JPH06348294A - Band dividing and coding device - Google Patents

Abstract

PURPOSE:To solve the problems of a conventional device which has a larger waveform distortion and an increased quantization error. CONSTITUTION:The device is provided with a band dividing filter 1 which divides digital signals of voice signals into plural bands, an electric power computing means 2 which obtains the power of the output of the filter 1 for every block and quantizes them in a form of 2sigma power, a quantizing means 6 which adaptively quantizes for every block and a quantized bit number control section 5 which adaptivley controls quantized bit numbers of each band of the means 6 based on the value of sigma. The section 5 regularly varies the quantization bit number for every sample data outputted from the filter 1 and sets the quantization bit number of the means 6, that is assigned to each band, to a number which is larger than zero including decimals. It is desired that the section 5 sets a bit number which is larger than zero and less than 1 for the band that has a bit number assigned to the means 6 is less than zero.

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 device for mainly coding a digital voice signal used in a voice IC recorder or the like.

[0002]

2. Description of the Related Art Conventionally, when a digital audio signal is compression-encoded, a method of dividing the audio signal into several sub-bands and encoding them has been adopted.

For example, Japanese Patent Application Laid-Open No. 4-1 by the applicant of the present invention
Japanese Patent No. 29430 (H03M 7/00) discloses an example thereof. In this publication, an input digital audio signal is first described as a band division filter group (BPF: Band Path Filtration).
Each of the divided digital audio signals is frequency shifted down by the center frequency of the pass band, where it is converted to a low frequency signal through a low pass filter, which is converted from the band pass filter bandwidth. Downsampling is performed at a determined sampling frequency, and the result is encoded by forward adaptive PCM (Pulse Code Modulation) for each block.

In the decoding unit of this publication, a reverse operation to the above operation is performed for each band to create a bandpass signal for each band, and these are added together to form an output voice waveform. .

As an advantage of such a technique, (1) it is easy to effectively perform bit allocation for each band, (2)
The quantization error in each band is related only to the signal in that band, and when the power in that band is small, even if the quantization distortion is large, it is only a small error from the viewpoint of the whole signal. The quantization error ideally becomes white noise, spreads widely in the frequency spectrum, and only a part of it falls in the band.

From the above (1) and (2), it is desirable to adaptively change the number of bits per sample allocated to each band according to the power of the signal in that band. In that case, the following equation It was said that it is most appropriate to divide according to the above (waveform distortion is minimized).

[0007]

[Equation 1]

Here, the electric power obtained for each block is set to 2σ
If quantized to the power of

[0009]

[Equation 2]

[0010] Further, for example, in the above equation 2, W _i = 1 /
4, and N = 4,

[0011]

[Equation 3]

[0012]

In fact, in this case, R _i may be obtained as a decimal number or a negative number, so that a conditional statement for making those numbers positive integers (including 0) is necessary. When the number of bits R _i assigned to each band becomes a decimal number, it must be a positive integer by rounding, but at that time, there is a problem that the waveform distortion becomes large.

When R _i becomes a negative number, the waveform distortion can be minimized by setting the number of allocated bits to 0. However, if the number of blocks to which 0 bit is allocated increases, there is a fear that an abnormal sound will be generated in the reproduced sound. Therefore, in order to prevent this abnormal noise, it is necessary to allocate 1 bit or more to every band, and at that time, there is a drawback that the number of quantization bits is insufficient and the quantization error increases.

[0015]

As in the prior art described above,
Number of bits R allocated to each band when waveform distortion is minimized
_i Is found in the above equation 1. But R_i Is a decimal place
If this is the case, a conditional statement is required to make it a positive integer by rounding.
And at that time, there is a problem that the waveform distortion becomes large.
there were.

When R _i becomes a negative number, the waveform distortion can be minimized by setting the number of allocated bits to 0. However, if the number of blocks to which 0 bit is assigned increases, an abnormal sound is generated in the reproduced sound. Therefore, in order to prevent this abnormal noise, it is sufficient to allocate 1 bit or more to the band, but at that time there is a problem that the quantization error increases.

The present invention has been made in order to solve the conventional problems of increasing the waveform distortion and increasing the quantization error.

[0018]

According to the present invention, a band division filter for dividing a digital signal such as a voice signal into a plurality of bands and an electric power for each block of the output of the filter are obtained.
Power calculation means for quantizing into the form of σ, adaptive quantizing means for each block, and adaptively controlling the number of quantizing bits for each band of the quantizing means by the value of σ. A quantization bit number control unit, wherein the quantization bit number control unit regularly changes the quantization bit number for each sample data output from the band division filter and assigns the quantization bit number to each band. Is set to a number greater than 0 including a decimal, and preferably, the quantization bit number control unit is set to 0 or more for a band in which the number of bits assigned to the quantization means is 0 or less. The number of bits is set to be less than one.

[0019]

In the above structure, the number of bits R _i to be assigned to each band when the waveform distortion is minimized is obtained by the equation 1 as in the prior art. When R _i becomes a decimal number, the number of bits assigned to each band can be made equal to or close to R _i by the above means, and waveform distortion can be reduced.

Further, when R _i becomes a negative number, the number of bits of at least less than 1 exceeding 0 is assigned to any band so that the audible feeling can be obtained without increasing the quantization distortion of other bands. The abnormal noise of can be removed.

[0021]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described in detail below with reference to the functional block diagram of FIG. 1 and the bit number allocation explanatory diagrams of FIGS.

In FIG. 1, 1 is a QMF (Qudrature Mir)
ror Filter), input audio is 1 / sampling frequency
The frequency is divided into a high frequency band and a low frequency band, and down-sampling is performed to ½ of the sampling frequency. In this embodiment, the QMF1 is used in two stages to divide the input voice into four bands and the filter. The output of 1 is dropped to the baseband.

Reference numeral 2 is a power calculation means for collecting the sound divided into bands for each band into blocks for every several samples (for example, for every 128 samples) and calculating the power in the block. Numeral 3 is a quantizing means for quantizing the calculated electric power to the power of 2 to the power of 2 and, in fact, expresses the electric power in a binary number, and searches for the place where 1 stands on the leftmost, and at what bit from the right It is checked whether or not it exists, and 1 is subtracted from the value to divide it by 2 to obtain σ _i for each band.

Reference numeral 4 is a normalizing means for normalizing the electric power in one block to be 1, which makes it possible to quantize digital audio signals having a wide dynamic range with high precision.

Reference numeral 5 denotes a quantization bit number control unit for allocating the number of quantization bits used by the quantizer of each band, and σ of each band
Can be assigned by the equation 4 using the value of

[0026]

[Equation 4]

Actually, R _i may be obtained as a decimal number larger than 0, and the value taken by R _{i in} the above equation 4 is x. 00,
x. 25, x. 50, x. There are four types, 75 (x is an integer). At this time, as shown by 1 (1) to (4) in FIG. 2, the average number of bits assigned to each band is made a decimal by regularly changing the number of bits assigned every other sample.

That is, in (1), all samples are quantized with x bits, in (2), 3 out of 4 samples are quantized with x bits, and the remaining 1 sample is quantized with x + 1 bits. ) Quantizes 2 out of 4 samples with x bits and quantizes the remaining 2 samples with x + 1 bits, (4) quantizes 1 sample out of 4 samples with x bits and 3 remaining samples with x + 1 bits Quantization yields the desired fractional bits.

In some cases, R _i may be obtained by 0 or a negative number. In that case, the number of bits greater than 0 and less than 1 is assigned as shown in FIG. That is, in (1), 1 out of 4 samples
Quantize the sample with 1 bit and set the remaining 3 samples to 0. In (2), quantize 2 out of 4 samples with 1 bit and set the remaining 2 samples to 0. (3) sets 4
Quantize 3 out of 3 samples with 1 bit and leave 1 remaining
By setting the sample to 0, the desired fractional bit is obtained.

Numeral 6 is a signal quantizing means for quantizing the signal normalized by the normalizing means 4 in accordance with the number of quantizing bits assigned by the quantizing bit number controller 5, and 7 is a quantizer for each band. It is a multiplexer that multiplexes the generated code and auxiliary information σ _i .

Reference numeral 8 is a demultiplexer for dividing the signal from the multiplexer 7 into codes of each band and auxiliary information σ _i .

Denoted at 9 is a decoding bit number control unit for calculating bit allocation in the same manner as the quantization bit number control unit 5, 10
Is a power decoding means for decoding the power of each band, 11 is a power decoding means for decoding the power of each band, and 12 is a signal that has been normalized using the decoded power. It is an inverse normalization means.

Reference numeral 13 is a decoding-side QMF for up-sampling and band-combining a band-divided and down-sampled signal.

With the above-mentioned structure, for any speech, band division is performed by the filter 1 to obtain power of each band, and the quantizing means 6 quantizes the value to the power of 2σ, Based on the value of, the number of quantization bits to be assigned to each band can be adaptively controlled by the number of bits including the decimal number in the quantization bit number control unit 5, so that optimum coding can be performed.

The encoded signal can be decoded into the original digital audio signal by following the operation reverse to the above operation.

[0036]

According to the apparatus of the present invention, when band-division encoding a digital audio signal, the power for each band is obtained, and the value is quantized to the power of 2 2 σ to obtain each band. It becomes possible to adaptively control the number of quantized bits to be assigned, and by regularly changing the number of quantized bits every other sample, it is possible to deal with the case where the number of assigned bits for each band becomes a decimal number. By reducing the quantization distortion, and further by allocating the number of bits greater than 0 and less than 1 at least to the band in which the number of allocated bits is 0 or less for minimizing the waveform distortion, Since it is possible to remove perceptual abnormal sounds without increasing the quantization distortion, it is extremely useful because it can optimally encode any voice.

[Brief description of drawings]

FIG. 1 is a block diagram for explaining an embodiment of the device of the present invention.

FIG. 2 is a diagram for explaining decimal bit allocation.

FIG. 3 is a diagram for explaining allocation of a bit number greater than 0 and less than 1.

[Explanation of symbols]

 1 Band Division Filter (QMF) 2 Power Calculation Means 5 Quantization Bit Number Control Unit 6 Quantization Means

Claims (2)

[Claims] 1. A band division filter for dividing a digital signal such as a voice signal into a plurality of bands, and a power calculation means for calculating the power of each block of the output of the filter and quantizing it to the power of 2 2σ. Quantization means for adaptively quantizing each block, and a quantization bit number control unit for adaptively controlling the number of quantization bits for each band of the quantization means by the value of σ, The number control unit regularly changes the number of quantization bits for each sample data output from the band division filter, and sets the number of quantization bits of the quantization means assigned to each band to a number greater than 0 including a decimal. A band division encoding device characterized by setting. 2. The band division coding apparatus according to claim 1, wherein the quantization bit number control unit is a bit greater than 0 and less than 1 in a band in which the number of bits assigned to the quantization means is 0 or less. A band division encoding device characterized by setting a number.