JPH0537395A - Band-division encoding method - Google Patents

Abstract

PURPOSE:To offer a band-division encoding method by which the number of quantized bits can be assigned to a band having information which is important to the auditory sense of a human being, as for the band-division encoding method by which a music signal and a sound signal is compressed and encoded. CONSTITUTION:An input signal is divided into the N pieces of bands by a filter means 1, an index value is detected from the signal of each band by a number of quantized bit assigning means 5, and the index Xi of each band is compared with a reference value TH(i) preliminarily set at each band. Then when a preliminarily set conditional expression; Xi <=TH(i) is fulfilled, the number of quantized bits to be assigned to the (i)th band is defined as 0.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a band division coding method for compressing and coding a music signal and a voice signal.

[0002]

2. Description of the Related Art In the conventional method of allocating the number of quantized bits for band-division coding, the product of the downsampling frequency and the number of quantized bits to be allocated, that is, the amount of information for each band is equal to the contribution to the clarity index of that band. The number of quantized bits was assigned as described above (for example, Kazuo Nakata; “High-efficiency coding of speech” 1986 (pages 56-61)).

[0003]

In the conventional method of allocating the number of quantized bits for band division coding, the number of quantized bits is allocated so that the amount of information for each band is equal to the degree of contribution to the clarity index. It was effective for high efficiency coding. However, music and voice are recognized by human hearing, and due to the characteristics of human hearing, there are cases where the human ear cannot perceive even if there is information. Assigning the number of quantization bits to a signal in such a band is
There was a problem that it could not be said that the information important to human hearing was encoded efficiently.

The present invention solves the above problems, and an object of the present invention is to provide a band division coding method for allocating a quantization bit number to a band having information important for human hearing.

[0005]

In order to achieve the above object, the present invention comprises a first step of dividing an input digital signal into N bands by using a band division filter,
A second step of assigning an appropriate number of quantization bits to each of the N bands, and a quantization bit assigned to each of the signals of the N bands obtained by the first step by the second step. And a third step of quantizing with a number, wherein the second step includes a step of detecting an index value for a signal of each of the N number of bands, and preset for each band. And comparing the reference value with the index value detected in the step, and setting the number of quantization bits to be assigned to the band to 0 when the preset condition is satisfied.

[0006]

When the band division coding method as described above is used for coding, information important to human hearing can be coded with high efficiency.

[0007]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. (FIG. 1) is a block diagram for explaining a quantized bit allocation method for band division coding according to the present invention. Reference numeral 1 is a filter means for dividing an input signal into N bands. It is composed of N band division filters of N. Numeral 5 is a quantizing bit number assigning means, a detecting means for detecting an index value from a signal of each band divided by the filter means 1 into N bands, a reference value preset for each band and the detecting means. Comparing means for comparing the index values detected by, and setting the number of quantization bits allocated to the band to 0 when the preset condition is satisfied. Reference numeral 6 denotes a quantizing means, which quantizes each signal divided into N bands by the filter means 1 by the quantizing bit number assigned by the quantizing bit number assigning means 5.

An embodiment of the quantizing bit number assigning means 5 will be described with reference to the flow chart shown in FIG.

First, in step 1, the maximum absolute value of the amplitude is detected as an index value for each band from a plurality of signals in each band. X ₁ represents the maximum value of the first band, X ₂ represents the maximum value of the second band, and X _N represents the maximum value of the Nth band. In step 2, i is set to 1, and in step 3, it is determined whether i is N or less. If i is N or less, in step 4, the reference value TH (i) for the i-th band is read from (Table 1).

[0010]

[Table 1]

In step 5, the maximum value X of the i-th band
_{When i} is compared with the reference value TH (i) and the conditional expression X _i ≦ TH (i) is satisfied, 0 bit is allocated to the i-th band in step 6. This reference value TH (i) represents the minimum audible level of the human ear for the i-th band, and signals below the reference value TH (i) are not perceived by the human ear, so this conditional expression is satisfied. It is not appropriate to give the band a quantization bit rate. If the above conditional expression is not satisfied, in step 7, an appropriate number of quantization bits is assigned to the i-th band. In step 8, 1 is added to i, and the process returns to step 3. While the condition of step 3 is satisfied, steps 4 to 8 are repeated until i is N
If it exceeds, it ends and moves to the next means. From the above, the number of quantization bits in all bands from 1 to N is determined. Here, step 1 corresponds to detection means, and steps 5 and 6 correspond to comparison means.

In this embodiment, the maximum absolute value of the amplitude of the signal in each band is used as the index value. (1) The maximum squared value of the amplitude of the sampled signal in each band (2) Each band The desired band division coding can be performed by using the maximum value of the number of significant digits when the absolute value of the amplitude of the sampled signal of is expressed by a power of 2.

[0013]

The present invention provides a band division coding method as described above, and since a signal in a band that is not important to human hearing is quantized with 0 bits,
It becomes possible to allocate a large number of quantization bits to a signal in a band necessary for human hearing, and it becomes possible to perform coding suitable for human hearing.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining an embodiment of a band division coding method according to the present invention.

FIG. 2 is a flowchart for explaining an embodiment of a quantization bit number assigning means according to the present invention.

[Explanation of symbols]

 1 Filter Means 2, 3, 4 Band Division Filters 5 Quantization Bit Number Allocation Means 6 Quantization Means

Claims (1)

Claim: What is claimed is: 1. A first step of dividing an input digital signal into N bands using a band division filter, and an appropriate number of quantization bits for each of the N bands. Band division including a second step of allocating and a third step of quantizing each signal of the N bands obtained by the first step with the number of quantization bits allocated by the second step. In the encoding method, the second step detects the index value for the signals of each of the N bands, and compares the reference value set in advance for each band with the index value detected in the step. Then, when the preset condition is satisfied, the number of quantization bits to be assigned to the band is set to 0 bit, and the band division encoding method.