JP3143359B2 - Audio coding method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a highly efficient speech coding method, and more particularly to an adaptive pulse code modulation (Adaptive Pulse Code Modulation).
Modulation, hereinafter referred to as “APCM”. ) Method and Adaptive Differential Pulse Code Modulation
e Code Modulation, hereinafter referred to as "ADPCM". 2.) Improvement of the method.

[0002]

2. Description of the Related Art ADPCM is used as a voice band compression method.
There is a way. This method between voice adjacent samples, for example in the audio data in the time t ₁ and time t _2, takes the difference between the audio signals in the prediction value and time t ₂ was calculated to time t _1,
The difference is coded into an ADPCM code to compress the voice, and then the code is decoded to obtain a quantized value of the difference signal, and the values are sequentially added to form a normal PCM code format. This is a method of reproducing sound. Further, the ADPCM method is characterized in that a quantization width required for obtaining a quantization value of a difference signal is appropriately changed according to an ADPCM code.

[0005] FIG. 5 is a diagram showing an A method for implementing the conventional ADPCM method.
FIG. 2 is a schematic configuration diagram of a DPCM encoding device 4 and an ADPCM decoding device 5, and functions of each component will be sequentially described below. It is assumed that n used below is an integer.

[0004] The first adder 41 is an ADPCM encoder 4
Input signal x _n difference d _n of the predicted signal y _n, the number 1

[0005]

(Equation 1)

[0006] It is determined according to:

[0007] The quantizer 42 number difference d _n obtained by the first adder 41 is 2, and the number 3

[0008]

(Equation 2)

[0009]

(Equation 3)

[0010] Based on the appropriate quantization width delta _n in accordance,
It has a function of obtaining the quantization values q _n.

Next, the encoder 43 encodes the quantized value q _n and outputs a code L _n (L _n = A _n ) to the memory 6.

The first quantization width updater 44 calculates

[0013]

(Equation 4)

The quantization width is adaptively changed / updated according to the above, and the quantization value Δn _{+ 1} is sent to the quantizer 42 for use in the next quantization.

Here, Table 1 shows a multiplier M used in Equation 4.
It is a table showing the relationship between (L _n ) and the symbol L _n .

[0016]

[Table 1]

On the other hand, the second adder 45 calculates

[0018]

(Equation 5)

A decoded signal w _n is obtained according to the following formula, and the decoded signal w _n is sent to the first predictor 46.

The first predictor 46 obtains the next predicted signal y _{n + 1} by delaying the decoded signal w _n by one sample, and this predicted signal y _{n + 1} is sent to the first adder 41.
The processing after the first adder 41 is the above described repetition.

On the other hand, the decoder 51 of the ADPCM decoding device 5 has

[0022]

(Equation 6)

[0023] outputs the quantized value q _'n in accordance with.

The second quantization width update 52 reads the code L _n from the memory 6 and calculates

[0025]

(Equation 7)

[0026] The quantization width delta _n changed or updated in accordance with this quantization width delta _{n + 1} is sent to the decoder 51, it is used for subsequent decoding. The value of M (L _n ) is as shown in Table 1.

The third adder 53 is given by the following equation (8).

[0028]

(Equation 8)

[0029] 'seek _n, the decoded signal w' w according _n is sent to the second predictor 54.

The second predictor 54 decoded signal w 'is delayed _n by 1 sample next prediction signal y' seek _{n + 1,} the prediction signal y _{'n + 1} is sent to the third adder 53 Can be

[0031] Next Figure 6 shows the relationship of the difference d _n of quantization values q _n when changing in accordance with the quantization width delta _n in ADPCM code, and the input signal x _n and the predicted signal y _n FIG. It is assumed that T <U in FIG.

Here, regarding the difference dn, _assuming that "[" and "]" include the boundary value in the range, and "(" and ")" do not include the boundary value in the range. 6 (a)
In case the value of the difference d _n is in the range of [0, T] is T /
2, 3T / 2 when in the range of (T, 2T),.
.., (3T, ∞), quantization is performed to 7T / 2.

When it is in the range of [-T, 0),-
-3T when T / 2 is in the range of [-2T, -T)
/ 2,..., [-∞, -3T), the quantization is performed to -7T / 2.

On the other hand, in FIG. 6B, similarly, when the value of the difference d _{n is in} the range of [0, U], (U, 2)
U], 3U / 2,..., (3U,
∞], it is quantized to 7U / 2. Also, when it is in the range of [-U, 0), -U / 2 and [-
2U, -U), -3U / 2, ...
, [-∞, -3U), quantization to -7U / 2 is performed.

[0035] Thus, by the time difference d _n is small, finely set the quantization width, when the difference d _n conversely large to roughly set the quantization width as shown in FIG. 6 (b), the efficiency Quantum Can be

[0036]

[SUMMARY OF THE INVENTION] However, since in the conventional art, when the abrupt change to a smaller value from the absolute value is larger the value of the difference d _n is stuck in the quantization width is large values, Since the minimum absolute value of the quantized value q _n is as large as U / 2 (see FIG. 6B), | d _n | ≪ (U /
In the case of 2), the quantization error increases.

[0037] In addition, the value of the difference d _n in the above prior art 0
Even in the case of, if quantization is performed, T / 2 (see FIG. 6A).
Or U / 2 (see FIG. 6B).
The quantization value is no longer 0. Further, in the silent section of the audio signal often value of the difference d _n is 0, there is a drawback that quantization error increases.

On the other hand, if the APCM process, for an input signal in which it is the difference d _n, has the same drawbacks as ADPCM method.

[0039] The present invention has been made in view of the problems described above, the absolute value abruptly changes to a smaller value from the larger value of the difference _{d n | d n | «(} T / 2), Or | d
_n | «while reducing the quantization error when it becomes (U / 2) and the value of the difference d _n is suppress the occurrence of the quantization error of 0, it is provided a novel speech coding method is there.

[0040]

In coding apparatus which realizes a speech encoding method of the means for solving problems] The present invention, a difference d _n between the predicted value y _n of the input signal x _n and the input signal x _n of speech, quantizing a quantizer for performing a quantization on the basis of the width delta _n, and encoder for encoding the code L _n quantized values q _n by quantization unit, said code L _n
Based on, update the quantization width delta _n, and the first quantization width updater to send this updated quantization width delta _{n + 1} to the quantizer, the quantization value by the quantizer q _n , and based on the predicted value y _n of the input signal, first determining the next prediction signal y _{n + 1}
A predictor, and a code L _n generated by the coding apparatus.
When one or more codes L _n among the above are decoded, quantization is performed so that the decoded value q ′ _n of the difference d _n becomes zero.

[0041]

DETAILED DESCRIPTION OF THE INVENTION The present invention By providing the means, the input signal x _n and the input signal x _n and the predicted value y _n of the input signal x _n
Difference d the absolute value of _n when the abrupt change to a small value from a large value, but the quantization width delta _n remains large value, decoding of the input signal x _n and the difference d _n at that time with By performing quantization so that the value can be set to 0, a quantization error generated can be reduced. That is, | d _n | ≪ (Δ _n /
During 2) is the quantized value of the difference d _n may be set to 0.

Further, even when the value of a signal which is frequently seen in a silent section of an audio signal is 0, the quantization value can be set to 0 so that the occurrence of a quantization error can be prevented.

[0043]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to FIGS.

FIG. 1 is a schematic block diagram of an ADPCM encoding device and an ADPCM decoding device for realizing the speech encoding method of the present invention. It is assumed that n used below is an integer.

[0045] First, the first adder 11 and the signal x _n input to ADPCM encoder a difference d _n of the predicted signal y _n

[0046]

(Equation 9)

Is determined according to

Next, the second adder 12 calculates

[0049]

(Equation 10)

[0050] determine the adjustment signal e _n accordance with.

The second adder 12 is given by the following equation (11).

[0052]

[Equation 11]

[0053] determine the adjustment signal e _n accordance with.

The quantizer 13 is represented by Expressions 12 and 13

[0055]

(Equation 12)

[0056]

(Equation 13)

The quantization value q _n is obtained according to the following.

The encoder 14 outputs L _n of Table 1 as a code to the memory 3 in accordance with the quantized value q _n (Table 1 is the same as the prior art and will not be described here).

The first quantization width updater 15 calculates

[0060]

[Equation 14]

[0061] adaptively modify and update the quantization width delta _n in accordance, the quantized value delta _{n + 1} is used during the next quantization.

The third adder 16 is given by equation (15).

[0063]

(Equation 15)

A decoded signal w _n is obtained according to the following formula, and the decoded signal w _n is sent to the first predictor 17.

The first predictor 17 obtains the next predicted signal y _{n + 1} by delaying the decoded signal w _n by one sample, and this predicted signal y _{n + 1} is sent to the first adder 11.

On the other hand, the decoder 21 of the ADPCM decoding device 2 has

[0067]

(Equation 16)

And outputs the quantized value q _'n in accordance with [0068].

The second quantization width update 22 reads the code L _n from the memory 3 and

[0070]

[Equation 17]

[0071] The quantization width delta _n changed or updated in accordance with this quantization width delta _{n + 1} is sent to the decoder 21, it is used for subsequent decoding. The value of M (L _n ) is as shown in Table 1.

The fourth adder 23 calculates

[0073]

(Equation 18)

'Seek _n, the decoded signal w' w according [0074] _n is fed to the second predictor 24.

The second predictor 24 delays the decoded signal w ′ _n by one sample to thereby generate the next predicted signal y ′.
_{n + 1} is obtained, and the prediction signal y'n _{+ 1} is sent to the fourth adder 23.

The operation of the ADPCM encoding apparatus 1 having the above configuration will be described with reference to the flowchart of FIG.

[0077] In step S10, subtracts a prediction signal y _n from the input signal x _n, obtains a difference d _n. In step S11, it is determined whether the difference dn obtained in step S10 is a positive number or a negative number. _If the difference dn is a positive number, the process proceeds to step S12. If the difference dn is a negative number, the process proceeds to step S12. Proceed to S13.

In step S12, the value obtained in step S10 is obtained.
Difference d_nTo the quantization width Δ_nOf the adjustment signal e
_n, And the process proceeds to step S14. Meanwhile, step
In S13, the difference d obtained in step S10_nQuantized to
Width Δ_nＥ of e _n, And then step S
Proceed to 14.

In step S14, equations (12) and (13)
After performing quantization and encoding in accordance with
Go to 5. In step S15, the code L _n obtained in step S14, and after the updating of the quantization width delta _n based on the quantization width delta _n, the process proceeds to step S16.

Finally, in step S16, the predicted values y _n ,
Then, the next predicted value y _{n + 1} is obtained by using the quantization value q _n .

Next, FIG. 3 shows the relationship of the difference d _n of quantization values q _n when changing in accordance with the quantization width delta _n in ADPCM code, and the input signal x _n and the predicted signal y _n FIG. It is assumed that T <U.

[0082] Here, when the prior art as well as looking at the difference d _n, the value of the adjustment signal e _n in FIG. 3 (a) (-0.5
T, 0.5T], it becomes 0, and (0.5T, 0.5T,
[1.5T], T,..., (2.5
T, ∞], it is quantized to 3T.

Also, when it is in the range (-1.5T, -0.5T), it is -T, when it is in the range (-2.5T, -1.5T), it is -2T, ... .., [-∞, -3.5T], it is quantized to -4T.

[0084] On the other hand, adjustment similarly in FIG. 3 (b) signal e _n
Is within the range of (-0.5U, 0.5U).
To 1U when in the range (0.5U, 1.5U),
..., (2.5U, ∞) is quantized to 3U.

Also, when it is in the range of (-1.5U, -0.5U], it is -U, when it is in the range of (-2.5U, -1.5U), it is -2U, ... .., (-∞, -3.5U], the quantization is to -4U.

Next, FIG. 4 is a flowchart of the processing performed by the ADPCM decoding device 2.

[0087] At step S20, the decoder 21 of the ADPCM decoder 2 reads the code L _n of the memory 3, obtains a quantized value q _'n from the code L _n, and the quantization width delta _n according to equation 16 The process proceeds to step S21. Step S
In 21 obtains the _{n + 1} 'next prediction signal y using the _n' quantization values q calculated in step S20, the process proceeds to step S22. Finally it updates the quantization width based on the code L _n at step S22.

[0088]

As is clear from the above description, in the present invention, the input signal x _n , or the input signal x _n and the input signal x _n are quantized so that the decoded signal can be zero. abruptly changes to a smaller value from the absolute value is larger the value of the difference d _n between the predicted value y _n of _n, even when it remains high value quantization width delta _n, by the quantized values 0 In addition, it is possible to reduce a quantization error that has conventionally occurred.

[0089] Further, an effect that the quantization error value is also quantized when the input signal x _n and the difference d _n is 0 becomes 0 is not generated.

[Brief description of the drawings]

FIG. 1 is an ADPCM that implements the speech encoding method of the present invention.
It is a schematic structure figure of an encoding device and an ADPCM decoding device.

FIG. 2 is a flowchart of an ADPCM encoding device of the speech encoding method of the present invention.

FIG. 3 is a conceptual diagram of optimal quantization used in the speech encoding method of the present invention.

FIG. 4 is a flowchart of an ADPCM decoding device of the speech encoding method of the present invention.

FIG. 5 is a block diagram of a conventional ADPCM method.

FIG. 6 shows a quantization value q _n , an input signal x _n, and a prediction signal y when the quantization width Δ _n is changed according to the ADPCM code.
It is a diagram showing the relationship of the difference d _n with _n.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 ... ADPCM encoder 13 ... Quantizer 14 ... Encoder 15 ... 1st quantization width updater 17 ... 1st predictor 2 ... ADPCM decoder 21 ... Decoder 22 ... Second quantization width updater 24 ... Second predictor 3 ... Memory

Continuation of the front page (58) Field surveyed (Int.Cl. ⁷ , DB name) H03M 3/04 H03M 7/36

Claims (2)

(57) [Claims] 1. A quantizer which calculates a quantized value q _{n of} an input signal x _{n by} a quantizer.
APCM for determining and quantizing the quantized value q _n with an encoder
A coding method, an input signal x _n of the audio input to the APCM encoding device
At the time of quantization by a quantizer, when the input signal x _n ≧ 0,
Obtains an adjustment signal _{e n = x n + Δ n} / 2 of the quantization values q _n, also
If the input signal x _n <0, the adjustment signal _{e n = x n -Δ n /} 2
A first step of obtaining the quantized values q _n of the quantization values q _n of the first step by encoding the code L _n
The quantization step Δn _{+ 1} is calculated based on the second step to be obtained and the code L _n by the second step.
Calculate and send the quantization width Δn _{+ 1} to the quantizer.
A step, the quantized values q _{n + 1} of the adjustment signal e _{n + 1} by the quantization width delta _{n + 1}
And a fourth step of determining . Wherein the prediction value y of the input signal x _n and the input signal x _n
The quantized value q _n of the difference d _n between the _n determined by the quantizer, quantum
ADPCM coding method for coding in the encoder the reduction value q _n
And an audio input signal x _n input to the ADPCM encoding apparatus.
And the difference d _n between the predicted value y _n of the input signal x _n and the
When the difference d _n ≧ 0, the adjustment signal e _n = d
seek _n + _{Δ n} / 2 of the quantization values q _n, also place the difference d _n <0
Determining if the adjustment signal _{e n = d n -Δ n /} 2 of the quantization values q _n
A first step, the quantized values q _n of the first step by encoding the code L _n
The quantization step Δn _{+ 1} is calculated based on the second step to be obtained and the code L _n by the second step.
Calculate and send the quantization width Δn _{+ 1} to the quantizer.
A step, the quantization value q _n, and based on the predicted value y _n, next prediction
A fourth step of obtaining a signal y _{n + 1,} the quantization value q _{n + 1} of the adjustment signal e _{n + 1} by the quantization width delta _{n + 1}
And a fifth step of obtaining.