JPH0420539B2 - - Google Patents

Description

[Detailed Description of the Invention] [Summary] The present invention provides a method for effectively utilizing encoded information bits in accordance with the redundancy of information amount of a signal to be encoded in encoding an audio or image signal.
Variable bit rate ( This makes it possible to encode the data at a certain rate.

[Industrial application field]

The present invention relates to a coding/transmission device, and particularly to a variable amount coding/transmission device for audio or image signals, etc.

High-efficiency transmission of audio signals, image signals, etc. is being applied in a wide range of areas, such as high-efficiency encoding transmission of audio signals and voice storage response services. In the encoded transmission of audio signals, there is a need for a method such as packetized transmission that can control the amount of encoded information according to the permissible transmission amount. On the other hand, in order to compress information while maintaining high playback audio quality in voice storage response systems, etc., a small amount of information is allocated to parts with a small amount of information generation, including silent periods, and to parts with a large amount of information generation. There is a need for a coding system that allocates a large amount of encoded information and is effective in allocating the amount of information.

[Conventional technology]

FIG. 3 shows a coding/transmission device disclosed in Japanese Patent Application No. 142307/1987 filed by the applicant on June 28, 1985, which includes a plurality of adaptive predictive coding units 1-1 to 1-1. 1-k and quantization error power calculation unit 20
, an optimal quantization determining unit 21, a selecting unit 22, and a multiplexing unit 23, and each adaptive predictive coding unit includes a quantizer Qi (i=1, 2...k) and an inverse quantization unit. vessel
Qi ^-1 (i=1, 2...k), a zero predictor Hz, a polar predictor Hp, and various adders Sia~Sid (i=1, 2...
k).

The current input audio signal X(n) is compared with each predicted value X'i(n) in the adder Sia, and the residual signal Ei(n) is
is quantized by a quantizer Qi to produce a quantized value Ii(n) of the residual signal, which is transmitted to the receiving side. The inverse quantizer Qi ^-1 (n) inversely quantizes the quantized value Ii (n) and outputs a value Ei' (n) in which the residual is restored. predictor
Hz and Hp generate the above-mentioned predicted value X'i(n) which predicts the current input based on the restored value Ei'(n).

In this case, the quantization step update rate of the quantizer and inverse quantizer Mi(I(n)) (i=1, 2...k)
differs for each adaptive predictive coding unit, and the residual Ei
(n) and its inverse quantized value Ei′(n).The quantization error power calculation unit 20 calculates the quantization error power based on the quantization error signal e _i =Ei(n)−Ei′(n). Calculate,
The optimum quantization determining section 21 determines the smallest one of them, the selecting section 22 selects the adaptive predictive coding section, and the multiplexing section 23 selects the signal Iopt from the selecting section 22.
(n) and the signal Mopt from the determining unit 21 are multiplexed.

[Problem that the invention seeks to solve]

Although such a coding transmission device is effective in itself, the number of quantization bits in the quantizer and inverse quantizer is fixed, and the coding transmission rate is constant. Therefore, it is not possible to make effective use of the encoded information bits by changing the encoding transmission rate according to the amount of information generated by the information source, and when the permissible amount of information to be transmitted is variable in packetized transmission, etc. There was a problem that it was not possible to encode within the amount.

SUMMARY OF THE INVENTION Therefore, an object of the present invention to solve this problem is to provide an encoding/transmission apparatus that can vary the transmission speed for companding according to the amount of generated information.

[Means for solving problems]

FIG. 1 is a diagram showing the principle of the encoder and transmitter of the present invention to achieve the above object, and 2-1, 2
-2,...2-k are a plurality of adaptive predictive coding units each having an adaptive quantizer with a different number of quantization bits and an inverse quantizer, and 3 is a reproduction of the input signal and each predictive coding unit for each frame. An adaptive predictive code is calculated according to a predetermined S/N threshold or noise level threshold determined according to the allowable encoding transmission rate while calculating the S/N or noise level evaluation function value between the output signal and the output signal. a transmission rate control unit that selects the quantized outputs of the quantization units 2-1, 2-2, ... 2-k and controls the transmission rate;
4 is an adaptive predictive encoding unit 2-1, 2 for each frame.
In order to avoid data discontinuity in the reproduced output when switching between -2,...2-k, the quantization step size and adaptive prediction parameters of the selected adaptive predictive encoder are changed to the other adaptive predictive encoders. This is a parameter copy control unit that performs copy control.

In addition, in the transmission rate control unit 3, when the input signal power is lower than the threshold value corresponding to the silent state,
Set the number of quantization bits to 0.

[Effect]

In FIG. 1 showing the present invention, adaptive predictive coding units 2-1, 2-2, ... 2-k each have, for example, a 1-bit to k-bit adaptive quantizer and inverse quantizer for each frame. , the transmission rate control unit 3 outputs the reproduced output signals X ₁ ′(n), X ₂ ′(n), . . . from each quantizer.
The evaluation function value of the S/N or noise level between X _k ′(n) and the input audio signal X(n) is calculated for each frame, and a predetermined S/N threshold determined according to the allowable encoding transmission rate Adaptive predictive coding units 2-1, 2-2, . . .
The transmission rate is controlled by selecting the quantized outputs I ₁ (n), I ₂ (n), . . . I _k (n) of 2-k. That is,
The appropriate number of bits to encode each frame is determined according to the obtained S/N evaluation function value or noise level evaluation function value and each threshold value, and information necessary for encoding is determined. More coded bits are allocated to parts of a signal with a large amount of signal with rapid changes, and fewer coded bits are allocated to parts where voice quality can be maintained even with a small number of coded bits. If the input signal power is equal to silence, the number of quantization bits is set to zero.

In order to avoid data discontinuity in the reproduced output when switching the adaptive predictive coding unit for each frame, the parameter copy control unit 4 sets the quantization step size Δl and adaptive prediction parameter P ₁ of the selected adaptive predictive coding unit. , P ₂ , ...P _k to other adaptive predictive coding units.

〔Example〕

Examples of the present invention will be described below.

FIG. 2 shows an embodiment of the coding/transmission apparatus of the present invention conceptually shown in FIG. 1, and the configuration of each adaptive predictive coding section 2-1, 2-2,...2-k is , adaptive predictive coding units 1-1, 1-2,...1- in FIG.
As in the case of k, quantizer Qi, inverse quantizer
Qi ^-1 , zero predictor Hz, polar predictor Hp, adder Sia,
Sic and Sid, but the adder Sib for generating quantization error power is not used. Each adaptive predictive coding unit receives a reproduced output signal Xi(n) from the adder Sic.
is occurring.

The transmission rate control circuit 3 includes each adaptive predictive encoder 2
-1, 2-2,...2-k playback output signal Xi
(n) and input signal X(n) for each frame.
An evaluation function value calculation unit 11 that calculates N or a noise level evaluation function value, receives allowable transmission amount information from a transmission path, sets a threshold value S/Nth or a noise level threshold value Nth, and corresponds to a silent state. Threshold adjustment circuit 1 that also sets the power threshold Pth to
2, and a comparator 13 that compares the calculated S/N evaluation function value or noise level evaluation function value with the threshold S/Nth or Nth from the threshold adjustment circuit 12.
and a quantization bit number determination unit 14 that determines the optimum number of quantization bits based on the comparison result of the comparator 13.
and a selection unit 1 that selects the quantized residual signal Ii(n) of the adaptive predictive coding unit corresponding to the number of quantization bits determined by the quantization bit number determination unit 14.
5, and a multiplexing unit 1 that multiplexes the selected residual signal Ii(n) and auxiliary information indicating the number of quantization bits.
6.

The transmission speed control circuit 3 also receives an input signal X(n)
a power calculation unit 17 that calculates the power from the signal, and a comparator 18 that compares the output of the power calculation unit 17 with a threshold value Pth corresponding to a silent transmission path state and generates a 0-bit quantization instruction signal. The quantization bit number determining unit 14 outputs a 0-bit signal in response to the 0-bit quantization instruction signal.

Note that a parameter copy control section 4 is provided to avoid data discontinuity between frames.

Next, the operation of the embodiment of the present invention shown in FIG. 2 will be explained.

First, to clarify the basic idea of this embodiment, a plurality of adaptive predictive encoding units having adaptive quantizers with different numbers of quantization bits are operated in parallel, and the locally decoded signal of each adaptive predictive encoding unit is Evaluate the voice quality of.

This evaluation result is compared with a threshold value, and the one with the smallest number of quantization bits is selected from among the encoding units that satisfy the conditions. If all the encoders do not satisfy this threshold, the number of quantization bits that provides the best reproduced audio quality is selected. Also,
Silence detection is performed for each frame, and quantization of 0 bits is instructed in the silent section, so that the residual signal is not encoded, thereby increasing the compression effect.

Such processing is performed for each frame to switch the encoding speed, but in order to avoid discontinuities at frame boundaries, the quantization step size, prediction coefficients, and tap data of the selected encoding unit are transferred to other encoding units. Copying may also be done.

Specifically, adaptive predictive encoding units 2-1, 2-
^In 2, . That is, for example, the adaptive predictive encoding unit Q
1 and Q1 ^-1 realize 1-bit quantization, Q2 and Q2 ^-1 realize 2-bit quantization, ...Qk and Qk ^-1 realize k-bit quantization, so the update speed of each quantization step size is
M ₁ (I ₁ (n)), M ₂ (I ₂ (n)), ...M _k (I _k (n)) are correspondingly different, so for a 1-bit quantizer Q1 , there are 2 types of output codes, 4 types for the 2-bit quantizer Q2, and 2 ^k types for the k-bit quantizer Qk, so the update speed M ₁ (I ₁ (n)), M ₂ (I ₂ (n)),...M _k (I _k (n)
))
There are also 2, 4, ...2 ^k ways.

Locally decoded signals of each adaptive predictive coding unit having a different number of quantization bits, that is, reproduced output signals
As a measure of the audio quality of Xi'(n) (an evaluation measure of how well each encoder gives reproduced sound),
For example, two types can be considered: intra-frame S/N and intra-frame noise level.

First, when the intra-frame S/N is used as an evaluation function, the S/N evaluation function SNi between the input signal X(n) and the reproduced output signal Xi'(n) is expressed by the following equation (1).

SNi=−10log ₁₀ (Σ(X(i)−X
′(n)) ² /ΣX(i) ² )...(1) Also, the evaluation function N2 of the intra-frame noise level
is expressed by the following equation (2).

Ni=10log ₁₀ (1/kΣ(X(i)−X′(n)) ² …(2) For the values of these evaluation functions, the threshold value is adjusted based on the allowable transmission amount information from the transmission line side. circuit 1
2 sets the threshold value SNth or Nth.

Now, if S/N is used as an evaluation function, calculation unit 11
For each frame, evaluation functions SN1 and SN2 are calculated from the reproduced output signals X ₁ ′(n), X ₂ ′(n), ...X _k ′(n) of each adaptive predictive coding unit and the input audio signal X(n). ,…SNk
is calculated and compared with the threshold value SNth by the comparator 13. Comparator 13 selects a threshold value from among the above evaluation functions.
Instructs the quantization bit number determination unit 14 to select the one with the smallest number of quantization bits as the optimal quantization bit number l among those giving a value larger than SNth, thereby minimizing the encoding transmission rate. do. Conversely, if the evaluation functions SN1, SN2, ...SNk are all smaller than the threshold value SNth, the smallest one of them is selected as the optimal number l of quantization bits, and the quantization bits are set so as to maximize the encoding transmission rate. Instruct the number determining unit 14.

This is similarly executed when the noise level N is used as the evaluation function.

Also, the input signal power calculated by the input signal power calculation unit 17 is compared with the threshold value Pth set by the threshold adjustment circuit 12 corresponding to the silent state, and the input signal power Pin within the frame is small. In this case, it is assumed that there is no need to perform encoded transmission, and the quantization bit number determining unit 14 is instructed to quantize "0" bits.

The number of quantization bits for each frame is determined as described above, and the selection section 15 selects the quantization output of the corresponding adaptive predictive coding section according to the result. The multiplexer 16 multiplexes the data with auxiliary information indicating the number of bits.

In order to avoid data discontinuity between frames, a patent application filed by the present applicant on April 9, 1988 entitled "High-efficiency coding transmission device" has been filed.
It is necessary to control copying of parameters as disclosed in No. 80063. That is, the number l of quantization bits to be sent from the quantization bit number determination section 14 to the selection section 15
Copy control is performed to match all adaptive predictive coding units. That is, in each adaptive predictive coding section, the quantization step size Δl is applied to the quantizer and inverse quantizer.
Copy (n) and copy Hzl and Hpl to the zero predictor and polar predictor, respectively.

As a result, the quantization step size at the end of the frame of the adaptive encoder determined to be optimal for the current frame, and the quantization step size at the start of the relevant frame of the adaptive predictive encoder determined to be optimal for the next frame. will always match, and there will be no discontinuity in the reproduced signal at the frame boundary.

〔Effect of the invention〕

As described above, according to the encoding and transmitting apparatus of the present invention, the optimum quantization bit is determined according to the allowable transmission capacity using the voice quality evaluation function (S/N, noise level) of each adaptive predictive encoding unit. Since the code transmission speed is controlled by selecting the number, uniform reproduction quality can be provided in accordance with the redundancy of the amount of information in audio encoding, and the compression effect is improved.

[Brief explanation of drawings]

FIG. 1 is a principle block diagram of a coding/transmission device according to the present invention, FIG. 2 is a circuit diagram showing an embodiment of the coding/transmission device according to the invention, and FIG. 3 is a block diagram showing a conventional coding/transmission device. Figure. In Figures 1 and 2, 2-1, 2-2,
...2-k is an adaptive predictive coding unit, 3 is a transmission rate control unit, 4 is a parameter copying unit, Q1, Q2,...Qk
is a quantizer, Q1 ^-1 , Q2 ^-1 ,...Qk ^-1 is an inverse quantizer, 11 is an evaluation function value calculation unit, 12 is a threshold adjustment circuit, 13 and 18 are comparators, and 14 is a quantizer 15 is a selection unit; 16 is a multiplexing unit;
17 is an input signal power calculation section. In addition, in the figure,
The same reference numerals indicate the same or equivalent parts.

Claims (1)

[Claims] 1. A plurality of adaptive predictive coding units 2-1, 2-2, ... 2-k each having an adaptive quantizer and an inverse quantizer with different numbers of quantization bits, and The evaluation function value of S/N or noise level is calculated between the input signal and the reproduced output signal of each predictive encoding unit, and the a transmission rate control unit 3 that selects the quantized output of the adaptive predictive encoding unit and controls the transmission rate; and a transmission rate control unit 3 that selects the quantized output of the adaptive predictive encoding unit and controls the transmission rate; , a parameter copy control section 4 that controls copying of the quantization step size and adaptive prediction parameters of the selected adaptive predictive coding section to other adaptive predictive coding sections. Device. 2. The transmission rate control circuit 3 inputs the audio input signal and each of the adaptive predictive coding units 2-1, 2-2, 2-k.
a calculation unit 11 that calculates an intra-frame S/N evaluation function value with respect to the reproduced output signal; and a threshold adjustment circuit 1 that generates an S/N threshold according to the allowable transmission amount.
2, a comparator 13 that compares the S/N threshold and the S/N evaluation function value, and an S/N evaluation function value that gives a value larger than the S/N threshold in the comparison result. a quantization bit number determination unit 14 that selects the one with the smallest number of quantization bits among them, and selects the largest one when all the S/N evaluation function values are smaller than the S/N threshold; a selection section 15 that selects an adaptive predictive coding section corresponding to the number of quantization bits based on a signal indicating the number of quantization bits outputted from the deciding section 14; a multiplexing unit 16 that multiplexes the quantized output signal of the encoding unit;
An encoding and transmitting device according to claim 1, comprising: 3 The transmission rate control circuit 3 inputs the audio input signal and each of the adaptive predictive coding units 2-1, 2-2, 2-k.
a calculation unit 11 that calculates an intra-frame noise level evaluation function value with the reproduced output signal; a threshold adjustment circuit 12 that generates a noise level threshold according to the allowable transmission amount; and the noise level evaluation function value.
3, and select the one with the smallest number of quantization bits among the noise level evaluation function values that give a value smaller than the noise level threshold in the comparison results, and select the one with the smallest number of quantization bits, so that all the noise level evaluation function values are equal to the noise level. A quantization bit number determining section 14 selects the minimum number when it is larger than a threshold value, and an adaptive predictive coding section corresponding to the quantization bit number based on a signal indicating the quantization bit number output from the determining section 14. a selection unit 15 that selects the number of bits to be quantized; a multiplexing unit 16 that multiplexes the signal indicating the number of quantized bits and the quantized output signal of the selected adaptive predictive coding unit;
An encoding and transmitting device according to claim 1, comprising: 4 The transmission rate control circuit 3 inputs the audio input signal and each of the adaptive predictive coding units 2-1, 2-2, 2-k.
a calculation unit 11 that calculates an intra-frame S/N evaluation function value with respect to the reproduced output signal; and a calculation unit 11 that generates an S/N threshold value corresponding to an allowable transmission amount and a power threshold value corresponding to a silent state. a threshold adjustment circuit 12; a comparator 13 that compares the S/N threshold and the S/N evaluation function value; and an S/N that provides a value larger than the S/N threshold in the comparison result. a quantization bit number determination unit that selects the one with the smallest number of quantization bits among the evaluation function values, and selects the largest one when all the S/N evaluation function values are smaller than the S/N threshold; 14, an input signal power calculation unit 17, a comparator 18 that compares the input signal power with a threshold corresponding to the silent state, and a signal indicating the number of quantization bits output from the determination unit 14. a selection unit 15 that selects an adaptive predictive coding unit corresponding to the number of quantized bits; and a multiplexing unit that multiplexes a signal indicating the number of quantized bits and a quantized output signal of the selected adaptive predictive coding unit. Part 16
The determining unit 14 determines the number of quantization bits to be 0 when the comparator 18 determines that the input signal power is smaller than the threshold value corresponding to the silent state. An encoding transmission device according to claim 1. 5 The transmission rate control circuit 3 inputs the audio input signal and each of the adaptive predictive coding units 2-1, 2-2, 2-k.
a calculation unit 11 that calculates an intra-frame noise level evaluation function value with the reproduced output signal; and a threshold that generates a noise level threshold corresponding to an allowable transmission amount and a power threshold corresponding to a silent state. Adjustment circuit 1
2, a comparator 13 that compares the noise level threshold and the noise level evaluation function value, and quantization among the noise level evaluation function values that give a value smaller than the noise level threshold in the comparison result. a quantization bit number determination unit 14 that selects the minimum number of bits, and selects the minimum number if the noise level evaluation function values are all greater than the noise level threshold; and an input signal power calculation unit 17. , a comparator 18 that compares the input signal power with a threshold value corresponding to the silent state, and a signal indicating the number of quantization bits output from the determining section 14 to perform adaptive prediction corresponding to the number of quantization bits. a selection section 15 that selects an encoding section; and a multiplexing section 16 that multiplexes a signal indicating the number of quantization bits and a quantized output signal of the selected adaptive predictive encoding section.
The determining unit 14 determines the number of quantization bits to be 0 when the comparator 18 determines that the input signal power is smaller than the threshold value corresponding to the silent state. An encoding transmission device according to claim 1.