to the calculation circuit (64). In this period Ti _{+ 1} , the value Ri _{+ 1} is obtained from the coefficients Gi _{+ 1} and Gi _{+ 6.
+1} is calculated and converted to data NMBR, and this data NMBR is supplied to the circuits (16) and (62).

Then, as shown in FIG. C and subsequent figures, data NMBR is similarly obtained for the period after the number, and the circuit (1
The bit length of the residual Et in (6) and (62) is controlled.

In the calculation circuit (64), data WNMB is formed from the data NMBR and NMBR of two block periods in which the residuals Et and Et in the data WE are not paired, and the data WNMB is transmitted for each block period. It is sent to the system (30).

Each data is transmitted from the encoder (10) to the transmission system (30).
(Not shown), the timing of the data is corrected according to the transmission format of the transmission system (30). Also, since the data WNMB is newly transmitted, the data amount in one field period is increased by 26 words to 983 words, which is smaller than the data amount of 1050 words in the current 8-mm video.

FIG. 2 shows an example of a decoder (40) corresponding to the above-mentioned encoder (10). However, in this case, the transmission system (30)
It is assumed that each data supplied from to the decoder (40) has been corrected to the timing required for the following decoding processing.

That is, the residual data WE is transmitted from the transmission system (30) every other block period for every two blocks forming a pair. As shown in FIG. It is supplied to the memory (71) every other period.
Then, as shown in FIG. C, when the data WE in the periods Ti ₊₆ and Ti ₊₁ (numbered period) is supplied to the memory (71), the data WE is supplied to the separation circuit (72). At the same time, the data NMBR of the period Ti _{+ 1} is supplied to the separation circuit (72), and the residual Et of the period Ti _{+ 1} is extracted from the data WE based on the data NMBR, and the residual Et is used as the gain control circuit (41). )
Supplied to

Then, in the next block period, as shown in FIG. D, the data WE of the periods Ti ₊₂ and Ti _-3 (numbered periods) are taken out from the memory (71) and supplied to the separation circuit (72). At the same time, the data NMBR of the period Ti _{+ 2} is supplied to the separation circuit (72), and the residual Et of the period Ti _{+ 2} is extracted from the data WE and supplied to the control circuit (41).

Thereafter, as shown in FIG. E and thereafter, the above processing is repeated with a cycle of two block periods, and the residual Et is sequentially supplied to the control circuit (41).

Further, at that time, the data WNMB from the transmission system (30) is supplied to the separation circuit (73), and for each block period, the data NMBR of the block period is extracted from the data WNMB and the separation circuit (73) 72).

Therefore, the decoded data t is taken out from the terminal (44) for each sample.

H Effects of the Invention Thus, according to the present invention, audio data Xt by ADPCM is transmitted. In this case, in particular, according to the present invention, two block periods are paired, and the residual Et in the block period of the pair is used. The number of bits of the residual Et is not increased, so the number of bits of the residual Et when the level changes rapidly can be increased, and large distortion and noise can be generated at the time of the change. Will not occur.

The data NMBR indicating the accommodation state of the number of bits is transmitted for each block period, and the data NMBR, NMB is used for the block period before and after the pair.
Since R has a two's complement relationship, even if an error occurs in one data NMBR, the errored data NMBR can be restored from the other data NMBR, and thus the number of bits of the residual Et can be more correctly restored. it can.

Furthermore, the data to be transmitted, because all are one word length (for the coefficients k ₁ to k ₃ can be separated in every 8 bits), considers these data and PCM audio in the current 8 mm video Recording and playback.

[Brief description of the drawings]

FIG. 1 is a system diagram of an example of the present invention, and FIGS. 2 to 10 are diagrams for explaining the same. (10) is an encoder, (30) is a signal transmission system, and (40) is a decoder.

Continuation of front page (58) Field surveyed (Int.Cl. ⁶ , DB name) H04B 14/00 H03M 7/00

Claims (1)

(57) [Claims] An A / D-converted sample is set as one block for each predetermined number, and for each block, data obtained by compressing the number of bits from the samples included in the block is obtained. Is transmitted in a predetermined number of bits, and the two blocks at a predetermined time interval in which the masking effect is effective are made into one pair, and the sum of the number of bits between the two blocks forming the pair is the other. Bits are flexibly transmitted by increasing or decreasing the number of bits of each block so as to have a relationship equal to the sum of the numbers of bits of the two blocks in the period, and data indicating the state of accommodation of the number of bits is also transmitted. The encoding method for digital audio data.