JPH04168500A - Signal coding method - Google Patents

Abstract

PURPOSE:To decide a real assignment bit number of each band area for restricting distortion of the wave form at minimum by truncating a fraction easily on the basis of a computed assignment bit number, and correcting the bit number to maintain the bit number of the whole constant. CONSTITUTION:A decision unit 6 computes a computed assignment bit number R1i on the basis of the data of the power Ui of each band area, and truncates a fraction of the number R1i less than a decimal point to obtain an integral bit number R2i which consists of only integral number. Shortage of the number R2i from the total bit number N - R of the whole is computed to obtain a residual bit number A, and simultaneously, one bit is respectively added to A number of the integral bit number R2i, which correspond to the number R1i and arranged in order of largeness of fraction of the number R1i, to eliminate the shortage on the basis of the number A. The total number of the integral bit number R3i after this addition is equal to the number N - R, and the number R2i is corrected to the number R3i by adding one bit to the number R2i after the truncation to restrict dislocation of the number R1i from the number R3i. The number R3i is thereby decided as a real assignment bit number of each band area, which makes distortion of the wave form minimum.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a signal encoding method applied to digital transmission of analog signals such as audio signals.

[Conventional technology]

Conventionally, in the field of digital transmission of audio signals, a signal encoding method using subband coaching (hereinafter referred to as ``s'') encoding has been proposed in order to minimize redundancy and encode efficiently. ing.

Regarding this SBC signal encoding, the document ``High Efficiency Coding of Speech'' (published by Hyoka Publishing - September 3, 1986)
As described on pages 58 to 61 and pages 106 to 110 of , first, an audio signal is divided into a plurality of (N) band components using a band pass filter.

Furthermore, each band component is converted to a low frequency (baseband) and then converted into a digital signal such as an adaptive PCM code by encoding processing.

At this time, the overall total number of bits is fixed to a constant value (integer value) based on the bit number limit of the transmission path, etc., but the number of bits for each band (number of assigned bits) that is allocated by dividing the total number of bits is: γi in formula 7.1 on page 107 of the above document is O
By frequency domain adaptive bit allocation based on the following equation 1, the optimum number of bits for each band is determined in calculations so that waveform distortion is minimized.

However, RI i + RI WJ + υi indicates the following values.

R, ;: Calculated number of allocated bits π for the i-th band
: Average number of allocated bits for all bands, which is 1/N of the total number of bits Ui : Lli/wi (Ui: Power of i-th band component) [Problem to be solved by the invention J In calculation of each band found from the above formula 1 The number of allocated bits is
It does not necessarily have to be an integer value, but generally includes fractions after the decimal point.

On the other hand, the actual number of bits allocated for each band must be an integer value, and the total number of bits or the average number of bits must be equal to the total number of bits or dice set.

Therefore, it is necessary to perform some processing (correction) on the calculated number of bits allocated to each band (I:'''-/) to determine the actual number of bits allocated to each band, but the specific method for this has not yet been invented. Not yet.

Note that if the calculated number of allocated bits for each band is simply rounded up, rounded down, or rounded to the nearest whole number L2 to obtain the actual number of allocated bits, the total number (or average number of bits) is
There will be excess or deficiency in the total number (or R).

And, the above-mentioned surplus and deficiency cannot be appropriately resolved,
If we adjust the number of bits allocated to each band appropriately,
A situation arises in which waveform distortion cannot be minimized.

An object of the present invention is to make it possible to easily determine the actual number of allocated bits for each band while keeping the overall number of bits constant and minimizing waveform distortion.

[Means to solve the problem]

In order to achieve the above object, in the signal encoding method of the present invention, each integer bit number is obtained by rounding down the fraction after the decimal point from each calculated calculated allocation B/ ) number, and the total number of bits is calculated. The number of remaining bits is obtained by subtracting each integer bit number from Determine the number.

0 action] In the case of the signal encoding method of the present invention configured as described above, each integer number of bits is subtracted from the decimal number of the calculated number of allocated bits for each band, and from the total number of bits set as a whole. Accordingly, when each integer number of bits is taken as the number of allocated bits for each band, the shortfall from the total number of bits is determined as the remaining bits.

Furthermore, the integer bit numbers of the remaining bit numbers in order from the largest fraction before rounding down are corrected by adding or subtracting one bit from each of the remaining bit numbers, thereby eliminating the shortage.

Then, each integer bit number after performing the increase/decrease correction is
It is determined by the actual number of allocated bits for each band and (-).

At this time, the total number of actually allocated bits of each determined band always matches the total number of bits.

Moreover, for a large fraction of the allocated bit numbers in each calculation where the error becomes large due to truncation, the error can be suppressed by 1-bit addition correction based on the remaining number of bits after truncation.

Therefore, by simply rounding down fractions and correcting the number of bits based on the calculated number of allocated bits, the actual number of allocated bits for each band can be adjusted to keep the overall number of bits constant and minimize waveform distortion. number is determined.

〔Example〕

One embodiment will be described with reference to FIGS. 1 and 2.

Figure 2 shows the block configuration when digitally transmitting an audio signal by SBC encoding and decoding.
The audio signal at the input terminal (2) of the N
The frequency is divided into N band components by N bandpass filters.

Furthermore, each band component is low-pass converted to a baseband signal by shifting the frequency axis to the lower frequency side by the center frequency of each band-pass filter by low-pass conversion B (4), and each band component is converted to a baseband signal. Power IJi is required.

Then, each band component that has been low-pass converted is supplied to the encoding unit (5), and the data of the power IJi is supplied to the transmitting side bit allocation decision 8 (6), and this decision 8 (6) will be explained next. Through frequency adaptive bit allocation, the number of allocated bits for each band is determined.

That is, in decision 8 (6), as shown in FIG. 1, the number of calculated bits (.R1,) to be allocated to each band is calculated from Equation 1 above, based on the data of the power IJi of each band.

At this time, the following equation 2 holds true between the overall total number of bits (-NR) (R: average number of bits allocated for each band) and each number of allocated bits R+i.

Next, the fraction below the decimal point (decimal B) of the calculated calculated number of allocated bits R1i is rounded down, and the number of allocated bits R1i
Find integer bins 1.1 to R2, which consist only of integer parts.

Furthermore, the remaining bit number A is obtained by calculating the shortage of the total number of integer bits R2i from the total number of vias Nm using the following equation 3.

At this time, the remaining bit number A is always 0 as shown below.
Or it will be a positive integer.

8.One plane (Rz++Rz□+...+RZN)≧0Also, at the same time as calculating the remaining number of bits, the allocated bit number R1i
Collect the fractions in order from largest to largest.

Next, 1 bit is added to each of the A integer bit numbers R2i corresponding to the A Rlis in order from the largest fraction to eliminate the shortage based on the remaining bit number A.

Part 1. Then, each integer bit number 1 after the above 1-bit addition
If 'lzi is R31, this number of bits R3i
Moreover, for the number of allocated bits R1i with a large fraction, 2. The integer bit number R2i after truncation is corrected to R3i by adding 1 pip I, and the deviation of the integer bit number R3i from Rli due to the truncation of fractions is suppressed as much as possible.

Therefore, even if the integer bit number R3i is the actual number of bits allocated to each band, the number of bits allocated to each band can be determined so that the total number of bits is kept constant and waveform distortion is minimized.

However, in this embodiment, since the number of bits allocated to each band is 1 or more, each integer bit number pzi is set to R3i.
After correcting, it is determined whether or not each of the integer bit numbers R3 is less than 1.

For example, if the integer bit number R1 of the a-th band is less than ], this integer via) number R3m is corrected to 1, and at the same time, the excess of the bit number of (1-R:l-) caused by this correction is corrected. , for example, (1,-R3
-) other integer bit numbers R3, 1 bit is subtracted from each of them to eliminate them and adjust the bit allocation.

With this adjustment, each integer bit number R3,' is always 1 or more, and the total number of bits is kept constant and the waveform distortion is minimized as before the adjustment.

Then, each integer bit number R3□ after bit allocation adjustment is determined as the actual allocated bit number for each band.

Note that if R3i after correcting each integer bit number R2i does not include anything less than 1, the above-mentioned bit allocation adjustment is not performed, and each integer bit number R3i after correction is used as the actual allocated bit for each band. determined as a number.

Then, through encoding control based on the determination by the determining unit (6), the encoding unit (5) converts each band component into a digital signal such as an adaptive PcM code for each determined actual allocated number of bits for each band. and encode it.

Furthermore, information on the digital signal of each band (main information) of the encoding unit (5) and the actual number of allocated bits of each band (sub information)
is connected from the serial conversion multiplexer (7) to the transmission line (8).
) and is sent to the receiving bag W (9).

In the receiving device (9), the signal from the transmission path (8) is separated by a demultiplexer 001 for parallel conversion.

By this separation, the digital signals of each band component are respectively supplied to the decoding section Ql), and information on the actual number of allocated bits of each band is supplied to the receiving side bit allocation determining section (to).

Then, by the decoding control based on the information on the number of allocated bits in the deciding section, the decoding section (11) converts the digital signal of each band component into the digital signal of each band component similar to the output signal of the low frequency conversion section (4). The signal is decoded into a baseband signal and supplied to the high frequency conversion section a.

The conversion unit 03 performs high frequency conversion on the frequency axis of each baseband No. 6, and returns the signal to the original band components similar to the output signal of the band division unit (3).

Then, each band component of the converting section a is frequency-synthesized and combined by the band combining section a, and based on this combination, a reproduced audio signal similar to the audio signal of the input terminal (2) is output from the output terminal 09. Output.

At this time, since each band component is digitally transmitted while minimizing waveform distortion, the audio signal at the output terminal 09 has extremely little waveform distortion.

In the case of FIG. 2, it was confirmed that the overall bit NR improved by about 3 dB after rounding off the calculated number of allocated bits for each band.

By the way, if it is not necessary to increase the number of allocated bits for each band to 1 or more, without adjusting the bit distribution described above,
Each integer bit number Rffi before this adjustment may be used as the actual allocated bit number for each band.

Further, in the embodiment described above, the case where an audio signal is encoded has been described, but it is of course applicable to the case where an analog signal other than an audio signal is encoded.

Furthermore, it can be applied not only to digital transmission of analog signals such as audio signals, but also to, for example, digital recording and playback.

〔Effect of the invention〕

Since the present invention is configured as described above, it produces the effects described below.

Subtract each integer bit number from the total number of bits in the field, and when each integer bit number is taken as the allocated bit number for each band, the shortfall from the said total number of bits is determined as the remaining bits, and the fractions are calculated in descending order of the number of bits before rounding down. Each integer bit number of the number of remaining bits was corrected by increasing or decreasing by 1 bit by distributing each 1 bit of the remaining number of bits, and the shortage was resolved and determined as the actual allocated bit number for each band. The total number of bits actually allocated to each band always matches the total number of bits.

Moreover, for a large fraction of the allocated bit numbers in each calculation where the error becomes large due to truncation, the error can be suppressed by 1-bit addition correction based on the remaining number of bits after truncation.

Therefore, by simply rounding down fractions and correcting the number of bits based on each calculated number of allocated bits, the actual allocated bits of each band can be adjusted to keep the overall number of bits constant and minimize waveform distortion. The number can be determined, and based on this determination, an optimal signal encoding of the subband coding can be performed.

[Brief explanation of the drawing]

1 and 2 show one embodiment of the signal encoding method of the present invention, FIG. 1 is a flowchart of the process of determining the number of allocated bits, and FIG. 2 is a block diagram of digital transmission. (3)...Band division section, (4)...Low frequency conversion section, (
5)... Encoding unit, (6)... Transmitting side bit allocation determining unit. Agent Patent Attorney Fuji 1) Ryuutabe 1 Zuken 2! la ~1 Ge・91 Machiharu Myofi 8・Buban Maki 3, Risame 41J Marshal
91 recognition road 4゛46 imago and f distribution
10 ・ Death IL-j7~shi9 sun (
p5... everyone? 4?
11 Zu 1 Ihi 4 children 6...υ2 made...
``Biwatodo'' Toharudai Q 12 Cost I C1'] Hit $1'' is also large I 17゛ °Maru 10? Bottom size (p
13 ・ Ship-ni 1 sad Ao) P14 Dwell f; oblique r

Claims (1)

[Claims] (1) Sub-band coding divides an analog signal such as an audio signal into multiple band components, and calculates the fraction below the decimal point of each band to keep the overall number of bits constant and minimize waveform distortion. Calculate the calculated number of allocated bits including the calculated number of allocated bits, and convert and encode each band component into a digital signal based on the actual allocated number of bits of the integer value of each band determined from the calculated number of allocated bits. In the signal encoding method, each number of integer bits is obtained by rounding down the fractions from each of the number of allocated bits in each calculation, the number of remaining bits is obtained by subtracting each number of integer bits from the total number of bits, and the number of remaining bits is obtained by subtracting the number of integer bits from the total number of bits, A signal encoding method, characterized in that the actual allocated bit numbers are determined by adding one bit of each of the bit numbers to each of the integer bit numbers in descending order of the fractional number.