JP2002123298A - Method and device for encoding signal, recording medium recorded with signal encoding program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for masking hearing sensation highly accurately by precisely estimating the position of a mountain and a valley in a spectral envelope curved line when vocal and musical sound signals are encoded. SOLUTION: The device is provided with a T/F converter 11 which performs a time-axis/frequency-axis conversion to an input signal to obtain a coefficient sequence X on a frequency-axis (n), an envelope calculation part 13 which calculates the spectral envelope on the basis of the coefficient sequence X (n), a mountain and valley estimation part 14 which estimates the position of the mountain and the valley in the spectral envelope, a weighting part 15 which performs weighting of amount of information at the position of the estimated mountain and valley in the spectral envelope, a hearing sensation weight calculation part 16 which calculates hearing sensation weight for quantization on the basis of the spectral envelope subjected to weighting of amount of information, and a quantization part 17 which performs quantization to the coefficient sequence X (n) on the basis of the hearing sensation weight for quantization.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a signal encoding method and apparatus for performing quantization by transforming an input signal on a time axis / frequency axis. The present invention relates to an auditory masking method for deforming the audio signal so as to be hard to perceive, and a signal encoding device using the auditory masking method.

[0002]

2. Description of the Related Art As an auditory masking method in a conventional signal encoding method for encoding voices and musical sounds, an input signal is converted on a time axis or a time axis / frequency axis, and the input signal is input by a linear prediction analysis method or the like. There has been a method of estimating a spectral envelope curve of a signal, performing a proper deformation operation on the estimated curve, obtaining a masking curve, and performing auditory masking. Alternatively, there is also a method in which a spectrum envelope curve is directly obtained from a signal obtained by converting an input signal into a time axis / frequency axis, and a masking curve is obtained by applying a proper deformation operation to the curve to perform quantization by auditory masking. .

[0003]

In the auditory masking method, as the masking on the frequency axis, the quantization noise near the valley of the spectrum envelope curve is reduced, and the quantization noise near the peak of the spectrum envelope curve is increased instead. By performing such noise shaping, it is possible to make the quantization noise inaudible to the human ear. Here, in the conventional method as described above, since the estimated positions of the peaks and valleys in the spectral envelope may be inaccurate, noise shaping is not properly performed, and as a result, the sound quality of the encoded reproduced sound is poor. was there.

Accordingly, an object of the present invention is to accurately estimate the positions of peaks and valleys in a spectral envelope curve,
Accordingly, it is an object of the present invention to provide a signal encoding method and apparatus capable of executing a highly accurate auditory masking method.

[0005]

SUMMARY OF THE INVENTION The present invention is for realizing signal encoding that can be quantized so that distortion on an auditory basis is minimized. A method of accurately estimating the positions of the peaks and valleys of the envelope curve and performing appropriate noise shaping from the accurately estimated positions of the peaks and valleys is employed. The position estimation of the peaks and valleys of the spectrum envelope curve is performed by removing fine irregularities as necessary from the accurate spectrum envelope curve of the signal subjected to the time-axis / frequency-axis conversion, and further performing first-order differentiation and second-order differentiation as necessary. And the exact positions of the peaks and valleys are determined from these differential values or the arithmetic mean of the differential values. Appropriate weighting is performed at the positions of the peaks and valleys thus obtained, and effective noise shaping is realized.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Next, a preferred embodiment of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram illustrating a configuration of a signal encoding device according to an embodiment of the present invention.

This signal encoding apparatus performs time / frequency axis conversion (T / F conversion) on a time-series input signal x (t), which is typically a voice signal or a tone signal, and performs frequency axis conversion. A T / F converter 11 for obtaining the above signal sequence X (n), and a quantum for obtaining a quantization index by performing vector quantization (VQ) and scalar quantization (SQ) on the signal sequence X (n). It has a conversion unit 12. Here, the T / F converter 11
Is, for example, MDCT (modified descrete cosine tra
nsform; modified discrete cosine transform)
X (n) indicates a conversion coefficient sequence or the like obtained by this conversion. Further, this signal encoding apparatus calculates an “auditory weight” for determining how much information amount is to be allocated to which frequency band, and when the quantization is performed by the quantization unit 11, the quantization is performed by the human ear. Perceptual weighting quantization based on this perceptual weight is performed so that noise is hard to hear. For the calculation of the auditory weight, this signal coding device
An envelope calculating unit 13 that calculates a spectrum envelope based on the signal sequence X (n), and a peak / valley estimating unit 14 that estimates the positions of peaks and valleys of the spectrum based on the calculated spectrum envelope.
Based on the estimated peak and valley positions of the spectrum, the distribution of the information amount is appropriate in the vicinity of the peak and the valley so that the distribution of the information amount is “particularly small at the peak position” and “particularly large at the valley position”. Weighting unit 15 for performing an appropriate weighting, and an auditory weight calculating unit 16 for outputting to the quantizing unit 12 as “auditory weight”
And Here, the reciprocal of the spectral envelope is used as the original form of the “auditory weight”.

As for the peaks and valleys, the signal sequence X (n) is plotted with the horizontal axis as the frequency axis, and when the signal sequence X (n) is averaged (smoothed), a portion where the value of the signal sequence is larger than the surroundings is indicated as a peak. The area where the value is smaller than the surrounding area is called a valley. As described later, the smoothing is performed by, for example, calculating an arithmetic mean (calculating a moving average based on the section length) in a certain section length (also referred to as an average section length). By changing the section length, fine peaks / valleys, slightly fine peaks / valleys, rough positions of peaks / valleys, and the like are estimated. Here, the arithmetic averaging is for smoothing a spectrum in one frame in a frequency section. In the present invention, by combining the estimation of the positions of the peaks and valleys with different degrees of smoothing,
This enables more accurate auditory masking.

Next, the operation of the signal encoding apparatus will be described.

A time-series input signal x (t) input as a time-series signal is converted by a T / F converter 11 into a signal sequence X (n) on the frequency axis. The signal sequence X (n) is supplied to the quantization unit 12 for vector quantization and scalar quantization, and is also sent to the envelope calculation unit 13 to calculate the spectrum envelope. Envelope calculator 13
Calculates the spectral envelope of the signal sequence X (n), and the peak / valley estimating unit 14 estimates the positions of the peaks and valleys in the spectrum based on the calculated spectral envelope, and weights the estimated position to a weighting unit 15. Output to The weighting unit 15 calculates the peak and valley positions of the spectrum based on the reciprocal of the spectrum envelope obtained by the envelope calculation unit 13, respectively.
Appropriate information amount weighting is performed in the vicinity of the mountain and in the vicinity of the valley so that the distribution of the information amount is “particularly small at the position of the mountain” and “particularly large at the position of the valley”. Specifically, the position of the peaks and valleys is raised using a weight function that raises the vicinity of the peaks high and lowers the vicinity of the valleys deeply, or lowers the vicinity of the peaks and raises the vicinity of the valleys so as to be shallow. Weighting operation. The weighting unit 15 is supplied with a spectrum envelope curve from the envelope calculation unit 13, and the weighted operation is performed to supply the weighted spectrum envelope curve to the auditory weight calculation unit 16.

The auditory weight calculator 16 calculates an auditory weight for quantization based on the weighted spectral envelope curve, and outputs it to the quantizer 12. As a result, the quantization unit 13 performs vector quantization and scalar quantization on the signal sequence X (n) from the T / F conversion unit 11 using the supplied auditory weights for quantization. As a result, the quantization unit 13 outputs a quantization index (output index) on which high-precision auditory masking has been performed.

Although the basic operation of the signal encoding apparatus according to this embodiment has been described above, in the present invention, the above-described weighting method and a conventionally used linear weighting method are used as auditory weighting methods. A method may be used in which a spectral envelope is predicted by a prediction analysis method or the like, and peaks and valleys of the envelope curve are rounded by exponentiation operation and weighted.

Next, the weighting process in this embodiment will be described in detail.

FIG. 2 is a block diagram showing a process of weighting peaks and valleys of a spectrum. Here, from the spectrum envelope curve obtained by the spectrum envelope calculator 13, the peak / valley estimator 14 estimates the frequency position of the fine peak / valley of the spectrum, and then the frequency position of the slightly fine peak / valley. This procedure is repeated as many times as necessary, and finally, a rough peak of the spectrum is obtained.
Estimate the frequency position of the valley. The weighting unit 15 performs a weighting operation on each of the estimated peaks and valleys by using an appropriate weighting function.

FIG. 3 is a block diagram showing details of the processing in the envelope calculation unit 13. The envelope calculation unit 13 obtains a spectrum envelope curve by performing arithmetic averaging processing on the signal sequence X (n) in the frequency domain. In the figure, arithmetic averages (1) to (k) are arithmetic averages in moving average sections having different section lengths. Here, first, the first arithmetic mean (1) is applied to the signal sequence X (n), and as a result, the second arithmetic average (1) is applied to Y ₁ (n).
The arithmetic mean (2) is applied, and as a result Y ₂ (n)
, The third arithmetic mean (3) is applied, and the arithmetic average of k times is sequentially performed. Here, k is an integer constant of 1 or more. The results of each arithmetic mean Y ₁ (n), Y ₂ (n),.
_k (n) is sent to the peak / valley estimating unit 14, respectively. The section length in each arithmetic averaging is determined according to each application, but mainly in the arithmetic averaging (1), the average section length is shortened to detect fine peaks and valleys. In the arithmetic averaging (2), the average section length is made longer than in the arithmetic averaging (1), and rough peak and valley positions are detected. Hereinafter, the same operation is performed up to the arithmetic averaging (k), and the average section length in each arithmetic averaging may be gradually increased. Further, the arithmetic operation of the above-described “arithmetic average (k)” may be performed a plurality of times by changing the average section length as needed.

Here, the value of k and the section length at each arithmetic mean will be described. k may be an integer of 1 or more, but is typically 2 or 3. The input signal is a normal audio signal, and the sampling frequency of the input signal is 16 kHz.
When the z and the frame length are 60 ms, the average section length of the arithmetic average (1) is about 200 μs, and the arithmetic average (2)
Is preferably about 1 ms, and the average section length of the arithmetic mean (3) is about 10 ms.

Next, the processing in the peak / valley estimating unit 14 will be described. FIG. 4 is a block diagram for explaining processing in the peak / valley estimating unit 14.

The peak / valley estimating unit 14 calculates a coefficient sequence Y representing the spectral envelope by the arithmetic averaging from each time from the envelope calculating unit 13.
_{Using 1} (n), Y ₂ (n),..., Y _k (n) as inputs, the positions of peaks and valleys are estimated for each coefficient sequence as follows. That is, the input coefficient sequence Y _j (n) (1 ≦ j ≦ k) is first converted to n
In differentiating 'seek _j (n), the sequence Y' series Y to take an arithmetic mean with the appropriate interval relative to _j (n), sequence obtained by removing the fine fluctuation component

[0019]

[Outside 1]

Is obtained. Further, this is differentiated again by n to obtain a series Y ″ _j (n), and a series Y ′ _j (n) obtained by removing minute fluctuation components from the series Y ″ _j (n)

[0021]

[Outside 2]

Is obtained. Then, as shown by the equation in FIG. 4, the positions of the peaks and valleys of the spectrum envelope curve are estimated from the positive / negative of these values. In addition, the above-described arithmetic operation of “arithmetic averaging” for removing minute fluctuation components may be performed a plurality of times by changing the average section length as necessary, or may not be performed. .

FIG. 5 is a diagram illustrating a manner in which peaks and valleys of the spectral envelope are detected from the coefficient sequence X (n) as described above. Here, k = 2, that is, the envelope calculation unit 13
Shows a case in which the arithmetic mean is obtained in two stages.
In this figure, the absolute value | X (n) | of the coefficient sequence X (n) before taking the average is represented by a series Y ₁ (n) of arithmetic mean (1).
The absolute value of | Y ₁ (n) | a, the absolute value of the arithmetic mean coefficient due (2) column _{Y 2 (n) | Y 2} (n) | and the. The position of the mountain estimated from the arithmetic mean (1) is m ₁ ,
m _2, ..., m _12, the position of the valley V _1, V _2, ..., expressed in V _11, M ₁ to position mountain estimated from the arithmetic mean _(2), M 2,
M ₃ and the position of the valley are represented by V ₁ and V ₂ . Here, the section length in the arithmetic mean (2) is longer than the section length in the arithmetic mean (1), and corresponds to a fine peak / trough frequency position. Frequency position.

Next, assuming that a plurality of types of frequency positions of peaks and valleys have been obtained in this manner, how to weight information amounts will be described. FIG. 6 is a diagram illustrating an example in which information amounts are weighted near peaks and valleys of a spectrum envelope curve. Here, for simplicity of explanation,
The explanation is given using rough waveforms.

In FIG. 6, the reciprocal (1 / | Y ₂ ) of the spectral envelope curve (| Y ₂ (n) |) estimated in advance is shown.
(n) |) is used as the original form of the auditory weight, and weighting is performed using a weight function in the vicinity of the estimated positions of the peaks and valleys. In the example of this figure, the auditory weight (W) in which the information amount is corrected at the positions of the peaks and valleys by multiplying the weighting function by
_L ) have created. The weighting function and
Although various shapes are possible, here, as an example, the section length to be weighted is 2t, and 0.5 is set at the center of the mountain.
The result is weighted by a linear function such that the magnification is 1.0 times at the edge of the valley, 2.0 times at the center of the valley and 1.0 times at the edge of the valley. As can be seen from FIG. 6, it is possible to estimate the exact positions of the peaks and valleys, create a weight that increases the amount of information near the valley, and allocates a smaller amount of information near the peak.

Here, the value of t is, for example, 100 to 2 when it is desired to weight the peak / trough structure representing the pitch frequency.
When it is desired to weight the peak and valley structure representing 00 Hz and the formant frequency, the frequency is preferably about 300 to 600 Hz.

Actually, weighting is performed by the above-described method in the vicinity of the peaks and valleys of the “fine curve” and “rough curve” of the spectral envelope. For example, FIG.
As shown in (1), when the positions of the peaks and valleys are estimated for each of the “fine curve” and the “rough curve” of the spectral envelope, the reciprocal 1 / | Y ₁ (n ) | Is the original form of the auditory weight, and in the vicinity of the peak and valley positions m ₁ , v ₁ , m ₂ , v ₂ ,... Of this envelope curve, the original form of the auditory weight is 1 as in FIG. / | Y
₁ (n) | is appropriately weighted, and in the vicinity of the peak and valley positions M ₁ , V ₁ , M ₂ , V ₂ ,... Of the curve representing the rough spectral structure, Appropriate weighting is performed on 1 / | Y ₁ (n) | which is the original form of the auditory weight.

Various functions can be considered as the weighting function for the peak and the weighting function for the valley. FIG.
Exemplifies such a weighting function.

In FIG. 7, (a) and (b) each show an example of a weighting function for a mountain, where (a) is composed of a straight line and (b) is composed of a parabola. is there.
In each case, a section of a total of 2t is set as a weighted section, with t on each side of the center n = M of the mountain. It is assumed that the value of the weighting function is 1.0 at both ends (M ± t) of the weighting section. Also, the value α of the weight at the center of the mountain n = M
Is usually a reasonable constant at 0 <α <1.0. Similarly, in FIG. 7, (c) and (d) show examples of the weighting function for the valley, where (c) is formed by a straight line and (d) is formed by a parabola. As in the case of the peak, the value of the weighting function for the valley is 1.0 at both ends (V ± t) of the weighting section.
The weight value β at the center of the valley n = V is usually β
Use a reasonable constant at> 1.0. However, in some cases, setting α> 1.0 and 0 <β <1.0 may be effective.

When the auditory weighting is performed in this way, the quantization noise is transformed as shown in FIG. That is, when the auditory weighting is not performed, the quantization noise is considered to be constant irrespective of the frequency (in the figure), but if the spectrum envelope of the input signal is as shown in the figure, By performing the above-described auditory weighting, the noise has its frequency characteristics deformed as shown in the figure and is hidden by the spectral characteristics of the input signal, making it difficult to hear audibly.

Therefore, auditory masking can be performed with higher accuracy than the conventional method, and high-quality coding can be performed.

Next, an example in which the above-described signal encoding method of the present invention is applied to auditory weighting of a general transform encoding method will be described. FIG. 9 shows the configuration of a signal encoding device that performs such auditory weighting.

The signal encoding apparatus shown in FIG. 9 includes an MDCT conversion unit 31 that performs MDCT on an input signal,
A spectrum flattening unit 32 for flattening the spectrum of the subsequent signal, a frame gain normalizing unit 33 for normalizing and quantizing the frame gain based on the flattened spectrum, and then outputting a gain index; The residual component is quantized (vector quantization or scalar quantization) based on the obtained frame gain, and a residual component quantization unit 34 that outputs a quantization index, and a spectral envelope is estimated from a signal spectrum after MDCT. A spectral envelope estimating unit 35, an auditory weight calculating unit 36 that calculates an auditory weight from the estimated spectral envelope to perform information weighting at the time of quantization in the residual component quantizing unit 34,
A spectrum information quantization unit 37 for quantizing the spectrum information based on the estimated spectrum envelope and outputting a spectrum index. In this signal encoding device, the MDCT conversion unit 31 corresponds to the T / F conversion unit 11 of the signal encoding device shown in FIG. 1, and the spectrum envelope estimating unit 35 is an envelope calculating unit of the device shown in FIG. 13 and the peak / valley estimating unit 14, and the hearing weight calculating unit 36 includes the weighting unit 15 and the hearing weight calculating unit 16 of the device shown in FIG. 1.
It consists of.

According to the signal encoding method of the present invention, peaks and valleys of a spectrum in an analysis frame can be accurately and finely analyzed, and highly accurate auditory masking can be performed at the time of quantization according to the shape. This auditory masking
It can be applied to vector quantization and subband scalar quantization.

FIG. 10 shows an example in which the auditory weighting of the present invention is applied to the encoder and decoder disclosed in Japanese Patent Application Laid-Open No. 8-44399. In the configuration shown in FIG. 10, encoder 110 includes a frame dividing section 114 for dividing an input signal applied to input terminal 111 into frames.
And a time windowing unit 115 for drawing a time window on the frame, and an MD for performing an N-order MDCT on the frame on which the time window is set
A CT unit 116, a linear prediction analysis unit 117 that performs a linear prediction analysis on a frame to which a time window is applied and outputs a prediction coefficient, a quantization unit 118 that quantizes the prediction coefficient to obtain an index I _p , A spectrum outline calculation unit 121 for obtaining a spectrum outline of a coefficient, and an MDCT unit 11
6, a normalization unit 122 that obtains a residual coefficient R (F) by normalizing the spectrum amplitude from the spectrum amplitude by a spectrum outline, a residual outline calculation unit 123 that calculates a residual coefficient outline E _R (F), A weight calculator 124 for calculating a weighting coefficient (vector W) based on the difference coefficient outline and the spectrum outline, and a quantum for quantizing based on the weighting coefficient and outputting an index _Im and a quantized small sequence vector C (m) And a residual coefficient normalizing unit 126 for normalizing the residual coefficient R (F) with the residual coefficient approximate form E _R (F) to obtain a fine structure coefficient, and normalizing the fine structure coefficient of the current frame. It is given to the quantization unit 125 as the normalized fine structure coefficient X (F) and the index _IG
And a denormalization unit 131 that denormalizes the quantized small sequence vector C (m) and outputs a quantized residual coefficient R _q (F) to the residual approximate calculation unit 123. It has.

In order to perform the auditory weighting based on the present invention in the encoder 110, the spectrum outline calculator 12
1, in addition to the conventional method, the same processing as the processing in the envelope calculating unit 13 and the peak / valley estimating unit 14 of the signal encoding device shown in FIG. 1 is performed, and based on the result, the weight calculating unit In 124, in addition to the conventional method, the same processing as the processing in the weighting unit 15 and the perceptual weight calculating unit 16 of the apparatus shown in FIG. What is necessary is just to supply.

[0037] The decoder 150 on the other hand, the reproduction unit 151 for reproducing the normalized fine structure coefficients from the index I _m
When, the normalized gain reproducing unit 152 for reproducing normalized gain from the index I _G, and power inverse normalization unit 153 and normalized fine structure coefficients by inverse normalization by the normalization gain obtain fine structure coefficients, fine structure coefficients Is denormalized with a residual approximate form ER to reproduce a residual coefficient R (F).
4 and a residual approximate shape calculation unit 155 for calculating the residual approximate shape E _R
And a reproduction / spectrum outline calculation unit 1 for reproducing a linear prediction coefficient from the index I _p and calculating a spectrum outline
56, an inverse normalization unit 157 that inversely normalizes the spectrum outline with a residual coefficient R (F) to reproduce a frequency domain coefficient, and an inverse MDCT unit that performs inverse MDCT on the frequency domain coefficient for each frame to obtain a time domain signal. 158, a windowing unit 159 for applying a time window to the time domain signal for each frame, and a frame overlapping unit 161 for obtaining a reproduced audio signal by superimposing frames on the windowed output and outputting the reproduced audio signal to an output terminal 191. And

In the encoder 110 shown in FIG. 10, the normalization unit 12 is provided without the denormalization unit 131.
Residual envelope calculation section 123 based on only the output of 2 it is possible to calculate the residual coefficients envelope E _R (F) and the index I _Q, the residual in this case, the decoder 150 envelope calculation section 155 calculates a residual envelope E _R based on the index I _Q.

Next, CEL which is a coding method in the time domain is used.
An example in which the present invention is applied to auditory masking of P (Code-Excited Linear Prediction) coding will be described. In CELP coding, since auditory masking is performed in the time domain, the auditory weighting based on the present invention is applied in the frequency domain, and the obtained auditory weight is returned to the time domain and then applied to quantization. FIG. 11 is a block diagram illustrating a configuration of a signal encoding device that performs such encoding.

The device shown in FIG.
An FFT unit 38 for performing FT (fast Fourier transform);
A spectral envelope estimating unit 35 for estimating a spectral envelope based on an output of the T unit (a signal sequence in the frequency domain), an auditory weight calculating unit 36 for calculating an auditory weight from the estimated spectral envelope, and Inverse FFT to return
Unit 39 and the C of the input signal based on the auditory weight in the time domain.
CELP that performs ELP encoding and outputs an index
And an encoding unit 40. In this signal encoding device, the FFT unit 38 corresponds to the T / F conversion unit 11 of the signal encoding device shown in FIG. 1, and the spectrum envelope estimating unit 35 is an envelope calculating unit of the device shown in FIG. 13 and the peak / valley estimating unit 14, and the auditory weight calculating unit 36 includes the weighting unit 15 and the auditory weight calculating unit 16 of the apparatus shown in FIG.

Further, FIG.
FIG. 1 shows an example in which the auditory weighting of the present invention is applied to the speech encoding device disclosed in FIG. The speech coding apparatus shown in FIG. 12 divides a speech signal input via an input terminal 201 into frames, performs linear prediction analysis, and determines a prediction coefficient, a prediction coefficient determination unit 202, a synthesis filter 203, Quantizing the prediction coefficients and synthesizing filter 203
A predictive coefficient quantizer 204 for setting a predictive coefficient in the adaptive codebook 217 for storing a plurality of pitch period vectors,
Noise codebook 218 storing a plurality of noise waveform vectors
And a gain unit 219 a for adding a gain to the pitch period vector selected from the adaptive codebook 217 and a noise codebook 218.
Codebook 219 having a gain section 219b for adding a gain to the noise waveform vector selected from, and gain section 219.
a prediction gain determination unit 215 for obtaining a prediction gain of the next noise waveform vector based on the past output power of b, and a gain unit 21
A prediction gain section 216 provided on the input side of the input section 9b and adding the prediction gain to a selected noise waveform vector;
The adder 209 adds the output vectors from 9a and 219b and supplies the resultant to the synthesis filter 203 as a drive vector.
And a synthesis filter 2 from the input speech vector (input signal)
03, a subtractor 211 that subtracts the output (synthesized speech vector) and outputs the resultant as distortion data, a perceptual weighting filter 220 that performs perceptual weighting on the distortion data, and calculates distortion power based on the perceptually weighted distortion data. Further, a distortion power calculation unit 212 that performs selection in each of the codebooks 217 to 219 so as to minimize distortion power is provided, and a code output unit 213 that outputs a code.

When performing the hearing weighting based on the present invention in this speech coding apparatus, the signal coding apparatus shown in FIG.
Or in combination with the auditory weighting filter 220. As a result, auditory weighting based on the present invention is performed on the distortion data. further,
Here, although not described with reference to the drawings,
In the speech coding apparatus disclosed in FIG. 2 of Japanese Patent No. 2298, the signal coding apparatus shown in FIG. 11 modified as described above can be used as the auditory weighting filter.

The signal encoding method and apparatus according to the present invention described above can also be realized by causing a computer (computer) to read a computer program for realizing the method and executing the program. The program for performing signal encoding is a magnetic tape or CD-RO.
It is read into a computer by a recording medium such as M or via a network. FIG. 13 is a block diagram illustrating a configuration of a computer that executes the above-described signal encoding method.

This computer has a central processing unit (CPU) 2
1, a hard disk device 22 for storing programs and data, a main memory 23, an input device 24 such as a keyboard, a mouse, and a microphone, a display device 25 such as a CRT or a speaker, a magnetic tape or a CD-ROM.
And the like, and a communication device 28 connected to a network. Hard disk device 22, main memory 23,
The input device 24, the display device 25, the reading device 26, and the communication interface 28 are all connected to the central processing unit 21. Instead of the hard disk device 22, a nonvolatile semiconductor storage device such as a flash ROM may be used. The computer attaches a recording medium 27 storing a program for performing signal encoding to a reading device 26, reads the program from the recording medium 27, stores the program in the hard disk drive 22, and executes the program stored in the hard disk drive 22. The central processing unit 21 functions as a signal encoding device when executed. Of course, a program for performing signal encoding may be downloaded to this computer via a network.

[0045]

As described above, according to the present invention,
When encoding a speech / tone signal, auditory masking can be performed with higher accuracy than the conventional method, and high-quality encoding can be performed. Specifically, when the time series signal is converted into a frequency-domain coefficient sequence by, for example, an MDCT transform and quantized, the present invention is used to perceive a quantization error using human auditory masking characteristics. As a result, it is possible to perform the distribution on the frequency axis with higher accuracy than the conventional method.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of a signal encoding device according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a process of weighting peaks and valleys of a spectrum.

FIG. 3 is a block diagram illustrating details of processing in an envelope calculation unit.

FIG. 4 is a block diagram illustrating details of processing in a peak / valley estimating unit;

FIG. 5 is a diagram illustrating an example of a state of peaks and valleys in a spectrum envelope detected by a peak / valley estimating unit.

FIG. 6 is a diagram illustrating an example in which weighting is performed around peaks and valleys of a spectral envelope.

FIGS. 7A to 7D are diagrams illustrating examples of weighting functions for peaks and valleys;

FIG. 8 is a diagram illustrating a state where quantization noise is masked into a spectral envelope by an auditory weighting process.

FIG. 9 is a block diagram illustrating an example of a configuration of a signal encoding device according to the present invention.

FIG. 10 is a block diagram showing an example of a configuration of an encoder and a decoder to which auditory weighting according to the present invention is applied.

FIG. 11 is a block diagram illustrating an example of a configuration of a signal encoding device.

FIG. 12 is a block diagram illustrating an example of a configuration of a signal encoding device.

FIG. 13 is a block diagram illustrating an example of a computer system used to configure the signal encoding device.

[Explanation of symbols]

 Reference Signs List 11 T / F converter 12 Quantizer 13 Envelope calculator 14 Peak / valley estimator 15 Weighter 16 Auditory weight calculator

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Naoki Iwagami 2-3-1 Otemachi, Chiyoda-ku, Tokyo Within Nippon Telegraph and Telephone Corporation (72) Inventor Takeshi Mori 2-chome Otemachi, Chiyoda-ku, Tokyo No. 1 Within Nippon Telegraph and Telephone Corporation (72) Inventor Kazuaki Chikira 2-3-1 Otemachi, Chiyoda-ku, Tokyo F-term within Nippon Telegraph and Telephone Corporation (reference) 5J064 AA01 BA16 BB03 BC16 BC21

Claims (15)

[Claims] 1. A signal encoding method for performing quantization on an input signal, comprising: performing a time axis / frequency axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; Calculating the spectrum envelope based on the sequence; weighting the information at the peak and valley positions of the calculated spectrum envelope; calculating the auditory weight for quantization based on the information weighted spectrum envelope Performing a quantization based on the auditory weight for quantization,
A signal encoding method comprising: 2. A signal encoding method for performing quantization on an input signal, comprising: performing a time axis / frequency axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; A step of performing a smoothing process by obtaining an arithmetic mean over the section length for the column to obtain a spectrum envelope, and a section used in obtaining the spectrum obtained last time with respect to the spectrum envelope obtained last time. Performing a smoothing process by obtaining an arithmetic mean over a section length longer than the length to obtain a spectral envelope at least once, and in any one of the spectral envelopes, a peak in each of the spectral envelopes is obtained. Performing a weighting of the information amount at the position of the valley to obtain a spectrum envelope weighted with the information amount, based on the spectrum envelope weighted with the information amount,
A signal encoding method comprising: calculating an auditory weight for quantization; and performing quantization based on the auditory weight for quantization. 3. A signal encoding method for performing quantization on an input signal, comprising: performing a time-axis / frequency-axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; Calculating a spectrum envelope based on the sequence; estimating the positions of peaks and valleys in the spectrum envelope; and performing weighting of information amounts to the estimated positions of peaks and valleys in the spectrum envelope. Calculating the auditory weight for quantization based on the weighted spectral envelope; andquantizing based on the auditory weight for quantization, wherein the estimating step comprises: Obtaining the first order differential value, calculating the arithmetic mean value of the first order differential value, and obtaining the first order differential value of the arithmetic mean value of the first order differential value to obtain the second order differential value And the second derivative Calculating the arithmetic mean of the first order differential value or the arithmetic mean value of the first order differential value changes from a positive value to a negative value, and in the vicinity of the change, 2
If the arithmetic differential value of the first-order differential value or the second-order differential value is always a negative value, the frequency is regarded as a peak position, and the first-order differential value or the arithmetic mean of the first-order differential value is a negative value. To a positive value, and in the vicinity of the change, the second derivative or 2
If the arithmetic mean of the differential values is always a positive value, the frequency is used as the position of the valley. 4. The step of weighting the amount of information is performed by using a weighting function that raises the vicinity of the hill high and lowers the vicinity of the valley deeply, or lowers the vicinity of the hill and lifts the vicinity of the valley to be shallow. The signal encoding method according to claim 1, further comprising a step of performing a weighting operation on the positions of the peaks and valleys. 5. The method according to claim 1, wherein the step of calculating the auditory weight for quantization is a step of obtaining an auditory weight for quantization based on an envelope curve in which information positions are weighted at peaks and valleys. Signal encoding method. 6. A signal coding apparatus for performing quantization on an input signal, comprising: a conversion unit for performing a time axis / frequency axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; Envelope calculating means for calculating a spectrum envelope based on a coefficient sequence; weighting means for weighting the amount of information at peaks and valleys of the calculated spectrum envelope; and quantizing based on the information weighted spectrum envelope. A signal encoding device comprising: an auditory weight calculating unit that calculates an auditory weight; and a quantizing unit that performs quantization based on the quantizing auditory weight. 7. A signal encoding apparatus for performing quantization on an input signal, comprising: a conversion unit for performing a time axis / frequency axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; The smoothing process is performed to obtain the spectral envelope by obtaining the arithmetic mean over the section length for the coefficient sequence, and then the previously obtained spectral envelope is used for the previously obtained spectral envelope. An envelope calculating means for performing one or more times of performing a smoothing process to obtain a spectrum envelope by obtaining an arithmetic mean over a section length longer than the section length, and the envelope calculation in any of the spectrum envelopes; Weighting means for weighting the amount of information at peak and valley positions in each spectrum envelope obtained by the means to obtain a spectrum envelope weighted with the amount of information; Based on the spectral envelope which is,
A signal encoding device, comprising: an auditory weight calculator that calculates an auditory weight for quantization; and a quantizer that performs quantization based on the auditory weight for quantization. 8. A signal encoding apparatus for performing quantization on an input signal, comprising: a conversion unit configured to perform a time axis / frequency axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; An envelope calculating means for calculating a spectrum envelope based on a coefficient sequence; a peak / valley estimating means for estimating a peak / valley position in the spectrum envelope; and an information amount to the estimated peak / valley position in the spectrum envelope. Weighting means for performing weighting, perceptual weight calculating means for calculating a perceptual weight for quantization based on the information-weighted spectral envelope, quantizing means for performing quantization based on the perceptual weight for quantization, The peak / valley estimating means obtains a first-order differential value of the spectrum envelope, obtains an arithmetic mean value of the first-order differential value,
Calculating the first order differential value of the arithmetic mean value of the second order differential value to obtain a second order differential value, obtaining the arithmetic mean value of the second order differential value, and calculating the arithmetic mean of the first order differential value or the first order differential value If the value changes from a positive value to a negative value and the second derivative or the arithmetic mean of the second derivative is always a negative value in the vicinity of the change,
The frequency is defined as the position of the mountain,
If the arithmetic mean of the second derivative changes from a negative value to a positive value, and if the second derivative or the arithmetic mean of the second derivative is always a positive value in the vicinity of the change, A signal encoding device that uses the frequency as a valley position. 9. The weighting means uses a weighting function that raises the vicinity of the hill higher and lowers the vicinity of the valley deeply, or lowers the vicinity of the hill and lifts the vicinity of the valley so as to be shallower. The signal encoding device according to claim 6, wherein a weighting operation is performed on the position. 10. The signal encoding apparatus according to claim 6, wherein the weighting means obtains an auditory weight for quantization based on an envelope curve obtained by weighting the amount of information to the positions of the peaks and valleys. 11. A recording medium readable by a computer, wherein the computer performs a time-axis / frequency-axis conversion on an input signal to obtain a coefficient sequence on a frequency axis, based on the coefficient sequence. Calculating the spectral envelope by calculating the amount of information to the peaks and valleys of the calculated spectral envelope; and calculating the auditory weight for quantization based on the information-weighted spectral envelope. A step of performing quantization based on the quantization auditory weight; and a recording medium storing a signal encoding program for executing the following. 12. A recording medium readable by a computer, wherein the computer performs a time axis / frequency axis conversion on an input signal to obtain a coefficient sequence on a frequency axis; Performing a smoothing process to obtain an arithmetic average over the section length to obtain a spectrum envelope, and comparing the previously obtained spectrum envelope with the section length used in obtaining the previously obtained spectrum. Performing a smoothing process by obtaining an arithmetic mean over a long section length to obtain a spectral envelope at least once, and in any one of the spectral envelopes, Performing an information amount weighting on the position to obtain an information amount weighted spectrum envelope, based on the information amount weighted spectrum envelope,
A recording medium storing a signal encoding program for executing the steps of: calculating a hearing weight for quantization; and performing quantization based on the hearing weight for quantization. 13. A recording medium readable by a computer, wherein the computer performs a time axis / frequency axis conversion on the input signal to obtain a coefficient sequence on a frequency axis; Calculating a spectrum envelope based on the spectrum envelope; estimating the positions of peaks and valleys in the spectrum envelope; performing weighting of information amounts to the estimated positions of peaks and valleys in the spectrum envelope; Calculating an auditory weight for quantization based on the obtained spectral envelope; and performing a quantization based on the auditory weight for quantization. Calculating a first derivative, calculating an arithmetic mean of the first derivative, calculating a first derivative of the arithmetic mean of the first derivative to obtain a second derivative, 2 above Calculating an arithmetic mean value of the first derivative value, wherein the first derivative value or the arithmetic mean value of the first derivative value changes from a positive value to a negative value, and In the vicinity 2
If the arithmetic differential value of the first-order differential value or the second-order differential value is always a negative value, the frequency is regarded as a peak position, and the first-order differential value or the arithmetic mean of the first-order differential value is a negative value. To a positive value, and in the vicinity of the change, the second derivative or 2
If the arithmetic mean of the differential values is always a positive value, a recording medium that stores a signal encoding program that sets the frequency as a valley position. 14. The step of weighting the amount of information is performed by using a weighting function that raises the vicinity of the hill high and lowers the vicinity of the valley deeply, or lowers the vicinity of the mountain and lifts the vicinity of the valley so as to be shallow. The recording medium according to claim 11, further comprising a step of performing a weighting operation on the positions of the peaks and valleys. 15. The step of calculating auditory weights for quantization includes:
12. The recording medium according to claim 11, wherein the step of obtaining an auditory weight for quantization is based on an envelope curve obtained by weighting the amount of information to peaks and valleys.