JPH10190498A - Improved method generating comfortable noise during non-contiguous transmission - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method that generates comfortable noise (CN) in a mobile terminal in a non-contiguous transmission (DTX) system. SOLUTION: This method for generating comfortable noise is to make the frequency characteristic of the comfortable noise close to actual background noise by making a random exciting signal pass through a spectrum controlled filter. In the form of other execution, a sending end identifies a voice coded parameter that does not satisfactorily represent actual background noise, and when such voice coded parameters are detected, the parameters are replaced with a mean value. In this method, an balancing operation is not disturbed by a faulty parameter.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to voice communication technology in general, and more particularly to discontinuous transmission (DTX), and to improvement of comfort noise (CN) in discontinuous transmission.

[0002]

2. Description of the Related Art In a mobile communication system, a discontinuous transmission system is used in which a radio transmitter is disconnected while speech voice is temporarily interrupted. By using such a discontinuous transmission scheme, the power consumption of the mobile station can be reduced, and thus the battery recharge interval can be lengthened.
In addition, the use of the discontinuous transmission method generally reduces the communication disturbance level, thereby improving communication quality.

[0003] However, if the channel is completely disconnected, the background noise included in the audio signal and transmitted is also disconnected during the transmission suspension period. Signal state).

[0004] One known method for solving this problem is to extract characteristic parameters of background noise without completely disconnecting the transmission during a transmission suspension period, and to convert the characteristic parameters into unvoiced descriptor (SID) frames. There is known a method of performing wireless transmission at a low transmission speed by using a wireless communication. The receiving side uses this parameter to reproduce the background noise so as to reflect as closely as possible the spectral and temporal substance of the background noise actually existing on the transmitting side. The parameters characterizing such background noise are called comfort noise (CN) parameters. The comfort noise parameters usually include a subset of the speech coding parameters, in particular the synthesis filter coefficients and the gain parameters.

In a method for calculating comfort noise in a speech codec, only a part of comfort noise coefficients is extracted from speech coding parameters, and other comfort noise parameters are
For example, it is extracted from a signal that is present in a speech encoder but is not transmitted wirelessly.

[0006] In conventional discontinuous transmission systems, it is assumed that the excitation signal can be approximated sufficiently well using noise having a flat spectrum (ie, white noise). That is, in the conventional discontinuous transmission method,
Comfort noise is generated by passing locally generated noise having flat spectral characteristics through a speech coder synthesis filter. However, in practice, such a method using white noise cannot generate high-quality comfort noise. The reason for this is that the spectrum of the ideal comfort noise is not flat, but has a slightly skewed spectral characteristic or, in some cases, a spectral characteristic that deviates significantly from the flat spectral characteristic. Depending on the type of background noise, the optimal comfort noise spectrum has, for example, low-pass or high-pass characteristics. That is, the conventionally used random excitation (random excitation) signal (white noise) is deviated from the optimized desired noise excitation, so that the comfort noise generated on the receiving side is the background noise on the transmitting side. Is perceived to be different. For example, the generated comfort noise may have a much more "brilliant" feeling than the desired comfort noise, or conversely a "dark" feeling. In discontinuous transmission, the spectrum of the background noise differs between during transmission (period during speech coding) and during transmission interruption (period during comfort noise generation). Such a change in the background noise is perceived by the user as a deterioration in communication quality.

[0007] such as full speed (FR), half speed (HR), enhanced full speed (EFR) voice channels in GSM systems,
In speech coding systems, the transmission of comfort noise parameters occurs at a low rate. For example, FR and FER
In the channel, the transmission rate of the comfort noise parameter is very low, once every 24 frames (ie, once every 480 milliseconds). That is, the comfort noise parameter is updated only twice per second. Because of the low transmission speed, the spectral characteristics or temporal characteristics of the background noise cannot be expressed sufficiently accurately. Therefore, in the discontinuous transmission method, the quality of the background noise is inevitably reduced. to degrade.

Another problem that occurs in discontinuous transmission in digital cellular systems such as GSM is related to the hangover of some voice frames before the transmission after the voice burst actually ends. If the audio burst is below a certain threshold, the audio burst is interpreted as background noise, in which case no hangover period occurs after the audio burst. During the hangover period, before ending the transmission, the transmitting side calculates an evaluation characteristic value of the background noise of the transmitting side to be transmitted to the receiving side as a comfort noise parameter message (unvoiced descriptor (SID) frame). Used for As described above, on the reception side, comfort noise having characteristics similar to the background noise existing on the transmission side at the end of transmission is generated using the transmitted background noise evaluation characteristic value.

In known discontinuous transmission schemes similar to GSM FR or HR, comfort noise quantization is performed by a non-predictive scheme. Therefore, the receiving side does not need to know whether a hangover period exists at the end of the voice burst. However, in the case of GSM EFR, an efficient predictive comfort noise quantization method is employed, and therefore, it is necessary to determine whether a hangover period exists when performing comfort noise inverse quantization on the receiving side. Therefore, some computational load is spent executing the program instructions.

If the background noise on the transmitting side is not constant but fluctuates considerably, the following problem further occurs. That is, in such a situation, some or all of the speech coding parameters of one or several of the frames in the averaging period may not correctly represent the average background noise. . A similar problem occurs when the end of the spoken voice is unvoiced, as determined by a voice state detection (VAD) algorithm, or when the constant background noise includes a strong impulse-like noise burst. In the existing discontinuous transmission method, since the averaging period is short, when an inappropriate voice coding parameter as described above occurs,
The result of the averaging process changes significantly, and the averaged comfort noise parameters do not accurately characterize the background noise. Therefore, between the comfort noise and the actual background noise,
Differences occur in levels and / or spectra. Therefore, the period during which the call is actually being made (the period during which speech and background noise are encoded in the usual way)
Then, the user feels that the background noise has changed between the period during which the call is interrupted (the background noise during this period is generated by the comfort noise), which impairs the communication quality.

[0011] More specifically, during the discontinuous transmission hangover period, while the frame determined to be a "non-voice frame" by the voice state determination algorithm is wirelessly transmitted, the first unvoiced descriptor frame is comforted. Speech coding parameters used to calculate noise parameters are temporarily stored. The first unvoiced descriptor frame described above is transmitted as soon as the discontinuous transmission hangover period ends. The length of the discontinuous transmission hangover period is determined by the length of the averaging period. Therefore, the averaging period needs to be fixed at a relatively short length to minimize the use of the system's channels.

Before describing the present invention, circuits and methods conventionally used to generate comfort noise parameters on the transmitting side and for generating comfort noise based on these parameters on the receiving side are described. The circuit and method are described below. First, FIGS.
See (d).

In FIG. 1A, a linear predictive coding (L
The PC) analysis block 101 calculates short-term spectral parameters 102 from the audio signal 100. Note that linear predictive coding (LPC) is a technique well known in the art. For simplicity, it is assumed here that the synthesis filter consists only of short-term synthesis filters, but in most conventional systems, such as GSM FR, EFR encoders, the synthesis filter is actually a short-term synthesis filter. This is realized by connecting a series synthesis filter and a long-term synthesis filter in series. It should be noted that a long-term synthesis filter is not required to explain the present invention. Further, in the conventional discontinuous transmission method, when generating comfort noise, the synthesis filter is usually cut off for a long period of time.

The LPC analysis produces a set of short-term spectral parameters 102 for each transmitted frame. The frame length at this time differs depending on the system. For example, in the GSM channel, the frame sizes are all set to 20 milliseconds.

The audio signal is passed through an inverse filter 103, where a residual signal 104 is generated. This inverse filter has a characteristic expressed by the following equation.

[0016]

[Equation 11]

Here, a (i) (i = 1,..., M)
Are filter coefficients, which are generated in the linear predictive coding analysis, and are updated for each frame. To make the change of the filter parameter between frames smoother,
The interpolation processing used in the conventional speech coding technology may be applied to the inverse filter. Here, the residual signal 104 generated by the inverse filter 103 is an optimal excitation signal. By passing this signal through a synthesis filter 1 / A (Z) 112 (FIG. 1B) on the receiving side, The original audio signal 100 is accurately restored. The excitation gain calculation block 105 measures the energy of the excitation signal sequence and calculates a scaling gain 106 for each transmission frame.

The average spectrum and temporal characteristics of the background noise are extracted from the average of the excitation gain 106 and the short-term spectral coefficient 102 in several transmission frames. Normally, the above averaging process is performed on four frames in the case of the GSM FR channel, while it is performed on eight frames in the case of the GSM EFR channel. The parameters to be averaged are blocks 107a and 108a (FIG. 1 (d)
) During the averaging period. The above averaging process is performed in blocks 107 and 108 to generate an average parameter characterizing the background noise. The average parameters generated here include the average excitation gain g _mean and the average short-term spectral coefficient. Modern speech codecs commonly use ten (M = 10) short-term spectral coefficients, which, as is done in GSM EFR discontinuous transmission systems, have a line spectrum pair (LSP) coefficient f _mean ( i) (i = 1,..., M).
These parameters are usually quantized before transmission, but for the sake of simplicity, a description of quantization will be omitted here. Note that quantization is not an essential part in understanding the operation of the present invention.

As shown in FIG. 1 (d), averaging blocks 107 and 108
a, 108a and through these buffers the signal 1
07b and 108b are input. Buffers 107a and 108a will be described in more detail when describing the present invention with reference to FIGS.

The calculation and averaging of comfort noise parameters is described in detail in the GSM Recommendation: GSM 06.62, "Characteristics of Comfort Noise in Enhanced Full Rate (EFR) Voice Traffic Channels". For the discontinuous transmission method, the GSM recommended standard GSM
06.81, "Discontinuous Transmission (DTX) Scheme on Enhanced Full Rate (EFR) Voice Traffic Channel", and for voice state detection (VAD), G
SM Recommended Standard GSM 06.82 "Extended Full Speed (EFR)"
Voice State Detection on Voice Channel (VAD) ". Therefore, descriptions of these various functions are omitted here.

FIG. 1B is a block diagram showing a conventional decoder used for generating comfort noise on the receiving side in a conventional voice communication system. The decoder receives two comfort noise parameters, the average excitation gain g _mean and a set of short-term spectral coefficients f _mean (i) (i = 1,..., M), and based on these parameters Comfort noise occurs. The operation of comfort noise generation on the receiving side is based on the fact that the parameters are used at a fairly low speed (eg once every 480 ms in GSM FR and GSM EFR channels) and that there is no excitation signal from the speech coder. Except for this, it is the same as the speech decoding operation. The excitation signal on the receiving side in the audio decoding operation is
An index obtained from a codebook containing a plurality of possible excitation signal sequences and corresponding to a particular excitation vector in the codebook is transmitted along with other speech coding parameters. Speech decoding and codebook reference operations are disclosed in detail in, for example, U.S. Pat. , All of which are incorporated herein by reference.

On the other hand, in generating comfort noise, the index of the codebook is not transmitted, and the excitation signal is obtained by using the random excitation signal (RE) generator 110 instead. The random excitation signal generator 110 generates an excitation signal vector 114 having a flat spectrum.
The excitation signal vector 114 is stored in the scaling unit 11
5, so that the energy is scaled by the average excitation gain g _mean so that its energy corresponds to the average gain of the excitation signal 104 at the transmitter. The scaled random excitation signal sequence 111 thus obtained
Is input to the speech synthesis filter 112, and a comfort noise output signal 113 is generated. At that time, the average short-term spectral coefficient f _mean (i) is used in the speech synthesis filter 112.

FIG. 1 (c) shows the spectra at various parts of the prior art decoder of FIG. 1 (b). The random excitation signal generator 110 generates a random excitation signal sequence 114 (and a scaled random excitation signal sequence 111) having a flat spectrum.
This spectrum is shown by curve A. The speech synthesis filter 112 converts this excitation signal into a signal having a non-flat spectrum as shown by the curve B.

As described above, the conventional comfort noise generating technology has many problems. One of these problems is that, as described above, the comfort noise due to random excitation generated on the receiving side is not properly comfortably excited comfort noise, and the actual background of the transmitting side is not. It is perceived differently from noise. It is an object of the present invention to completely eliminate or reduce such problems.

A first object and advantage of the present invention is to provide an improved method for generating comfort noise in discontinuous transmissions, thereby minimizing signal degradation that occurs in discontinuous transmission schemes. .

It is another object and advantage of the present invention to provide an improved method of generating comfort noise that can better extract the characteristics of background noise, and thereby further provide improved quality comfort. It is an object of the present invention to enable communication with improved noise and improved quality in a discontinuous transmission scheme.

Yet another object and advantage of the present invention is to provide an improved method of generating comfort noise that can better represent actual background noise in discontinuous transmissions within a short averaging time. It is.

[0028]

SUMMARY OF THE INVENTION To overcome the above and other problems, an object and advantage of the present invention is a method for generating comfort noise (CN) in discontinuous communications (DTX). Is achieved by an improved method.

The present invention also provides an improved method for generating comfort noise, wherein a random excitation signal has a frequency characteristic similar to that of background noise using a spectrum control filter.

In the present invention, the generation of comfort noise
A random excitation signal having a flat spectral distribution is not used. Instead, the random excitation signal is appropriately modified so that comfort noise can be generated that more accurately characterizes the spectrum of the background noise actually present at the transmitting end of the communication. In this way, comfort noise with improved quality can be obtained.

The method of the present invention includes calculating random excitation spectrum control (RESC) parameters at the transmitting end. On the other hand, on the receiving side, the random excitation signal is transformed using the above-mentioned spectrum control parameters so that the spectral characteristics of the generated or generated comfort noise can accurately represent the background noise actually present on the transmitting side. When the speech is interrupted, a random excitation spectrum control (RESC) parameter and other comfort noise parameters are calculated and transmitted to the receiver.

In the method of the present invention, in the first step, random excitation spectrum control (R
ESC) parameters are calculated. These parameters are sent to the receiver together with the CN parameters. On the receiving side, after shaping the spectral characteristics of the excitation signal using the RESC parameters, the shaped excitation signal is input to the synthesis filter.

Further, according to the present invention, an average of the parameters is obtained, and if a bad speech coding parameter exists during the averaging period, all or a part thereof is removed or replaced with a median value. In an embodiment of the present invention, the distance between the speech coding parameters between the individual frames in the averaging period is measured with respect to each other, these parameters are arranged in the order of the measured distances, and the other parameters in the averaging period are The parameter having the largest distance is identified, and if the maximum distance exceeds a predetermined threshold, the parameter having the maximum distance is compared with another parameter within the averaging period. Replace with the parameter with the smallest distance (ie, the median). The parameter having the median value can be regarded as having a value that most accurately represents the characteristic of the background noise among the parameters obtained in the averaging period.
After performing the above processing, the average of the speech encoding parameters is obtained by using an appropriate arbitrary method. The embodiment of the present invention does not limit how to receive and use the comfort noise parameter on the receiving side of the discontinuous transmission system.

In addition to improving comfort noise quality by removing bad CN parameters within the averaging period, embodiments of the present invention have other advantages. For example, in the discontinuous transmission system according to the related art, it is necessary to lengthen the averaging period in order to reduce adverse effects due to bad parameters in the averaging period. In the present invention, since a measure is taken to reduce the influence of the bad parameter on the averaging, the averaging period may be shorter than the averaging period required in the related art.
Also, in the prior art, since the averaging period is long,
The hangover period also becomes longer, which increases the channel usage time. On the other hand, in the present invention, since the averaging period can be shortened, the discontinuous transmission hangover period can also be shortened, and thus the channel use time is also reduced. Further, in prior art discontinuous transmission systems,
Due to the long averaging period, a fairly large amount of static memory is required to implement the CN averaging algorithm. On the other hand, in the present invention, the averaging period is short, and a small-capacity static memory is sufficient to execute the CN averaging algorithm.

[0035]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The prior art for encoding and decoding comfort noise has been described above. A circuit and a method according to the first embodiment of the present invention will be described below with reference to FIGS. Note that FIG.
In FIGS. 1A and 2B, the same reference numerals are used for elements corresponding to FIGS. 1A and 1B.

The “silent descriptor averaging period” is a term used in GSM, and is described in IS-641 Rev. A
The term "comfort noise averaging period" or "C
"N averaging period" is used. In the following description of the present invention, these two terms will be used interchangeably. Similarly, “SID frame” and “comfort noise parameter message” or “CN parameter message”
Is used synonymously.

FIG. 2 (a) is a block diagram showing an apparatus for generating a comfort noise parameter on the transmitting side according to the present invention. New parts unique to the present invention are indicated by broken lines 204 to distinguish them from parts according to the prior art. In this embodiment of the present invention, the residual signal 1 output from the inverse filter 103 is
04 is further analyzed (eg, LPC analysis) to generate a further set of filter coefficients. For this second analysis, the present invention uses a random excitation (RE) linear predictive coding (LPC) analysis 200, which is generally of lower order than the linear predictive coding analysis performed in block 101.
The spectral parameters 201 output from the random excitation linear prediction coding analysis block 200 are averaged for several consecutive frames by an averaging block 203, whereby the random excitation spectrum control (RE) is performed.
SC) Parameter r _mean (i) (i = 1,..., R)
Is required. The spectrum of the excitation signal is characterized by this random excitation spectrum control parameter.

It should be noted that the random excitation spectrum control parameter is not a subset of the speech coding parameter but a parameter generated and used only in generating comfort noise. It should be noted that the inventors of the present invention have found that a sufficiently good result can be obtained by a first-order or second-order linear predictive coding analysis in order to generate a random excitation spectrum control parameter (R = 1 or 2). I found However, it is also possible to use a spectral model other than the all-pole linear predictive coding method. Alternatively, the averaging process by the random excitation linear prediction coding analysis block 200 may be performed by averaging the autocorrelation coefficients in the calculation of the linear prediction coding parameters. May be calculated using other suitable averaging methods. The averaging period of the random excitation spectrum control parameter may be the same as the averaging period of the other comfort noise parameters, but it is not necessary that the averaging period be the same. For example, it is desirable to set the averaging period of the random excitation spectrum control parameter to be longer than the averaging period conventionally used for the comfort noise parameter. That is, it is desirable to adopt a longer averaging period, for example, 10 to 12 frames instead of employing 7 frames as the averaging period.

Prior to calculating the signal gain, the linear prediction coded residual signal 104 is converted to a second inverse filter H _RESC
(Z) 202 to be processed. This filter 2
02 transforms the linear prediction coded residual signal 104 into a residual signal 205 that generally has a flatter spectrum. Inverse random excitation spectrum control filter H
_{As RESC} (Z), an all-zero filter of the following form can be used, but is not limited to this.

[0040]

(Equation 12)

Next, the excitation signal gain is calculated from the residual signal 205 having the flattened spectrum. In other respects, the operation of FIG. 2A is similar to the operation of FIG. 1A described above.

FIG. 2 (b) is a block diagram showing a decoder used to generate comfort noise on the receiving side according to the present invention. Excitation signal 21 by this decoder
The formation of 2 is performed as follows. First, the white noise excitation signal sequence 114 is generated by the random excitation signal generator 11.
0, which is then scaled by g _mean by scaling block 115.

The noise signal sequence 111 having a flat spectrum is then subjected to random excitation spectrum control (RE
SC) is processed by a filter 211 to generate an excitation signal having a desired spectrum. The random excitation spectrum control filter 211 performs a process reverse to that of the random excitation control inverse filter 202 of the encoder in FIG. When a random excitation spectrum control inverse filter having the characteristic of Equation (2) is used on the transmission side, on the reception side:
A random number control spectrum control filter 211 having the following characteristics is used.

[0044]

(Equation 13)

Filter coefficients b (i) (i = 1,...,
R) defining the random excitation spectrum control parameter r _mean (i) (i = 1,..., R) are sent to the receiving side as part of the comfort noise parameter, and the random excitation spectrum control filter 211 uses this random excitation spectrum control parameter to generate an excitation signal having an appropriately weighted, and therefore generally non-flat, spectrum and provides it to synthesis filter 112. Random excitation spectrum control parameter r
_mean (i) (i = 1,..., R) is a filter coefficient b
(I) may be the same as (i = 1,..., R);
Alternatively, any other suitable parameter representation, such as a line spectrum pair (LSP) coefficient, which can be efficiently quantized and transmitted, may be used. FIGS. 11A to 11
(G) shows a random excitation spectrum control filter 211.
Shows the typical frequency response of

As described above, the comfort noise excitation signal generator 210 of the present invention has features not found in the past. That is, the comfort noise excitation signal generator 210
Generates a random excitation signal having a flat spectrum using the random excitation signal generator 110. The excitation signal having this flat spectrum is supplied to the random excitation vector control filter 211 after being appropriately scaled by the average gain scaler 115. In order to prevent a mismatch between the spectrum of the comfort noise and the spectrum of the background noise, comfort noise having an appropriate spectrum is generated, and a random excitation signal is supplied via the random excitation vector control filter 211. next,
The excitation signal 21 whose spectrum is controlled as described above
2, speech synthesis filter 112 generates comfort noise whose spectrum better matches the background noise actually present at the transmitting end.

The random excitation spectrum control parameter is not a subset of speech coding parameters used for speech signal processing, but a parameter calculated and used only in comfort noise generation. That is, the random excitation spectrum control parameters are calculated and transmitted only for the purpose of generating and generating improved comfort noise during pauses in speech speech. Also, the random excitation spectrum control inverse filter 202 of the encoder
And a random excitation spectrum control filter 21 of the decoder
1 is used only for the purpose of controlling the spectrum of the random excitation signal.

FIG. 2 (c) shows the spectrum of the signal in the decoder of FIG. 2 (b) when comfort noise is generated according to the present invention. Random excitation signal generator 11
0 generates a random number sequence having a flat spectrum as shown by the curve A. This spectrum is the same as that shown by the curve A in FIG. Both signal 114 and signal 111 have such a flat spectrum. That is, the scaling of the block 115 does not affect the shape of the spectrum. The white noise sequence 111 is input to the random excitation spectrum control filter 211, and the excitation signal 21
2 is supplied to a linear predictive coding synthesis filter. The excitation signal sequence 212 with improved properties generally has a non-flat spectrum (curve C). Therefore, the spectrum of the output signal 113 of the synthesis filter 112 becomes as shown by a curve D in FIG. 2C, for example. The excitation signal sequence 212 may be low-pass or high-pass, or may include more complex frequency components (depending on the order of the random excitation spectrum control filter). The control of the spectrum is determined by the random excitation spectrum control parameters calculated at the transmitting side and transmitted to the receiving side as part of the comfort noise, as described above.

FIGS. 3A and 3B show another embodiment of the present invention. FIG.
Compared with (a), in the case of the present embodiment, the excitation signal gain is
Is calculated from the linear prediction coded residual signal 104, not from the residual signal output from. Therefore, in the present embodiment shown in FIG. 3A, the random excitation spectrum control inverse filter 202 is not necessary. FIG. 3B shows a decoder used on the receiving side corresponding to the encoder shown in FIG. The excitation signal scaling block (block 11) is compared to the decoder of FIG.
5) is different in that the random excitation spectrum control filter 211 is moved from the input side to the output side. In other respects, the operation of the encoder and decoder of FIGS. 3A and 3B is the same as the operation of the encoder and decoder of FIGS. 2A and 2B.

FIG. 4 is a block diagram showing a circuit for calculating a comfort noise parameter on the transmission side according to still another embodiment of the present invention. The present embodiment addresses the above-described problem that occurs when one or several frames in the average period have some or all of the speech coding parameters significantly deviated from the average background noise. It is. Parts specific to this embodiment of the invention are surrounded by dashed lines 300 and 310 to distinguish them from prior art parts. In this embodiment of the present invention, the speech coding parameters buffered in blocks 107a and 108a are first subjected to a median threshold replacement process before averaging blocks 107 and 108.
To calculate the average excitation signal gain g _mean and the average short-term spectral coefficient f _mean (i). By this processing, if the averaging period includes a parameter that deviates from the average value of the background noise by a certain amount or more,
The parameter is replaced with the value of the parameter representing the actual background noise, ie, the median.

First, the calculation processing of the scalar excitation signal gain parameter g by the block 300, which is performed before the averaging processing by the block 107, will be described.
A series of excitation signal gain values 107b during the averaging period are:
After being temporarily stored in the buffer block 107a, these values are input to the block 301 and are rearranged in the order of the values. These series of excitation gain values are:
Each is given an index. The sorted series of gain parameters 302 is
03, the L values of the excitation signal gains whose difference from the median exceeds a predetermined threshold value are replaced with the median of the series of parameters. The difference between the value of the individual parameter and the median is calculated by block 304, and if the difference exceeds a threshold, the index corresponding to the value of the excitation signal gain is provided to the median replacement block 303 as signal 305. Is done.

An odd value is preferably selected for the length N of the averaging period. In this case, the ((N + 1) / 2) -th value of the rearranged series of excitation gains is the center value. The variable L that determines the number of parameters to be replaced is 0
To N-1 and is determined in advance (for example, a constant).

For a value of the excitation gain whose difference from the center value exceeds a predetermined threshold value, the selector 3
07 is switched, and the output signal 308 of the center value replacement block 303 is supplied to the averaging block 107 as a value 309 of the excitation gain. On the other hand, for individual excitation signal gain values whose difference from the median does not exceed a predetermined threshold, the output of the buffer block 107a is used as the input parameter value 309 directly as the averaging block 1
The selector 307 is switched so as to be supplied to the selector 07.

The selection state of the selector 307 is controlled by a signal 306 output from the threshold block 304.

Next, before the averaging process by the block 108, the process regarding the line spectrum pair (LSP) coefficient f (k) (k = 1,..., M) performed by the block 310 will be described. During the averaging period, a series of LSP coefficients 108b are buffered in block 108a before being provided to block 311. The j-th LSP coefficients of the frame f _j of the same averaging period of the LSP coefficients of the i-th frame in the averaging period f _i (k)
The spectral distance to (k) can be approximately expressed by the following equation.

[0056]

[Equation 14]

Here, M is the order of the linear predictive coding model, and f _i (k) is the k-th LSP parameter of the i-th frame in the averaging period.

During the averaging period of length N, the LSP coefficients f _i (k) of the _ith frame are compared with the LSP coefficients of all other frames j = 1,..., N (i ≠ j). To determine the spectral distance ΔS _i , the sum of the spectral distances ΔR _ij is calculated for all i = 1,.
.., N.

[0059]

(Equation 15)

However, ΔR _ij = 0 (that is, the distance from itself is 0). The processing represented by the above equations (4) and (5) is performed in block 311.

The spectral distance is determined by various other L
Approximately, it can be obtained by expression using a PC filter. Specific examples thereof include, for example, AH Gray Jr., JD Markel, “Distance Measure in Audio Processing” (IEEE Sound, Audio, and Signal Processing). 24, p.
380-391, 1976). Also, an immittance spectrum pair (ISP) can be used in the same manner as a line spectrum pair. See, for example, Wybystritz, Esperell, "Imittance Spectrum Pairs in Speech Coding (ISP)".
(Proceedings of the IEEE International Conference on Sound, Voice, and Signal Processing, Minneapolis, MN, Vol. 2, p. 9-1
2, April 27-30, 1993).

The spectral distance ΔS _i of each of the LSP vectors f _i during the averaging period determined by the block 311 is supplied to the block 313. In the alignment block 313, the spectral distances are rearranged in the order of their magnitudes. Note that each of these spectral distances is associated with the LPS vector of the averaging period by an index. In the averaging period, vector f _i (i with the smallest distance [Delta] S _i =
1,..., N) are regarded as the median vector f _med of the averaging period, and the distance is represented by ΔS _med .

A series of LSP coefficient vectors f _i in the averaging period are divided into blocks 3 according to the order of the spectral distance.
13 aligned. The series of LSP vectors 314 aligned by the block 313 is supplied to the median replacement block 315. In block 315, LSP vector f _i of P (0 ≦ P ≦ N-1 ) is,
_{Replaced by the} median f _med . The index of these P vectors is given at block 316 as ΔS _i (i
= 1, 2,..., N) and the median value ΔS _med . The index of f _i where ΔS _i −ΔS _med is greater than a predetermined threshold is communicated to median replacement block 315 by signal 317.

For some i = 1, 2,..., N, if the difference ΔS _i −ΔS _med is larger than the threshold value, the output signal 320 from the median substitution block 315 is averaged as a parameter 321. The selector 319 is switched so as to be supplied to the block 108. On the other hand, ΔS _i −
If ΔS _med is less than the threshold value for all i = 1, 2,..., N, the input signal 321 to the averaging block 108 is
The position of selector 319 is switched, as given as 08b.

The selection state of the selector 319 is controlled by a signal 318 output from the threshold block 316.

FIG. 5 is a diagram showing another embodiment of the present invention. Parts specific to this embodiment of the invention are surrounded by dashed lines 400 to distinguish them from prior art parts. Figure 4 above
In the embodiment, each median of the excitation signal gain g and LSP vector f _i is processed independently, Fig. 5
In this embodiment, these parameters are processed simultaneously, as described below.

[0067] a frame of parameters, it is determined as to be replaced with the median value, Both the excitation signal gain g and LSP vector f _i of that frame are replaced with the respective parameters of the frame having a central .

The i-th frame and the j-th frame in the averaging period are used to align the frames for performing the median replacement.
Instead of equation (4), which represents the approximate distance ΔR _ij between the parameters of the th frame, the excitation signal gain g and the LSP
The following equation (6) in which both of the vectors f _i are considered is used here.

[0069]

(Equation 16)

Here, M is the order of the linear predictive coding model, and f _i (k) is the k-th LPS parameter of the i-th frame in the averaging period. Also, q
_i is the excitation signal gain parameter of the ith frame.

First, ΔT _ij is calculated, and then, using the equation (5), the parameters of the frame i in the averaging period having a length of N and other frames j = 1... N (i The distance ΔS _i to ≠ j) is obtained for all i = 1,..., N. The ΔT _ij obtained by using equation (6) is calculated by using equation (5).
In applying the formula (1), ΔT _ij is used instead of the distance ΔR _{ij in} the equation (5). The processing according to the above equations (5) and (6) is executed by the block 401. The weight coefficient w in the above equation (6) is appropriately selected depending on whether the median value replacement processing is performed with emphasis on the excitation signal gain or the spectral distance. This weight coefficient is set to an appropriate value empirically obtained by the user.

The distance ΔS _i of each frame during the averaging period obtained by the block 401 is supplied to the alignment block 403 as 402. The alignment block 403 rearranges distances according to these values. Each of these distances is associated with one frame within the averaging period by an index. Within the averaging period, the minimum distance ΔS _i (i = 1, 2,.
, N) are taken as the median frames for the parameters g _med and f _med of their averaging period, and their distances are denoted as ΔS _med .

The excitation signal gain to be aligned by block 403 is provided as signal 107b from buffer 107a to block 403, while the LSP coefficient is
8b is supplied from the buffer 108a to the block 403. As already mentioned, the sequence of parameters of the averaging period is ordered by block 403 in order of its spectral distance ΔS _i . The set of parameters aligned by block 403 is provided as signals 404 and 405 to median replacement block 406 and block 4
At 06, the parameters g _i and f _i of the L (0 ≦ L ≦ N−1) frames are replaced with the parameters g _med and f _{med of the} median frame. The index of these L vectors is given by block 407 where ΔS _i (i =
1, 2,..., N) with the median ΔS _med and the determined index is the signal 40
8 is transmitted to the median replacement block 406. ΔS
_{If i-} ΔS _med is greater than a predetermined threshold, block 407
, The median replacement block 406 replaces the corresponding parameters g _i and f _i with g _med and f _med . Note that the minimum value and the maximum value of the value of L may be determined in advance.

If any of i = 1, 2,..., N is greater than the distance ΔS _i −ΔS _med , the signal 411 output from the median substitution block 406 is The signal 41 is input as a parameter 321 and output from the median replacement block 406.
The position of selector 410 is switched so that 2 is input to averaging block 107 as parameter 309. On the other hand, ΔS _i −ΔS _med indicates that all i = 1, 2,.
.., N, the input signal 321 of the averaging block 108 is
The position of the selector 410 is switched such that the input signal 309 of the averaging block 107 is directly supplied as the signal 107b from the buffer block 107a. The position of the selector 410 is determined by the signal 4 from the threshold block 407.
09.

As described above, the median distance is subtracted from each distance value (that is, ΔS _i -ΔS _med is obtained).
Instead, use the quotient of each distance divided by the median distance (ΔS _i / ΔS _med ) in blocks 316 and 407 to represent the difference between the individual distance and the median distance. In many cases, it is more preferable to use the quotient in this way. Using the quotient gives the relative displacement of the individual distance values from the median distance, ie the normalized displacement, which is independent of the absolute value of ΔS _i or ΔS _med , The deviation can be expressed.

Next, another embodiment of the present invention will be described. Before proceeding, the discontinuous transmission system of the speech encoder on the transmission side will be described with reference to the block diagram of FIG. Signal 601 from A / D converter 600 is processed by speech encoder 602 for each frame. As in the above case, the length of the frame is typically 20 milliseconds. The sampling frequency of the audio signal 601 is generally 8 kHz. Speech encoder 602 encodes the input speech into a set of parameters 603 for each frame. The parameters thus obtained are transmitted from the radio subsystem 611 of the digital mobile radio unit to the reception (RX) side.

The discontinuous transmission operation is indirectly controlled by the voice state detection device on the transmission side. Voice state detection device 6
The basic functions of the 04 are noise present in the audio signal,
This is to discriminate the noise when no noise exists but only noise exists. The voice state detection device 604 continuously monitors and determines whether the input signal includes voice or no voice. The operation of the speech state detection device 604 is performed based on the speech encoder 602 and its internal variables 605. The voice state detection device 604 outputs a binary voice state flag 606 that takes a value of 1 when voice is present and takes 0 when voice is not present. The operation of the voice state detection device 604 is based on the standard GSM0.
As specified in 6.82, the processing is performed in units of frames.

The traffic frames, each associated with a binary SP flag 607, are sent continuously from the speech encoder discontinuous transmission handler 612 to the radio subsystem 611. The radio subsystem 611 determines whether the traffic frame transmitted from the discontinuous transmission handler 612 is a voice frame (SP flag = “1”) or a so-called unvoiced identifier (SID) frame (that is, a comfort noise parameter message) ( Whether the SP flag is "0") can be known from the SP flag 607. The wireless subsystem 611 transmits the frame to the air as a radio wave at a timing based on the state of the SP flag 607.

The basic problems in the above discontinuous transmission are as follows. That is, the audible background noise that has been transmitted together with the voice disappears when the transmission is suspended, so that the background noise becomes discontinuous on the receiving side. If the transmission and the transmission pause are repeated frequently within a short time, the discontinuity of the background noise gives the receiving user a hard-to-hear discomfort. This unnaturalness is particularly noticeable in an environment with a large background noise such as a car. In the worst case, the discontinuity may make it impossible to determine the sound.

As one known solution to this problem, when transmission is interrupted, the receiving side synthesizes noise similar to background noise present at the transmitting side (ie, comfort noise). Have been done. As already mentioned, the parameters required to generate comfort noise are determined by the speech coder at the sender (block 608 in FIG. 6) and are transmitted to the receiver by means of SID frames before the transmission is interrupted. Is done. Thereafter, the parameters are repeatedly transmitted at a low speed. By doing so, it is possible to make the comfort noise generated on the receiving side correspond to the change of the actual background noise on the transmitting side.

If the comfort noise parameter determined on the transmitting side appropriately represents the actual audible background noise level and the spectral envelope, high-quality comfort noise can be generated on the receiving side. It is known that These characteristics of the background noise change slightly over time. Therefore, in order to generate good comfort noise, the parameters representing the level of the background noise and the envelope of the spectrum are generated by the speech encoder, and these parameters should be averaged over several speech frames. is necessary. In the discontinuous transmission scheme in the GSM full-rate speech encoder and the extended full-rate encoder (see GSM 06.31 and GSM 06.81), the length of the SID averaging period is 4 and 8 frames, respectively, and 1 The length of the frame is 20 milliseconds.

Before the transmission is interrupted at the end of the voice burst, the optimized noise parameters are determined and included in the first SID frame to be transmitted to the receiver, as described above. , A hangover period is used. The hangover is a period during which the voice state detection device 604 determines that no voice is present (ie, the VAD flag 606 = “0”), and the transmission of the voice frame has not been stopped yet (ie, the SP flag 607 = "1") period. In this regard, FIG.
See also During the hangover period, since the speech state detection device 604 has already determined that there is no speech, the speech frame is guaranteed to contain only noise (no speech), and therefore these By averaging the speech coding parameters of the hangover frame of the hangover frame, a comfort noise parameter can be obtained.

The length of the hangover period is determined by the length of the SID averaging period. That is, the hangover period needs to be long enough to perform the parameter averaging before transmitting the obtained comfort noise parameter in the SID frame. In the discontinuous transmission method of the GSM full-rate speech encoder, the length of the hangover period (the length of the SID averaging period) is determined in order to determine the comfort noise by a method of updating the SID frame using the parameters of the previous frame. )
Four frames are sufficient. On the other hand, in the discontinuous transmission method of the GSM extended full-rate speech encoder, the length of the hangover period is 7 frames (the length of the SID averaging period-1), and the length of the eighth frame of the SID averaging period is The parameters are obtained while the speech encoder is processing the first SID frame. FIG. 7 shows the hangover period and S in the GSM enhanced full-rate speech coding discontinuous transmission scheme.
3 shows the concept of an ID averaging period.

When the hangover period ends, the first SID
The frame is transmitted. The evaluation of the background noise characteristics using the comfort noise calculation algorithm is subsequently performed, and the updated SID frame is sent to the wireless subsystem 611 on a frame basis. The transmission of the SID frame is continuously performed while the voice state detection device 604 determines that no conversation voice exists. When the SID averaging period ends, the discontinuous transmission handler 61 on the transmitting side
2 is a SID averaging period comfort noise calculation algorithm 6
08, the flag 609 notifies that the SID averaging period has ended. The flag 609 is normally in a reset state of “0”, but is set to “1” when an updated SID frame is sent to the wireless subsystem 611. When the flag 609 is set, the comfort noise calculation algorithm 608 performs parameter averaging to create an updated SID frame for use by the wireless subsystem 611.
The updated SID frame is transmitted to the wireless subsystem 611.
And written to the SID memory block 610 as well. The latest SID frame is always stored in the SID memory block 610 and is used in later processing.

When the voice burst ends, the last SI
If the time elapsed since the D frame was sent to the wireless subsystem is 24 frames or less, the last S
The ID frame is repeatedly read from the SID memory 610 and sent to the wireless subsystem 611. The above operation is continued until an updated new SID frame is prepared, that is, until the next SID averaging period ends. In this way, when the short spike-like background noise is mistaken for speech, it is not necessary to insert a hangover period for calculating a new SID frame at the end of the speech burst, thereby improving transmission efficiency. be able to.

FIG. 8 shows, by way of example, a speech burst for the longest possible duration without a hangover. The binary flag 613 indicates that the SID memory 61
0 is used to indicate when to store the updated new SID frame in the SID memory 610 and when to send the newest updated SID frame from the SID memory 610 to the wireless subsystem 611. S
If the P flag 607 is “0”, the SID memory 6
10 determines in each frame whether the SID frame should be stored or transmitted.

In a GSM enhanced full rate speech coder discontinuous transmission system, a binary flag 614 is also needed to inform the noise estimation algorithm of the end of the hangover period. Flag 614 is normally reset to "0", but is raised to "1" for one frame period during which the first SID frame is transmitted after the hangover period after the voice burst.

FIG. 9 shows reception (R) of the discontinuous transmission system.
It is a block diagram which shows the audio | voice decoder in X) side. A series of speech coding parameters 701 input from the radio subsystem 700 of the digital mobile radio unit are processed by the speech decoder 702 for each frame, and a speech signal 703 is synthesized. The obtained audio signal 703 is D /
The signal is supplied to an A (digital-analog) converter 704. D / A converter 704 generates an audible audio signal to the receiving user.

The receiving discontinuous transmission system receives the binary SP flag 705 from the radio subsystem. This flag operates in contrast to the sending SP flag. That is, when a voice frame is received, S
P flag = “1”, SP flag = “0” when SID frame is received or transmission is interrupted
Becomes Also, the binary flag 706 from the wireless subsystem 700 notifies the comfort noise generation algorithm 707 of the presence or absence of the newly received SID frame. That is, this flag is normally reset to "0", but is set to "1" when a new SID frame is received when the SP flag 705 is "0".

When the SP flag 705 = “0” and the transmission is in a pause state, the comfort noise generation block 707 of the speech decoder 702 indicates the characteristic of the background noise on the transmission side included in the SID frame. Based on the data,
Generates comfort noise. During the transmission suspension period,
The updated SID frames are sent at a slower repetition rate, but by determining the decoding comfort noise parameters between these updated SID frames by interpolation, the characteristics of the comfort noise change more smoothly. Can be

In the GSM full-rate speech coding discontinuous transmission system, when a new updated SID frame is calculated and transmitted to the radio subsystem 611 (FIG. 6), parameters representing the characteristics (level and spectrum) of the background noise are used. , SID are averaged during the SID averaging period and then scalar quantized using the same quantization method as in the normal speech coding mode. Similarly, if the SID frame is G
Upon arriving at the SM full-rate speech decoder 702, the unvoiced descriptor parameters are decoded using the same inverse quantization method used for normal speech decoding mode (see, eg, GSM 06.12).

In the GSM extended full-rate speech coding discontinuous transmission system, when calculating a new SID frame, a parameter (LSP) representing the spectrum of background noise is calculated.
The average of the parameter SID during the SID averaging period is calculated, and the quantization of the vector is performed using the same quantization table used to quantize the parameter in the normal speech coding mode. These spectral parameters are inversely quantized by the decoder 702 using the same table as the predictive inverse quantization table used in the normal speech coding mode. When calculating a new SID frame, a parameter (fixed codebook gain) representing the level of background noise is set to SI
After being averaged in the D averaging period, quantization is performed using a scalar prediction quantization table used for quantization in a normal speech coding mode. At the decoder, these parameters are dequantized using the same table as the predictive dequantization table used in the normal speech decoding mode.

However, when adaptively controlling the predictive quantizer, it is difficult to use the above-described quantization method for quantizing the comfort noise parameter to be transmitted by the SID frame. The reason is that it is not possible to maintain frame-by-frame synchronization of the quantizer and inverse quantizer predictors of the encoder and decoder, respectively, due to the transmission being interrupted when speech is lost Because it is. However, it is possible to locally estimate the value of the quantizer predictor in the encoder and the decoder in the same way, as described below. That is, the newest 7
The quantized LSP and fixed codebook gain parameters of the speech frames are stored locally in both encoder 602 and decoder 702. At the end of the hangover period at the end of the voice burst, the average of these stored parameters is determined. The average parameter thus obtained, ie, the reference LSP parameter vector f ^ref , the reference fixed codebook gain g
_c ^ref has the same value in the encoder 602 and the decoder 702. The reason is that in quantization, both the quantized LSP and the fixed codebook gain have the same value in the encoder 602 and the decoder 702 in the normal speech coding mode. The average value of the reference LSP parameter vector f ^ref and the reference fixed codebook gain g _c ^ref is kept frozen until a hangover period after the next voice burst. This frozen average is
Instead of the usual predictor of the quantization algorithm, it is used for quantization of comfort noise parameters.

Returning to FIG. 9, the discontinuous transmission handler 708 on the receiving side receives the SP flag 705 as an input,
If a hangover period occurs after a voice burst, 1
The binary flag 709 is set to “1” for the frame period and output. Note that this flag 709 is reset to “0” in other periods. The flag 709 is used in the discontinuous transmission system of the GSM extended full-rate speech decoder 702 to set the comfort noise generation algorithm 7
07, the reference LSP parameter vector f ^ref by the averaging process and the reference fixed codebook gain g _c ^ref
(G)
SM06.62). For a method of determining the value of flag 709, see Finnish Patent Application No. 95325.
No. 2 and corresponding US patent application Ser.
No. 672,932 (filed Jun. 28, 1996) and International Application No. PCT / FI96 / 00369, all of which are incorporated herein by reference.

As described above, in recent speech encoders, speech coding parameters are quantized using a prediction method. That is, the quantizer attempts to predict the value to be quantized as accurately as possible. In such a predictive quantizer, the difference or quotient between the actual parameter value and the predicted parameter value is usually quantized and transmitted to the receiving side. The inverse quantizer corresponding to the above-described quantization on the receiving side has the same predictor as the quantizer. Therefore, by adding or multiplying the received difference or quotient value to the predicted value, the value of the parameter quantized on the transmission side can be reproduced.

In such a predictive quantizer, usually, an adaptation for updating a predictor after each quantization using a result of the quantization is performed. Both predictors of the quantizer and the inverse quantizer are updated using the reproduced quantization parameter values so that synchronization between them is maintained.

However, it is difficult to use such a quantization method for adapting the predictive quantizer to quantize the comfort noise parameter sent by the SID frame. That is, while the speech is interrupted, the transmission is interrupted, so that the quantizers of the encoder 602 and the decoder 702 and the predictors of the inverse quantizer cannot be synchronized in frame units.

If it is possible to use the same quantization table used by the predictive quantizer to encode normal speech speech for quantization of comfort noise parameters,
That is more desirable. For that purpose, it is necessary to make prediction during discontinuous transmission without performing adaptation. In order for the quantizer to be able to encode parameter values in response to changes in background noise characteristics,
It is necessary that the predictor has a value as close as possible to the value of the current background noise average parameter. Also, it is desired that the same predicted value can be used in both the quantizer and the inverse quantizer.

As already mentioned, in quantizing the comfort noise sent by the SID frame, one way to obtain a good prediction is to hang over the parameter values quantized by the normal speech coding mode. During the period, the average of the stored quantized parameter values is determined at the end of the hangover period, and the average of the predictors thus obtained is frozen until the next hangover period. It is to keep it. However, this method has the problem that in GSM or similar discontinuous transmission schemes, the speech decoder 702 cannot know when the last hangover period of the speech burst has occurred.

In view of the above problems, it is an object of the present invention to provide a method for informing a speech decoder 702 of the occurrence of the last hangover period of a speech burst. The above object is preferably achieved by the speech encoder 60
2 to the speech decoder 702 by transmitting the hangover period information as side information in a SID frame (ie, comfort noise parameter message).

This method of the present invention will be further described with reference to FIG. In the scheme in FIG. 10, the binary flag 709 does not use the value generated by the receiving side discontinuous transmission handler, but uses the SI over the transmission channel.
The value sent from the encoder 602 by the D frame is used. Therefore, when the inverse quantization is performed by the prediction method of the present invention, the flag 709 is locally used in the decoder 702.
Need not occur, and therefore the discontinuous transmit handler block 708 at the receiving end is no longer needed. In this scheme of the invention, the flag 709 is raised to "1" in the first SID frame after the hangover period. However, if there is no hangover period preceding the first SID frame, the flag 709 of the first SID frame is reset to “0”.
The flag 709 is always reset to “0” in the second and subsequent SID frames during the comfort noise insertion period.

An advantage of this method of the present invention is that the speech decoder does not require a discontinuous transmission handler 708 to locally determine whether a hangover period exists at the end of a speech burst. Therefore, the program instruction used by the discontinuous transmission handler 708 on the receiving side becomes unnecessary, and the calculation load on the audio decoder 702 is reduced.

Another advantage of the present invention with respect to the method of providing information about the presence of a hangover period to decoder 702 is that each time the hangover period ends, a pseudo-noise excitation signal is generated at encoder 602 and decoder 702. It is possible to reinitialize the generator synchronously. Yet another advantage of the method of the present invention for providing information about the presence of a hangover period to the decoder 702 is that the received comfort may be dependent on whether the hangover period is present or absent at the end of the speech burst. The ability to interpolate the noise parameters using a variety of methods, resulting from the staircase changes in comfort noise level or spectrum that can occur when speech bursts are short. The perceptual unnaturalness of the comfort noise can be reduced.

Before describing the present invention in further detail, FIG.
A wireless user terminal or mobile station 10 (eg, but not limited to a cellular phone or a mobile phone) to which the present invention is applicable will be described with reference to FIGS. The mobile station 10 has an antenna 12 for transmitting and receiving signals to and from the base station 30. Base station 30 is part of a cellular communication network that includes, for example, a base station / mobile switching center (BMI) 32 including a mobile switching center (MSC) 34. The MSC 34 connects the mobile station 10 to a ground line when the mobile station 10 performs a call.
In the present invention, it is assumed that the mobile station 10 is on the transmitting side, while the ground station is on the receiving side. It is also assumed that the base station 30 is provided with a suitable receiver and speech decoder for receiving and processing the encoded speech parameters and the discontinuous transmission mode comfort noise parameters as described below.

The mobile station comprises a modulator (MOD) 14A, a transmitter 14, a receiver 16, a demodulator (DEMOD) 16A, and a transmitter 14, a receiver 16, and a controller 18 for transmitting and receiving signals. The signal includes signal information corresponding to the radio interface standard of the cellular system and the voice of the user and / or the data generated by the user. In the present invention, it is assumed that the radio interface standards include standards for physical and logical frame structures. However, the present invention
It is not limited to these specific physical and logical frame structures. Also, the present invention relates to a mobile station having IS-136 or a similar compatible mobile station or TDM.
However, the present invention is not limited only to the application to the A-type system. Also, in the present invention, it is assumed that the radio interface standard supports discontinuous transmission mode operation.

The controller 18 also includes circuits for implementing the necessary voice and logic functions in the mobile station. For example, the controller 18 includes a digital signal processor, a microprocessor, an A / D converter, a D /
It consists of an A-converter and other various circuits. Mobile station control and signal processing functions are assigned to these devices according to their capabilities. It is assumed that controller 18 has the speech coder and other features required in the improved method and apparatus for generating comfort noise in discontinuous transmissions according to the present invention. All of these functions can be realized by software, and all can be realized by hardware. Alternatively, it can be realized using both hardware and software.

As a user interface, a commonly used earphone or speaker 17, A /
Microphone 1 combined with D converter and speech coder
9 and a user input device such as a delay circuit 20 and a keypad 22.
These are all connected to the controller 18. The keypad 22 includes the normal numbers (0-9) and associated keys (#, *) 22a required to operate the mobile station 10.
And other keys 22b. The other keys 22b include, for example, a “transmit” key, various menu scroll keys, soft keys, a power switch key, and the like. Further, the mobile station 10 has a battery 26 for supplying power to various circuits necessary for the operation of the mobile station.

The mobile station 10 has various memories, and these memories are collectively represented as a memory 24. These memories store various constants and variables required by the controller 18 in the operation of the mobile station. For example, memory 24 includes various cellular system parameters and number assignment modules (NAMs).
Is stored. Further, the memory 24 includes a controller 18.
Is also stored (usually, R
OM). Further, the memory 24 stores the BMI3
It is also used to store data such as a user message received from 2 before displaying it on the display. Also, in the discontinuous transmission operation, a routine for executing the method described below for transmitting the comfort noise parameter is stored in the memory 24.

[0109] The mobile station 10 may be a device mounted on a vehicle or a portable device. The mobile station 10 may be based on various radio interface standards, modulation schemes, and access schemes. For example, the mobile station is not limited to IS-136, but may be based on any other standard, such as the GSM standard. That is, it will be apparent that the present invention is not limited to a particular type of mobile station or a particular radio interface standard.

Next, as a specific example, the present invention will be described with reference to IS-
An embodiment in which the present invention is applied to 136 will be described. However, it is repeatedly described here that the present invention is not limited only to the radio interface standard.

In the discontinuous transmission scheme on the digital traffic channel (IS-136.1, Rev. A, section 2.3.11.2), in the discontinuous transmission high power state, the transmitter 14 is moved. The station 10 transmits at the power level specified by the most recent power control command received (initial traffic channel specification message, digital traffic channel (DTC) specification message, handoff message, dedicated DTC handoff message, or physical layer control message). Do.

On the other hand, in the discontinuous transmission low output state,
The transmitter is kept off. CDVCC is not transmitted, except for Fast Associated Control Channel (FACCH) messages. In the discontinuous transmission low output state, the mobile station 1
0 from all low-speed associated control channels (SA
CCH) message is transmitted as a FACCH message, after which the transmitter 14 transmits a discontinuous transmission (DTX)
If is not prohibited, the state returns to the off state.

When the mobile station 10 switches from the discontinuous transmission high power state to the discontinuous transmission low power state, the mobile station 10 completes all the SACCH messages in progress in the high power state, or interrupts the SACCH message transmission. Then, the entire interrupted SACCH message is retransmitted again as a FACCH message in a discontinuous transmission low power state.

In the transition state when the mobile station switches from the discontinuous high output state to the discontinuous transmission low output state, the transmission output is discontinuous transmission until the transmission of the FACCH message being transmitted is completely completed. Maintained at a high output level.

In the preferred embodiment of the present invention, mobile station 10 terminates transmission of all comfort noise blocks (including six DTX hangover slots and associated comfort noise parameter messages). Until you do, it is in a transition state. The comfort noise block is transmitted without interruption. Any other F
If the ACCH message slot occurs simultaneously with the transmission of the comfort noise block, the mobile station 10
Delay either the CH message or the comfort noise block so that one is sent first and the other is sent later. In either case, the FACCH messages are grouped or separated so as not to interrupt or enter the slots used to transmit comfort noise blocks. This allows the base station's speech / comfort noise decoder to generate the best quality comfort noise.

In this regard, co-assigned, co-pending US patent application Ser. No. 08 / 936,755 entitled "Transmission of comfort noise in discontinuous transmission" (September 2, 1997).
5th filing, Sepo Alanara and Pekka Kapanen)
checking ...

According to a specific embodiment, the comfort noise (CN) parameter message, shown in Table 1 below, is transmitted on a feedback digital traffic channel (RDTC), more specifically on a FACCH logical channel. You.
The comfort noise parameter message consists of 38 bits, of which 26 bits are quantized using the same split vector quantization (SVQ) codebook used in the IS-641 speech codec.
Used to represent the residual vector. Note that the quantization / inverse quantization algorithm of the audio codec has been modified so that the codebook can be used. LS
The F parameter represents the spectral envelope of the background noise on the transmitting side, preferably using a 10th order LPC model of the spectrum.

The next 8 bits are used as a comfort noise energy quantization index representing the energy of the background noise on the transmitting side. The remaining four bits of the message are used to send a random excitation spectrum control (RESC) information element.

[0119]

[Table 1]

That is, the present invention is to solve the above-mentioned problem in the prior art by generating synthetic noise similar to background noise existing on the transmitting side on the receiving side. Before interrupting the transmission of the conversation voice, the comforting noise (CN) parameter is calculated on the transmitting side and transmitted to the receiving side. Thereafter, comfort noise parameters are transmitted at a constant slow repetition rate.
According to this method, it is possible to adapt the comfort noise to a change in noise on the transmission side.
The discontinuous transmission method according to the present invention evaluates acoustic background noise using the voice state detection (VAD) function 21 (FIG. 12) on the transmitting side and the controller 18 on the transmitting side.
A function of transmitting the characteristic parameter of the background noise to the receiving side,
A function of generating noise similar to background noise on the transmitting side, called comfort noise, on the receiving side while transmission of conversational speech is interrupted is required.

In addition to these functions, if it is determined that a parameter that has reached the receiving side has a significant error, the data is replaced, and voice or comfort noise is generated from the replaced data. In order to prevent the user on the receiving side from feeling uncomfortable.

Further, by the discontinuous transmission function on the transmission side, the traffic frame marked with the flag SP is continuously transmitted from the radio transmitter 14. Here, when the SP flag = "1", it indicates that the frame is a voice frame, and when the SP flag = "0", a set of comfort noise parameters encoded by the frame is transmitted. Indicates that The timing of transmitting a frame to the wireless interface is controlled by the wireless transmitter 14 based on the SP flag.

In the preferred embodiment of the present invention, before resetting the mobile station 10, all frames are infinitely long voice frames to enable accurate verification of the discontinuous transmission function of the transmitting side. Treated as if it were. Therefore, the first six frames after reset are always marked with the SP flag = "1", even if the VAD flag = "0" (ie, the hangover period, see Fig. 14).

The voice state detector (VAD) 21 continuously monitors whether or not the signal input from the microphone 19 includes speech voice, and for each frame, a binary flag (VAD flag = “1” or VAD flag =
"0") is output.

The VAD flag indirectly controls the entire discontinuous transmission operation on the transmission side via the transmission-side discontinuous transmission handler operation described below.

If the VAD flag = "1",
The speech coded output frame with the SP flag = “1” is directly supplied to the wireless transmitter 14.

At the end of the voice burst (transition from VAD flag = "1" to VAD flag = "0"), the next seven consecutive frames are used to generate an updated new set of CN parameters. used. Normally, the first six speech encoded output frames after the end of the speech burst are supplied directly to the radio transmitter 14 with the SP flag = "1", thereby forming a "hangover period". You. Subsequently, as the seventh frame after the end of the voice burst, the first new set of CN parameters is supplied to the wireless transmitter 14 with the SP flag = "0" (see FIG. 14).

However, if the time elapsed since the last CN parameter was calculated and supplied to the radio transmitter 14 at the end of the voice burst is 24 frames or less, the newly updated pair CN parameters (VA
The immediately preceding CN parameter is repeatedly supplied to the wireless transmitter 14 until the D flag = "successive seven frames to which" 0 "is added) is ready. By doing so, the spike-like short background noise is erroneously interpreted as speech voice, and "hangover" occurs to start the calculation of the CN parameter. Can be reduced. FIG. 15 shows, by way of example, a speech burst for the longest possible duration without a hangover.

After the first set of CN parameters after the voice burst is calculated and supplied to the radio transmitter 14, as long as the VAD flag is “0”, the discontinuity of the transmitting side continues. The transmission handler calculates the updated CN parameter and adds the SP flag = “0” to the wireless transmitter 1.
4

[0130] If the SP flag = "1", the speech coder operates in the normal speech coding mode, while
When the SP flag is “0”, the operation in the simple mode is performed because not all of the encoding functions are required to calculate the CN parameter.

The wireless transmitter 14 transmits traffic frames in the following order. That is, all S
P flag = frame with "1"; SP flag = first SP flag that appears after one or more frames with "1" have been transmitted = frame with "0"; SP flag = A frame that is “0” and is used for transmitting a CN parameter update message.

Thus, when the speaker stops speaking, the CN parameter message is transmitted, and the state transits to the discontinuous transmission low output state. While the conversation voice is interrupted, transmission is restarted at regular intervals, and a CN parameter message for updating comfort noise generated on the receiving side is transmitted.

[0133]

[Outside 1]

[0134]

[Outside 2]

[0135]

[Outside 3]

[0136]

[Outside 4]

The CN parameter message is always transmitted before the end of the voice burst and before the wireless transmission is interrupted.
Therefore, the CN parameter message also plays a role of initializing the generation of comfort noise on the receiving side.

The timing at which the CN parameter message and the voice frame are transmitted on the radio transmission path has already been described with reference to FIGS.

The evaluation of the background noise involves determining the average of three different types of average parameters, namely the LSF parameter, the random excitation signal gain parameter, and the RESC parameter. The comfort noise parameter encoded in the comfort noise parameter message consists of N = 7 consecutive frames and during the comfort noise averaging period with VAD = "0", as described in detail below. Is calculated.

Before calculating the average value of the LSF parameters during the comfort noise averaging period, first, a median replacement process is performed on a series of LSF parameters whose average value is to be calculated, and the characteristic of the background noise on the transmission side is obtained. Remove parameters that do not represent well. First, another LSF of the LSF parameter vector f (i) in the comfort noise period
Parameter vector f (j) (i = 0... 6, j = 0
.. 6 (i ≠ j)), and approximate using the following equation.

[0141]

[Equation 17]

Here, f _i (k) is the LS of frame i.
This is the k-th LSF parameter of the F parameter vector f (i).

All other frames j = 0 of the LSF parameter vector f (i) within the comfort noise averaging period
.. LSF parameter vector f of N (j ≠ i)
To determine the spectral distance ΔS _i for (j),
The sum of the spectral distances ΔR ^ij is obtained for all i = 0... 6 (j ≠ i) according to the following equation.

[0144]

(Equation 18)

All the LSFs within the comfort noise averaging period
Among the parameter vectors, the minimum spectral distance ΔS
^The LSF parameter vector f (i) having ⁱ is adopted as the central LSF parameter vector f _med of the averaging period, and its spectral distance is represented by ΔS _med . The central LSF parameter vector can be considered to be the LSF parameter vector that best represents the short-term spectral characteristics of the background noise among all the LSF parameter vectors within the averaging period. Next, the LSF parameter vector f within the comfort noise averaging period
It is determined whether any of (j) satisfies the following expression (6).

[0146]

[Equation 19]

Here, TH _med = 2.25 is a median replacement threshold value. If there is an LSF parameter vector that satisfies the above equation, before calculating the average of the LSF parameter vector f _mean , a maximum of two LSF parameter vectors (that is, from the top of the LSF vector larger than TH _med ) are calculated. 2) with the central LSF parameter vector.

A set of LSF parameter vectors obtained as a result of the above-described median value substitution processing is represented by f ′ (ni).
Where n is the index of the current frame and i
Is an averaging period index (i = 0... 6).

When the median replacement processing is performed at the end of the hangover period (first comfort noise updating processing), the immediately preceding 6
One frame (i = 1 ... 6 within the hangover period)
All LSF parameter vectors f
(Ni) has a quantized value, while the LSF parameter vector f (n) of the newest frame n has an unquantized value. In the subsequent updating of the comfort noise, the LSF parameter vector of the comfort noise averaging period of the frame overlapping the hangover period has a quantized value, but the parameter vector of the newest frame of the comfort noise averaging period is It has an unquantized value. If the period of the 7 most recent frames does not overlap with the hangover period, the LSF
The median parameter replacement processing is performed using only the parameter values that are not quantized.

The average LSF parameter vector f ^mean in frame n is calculated using the following equation.

[0151]

(Equation 20)

Here, f ′ (n−i) is the newest seven frames (i = 0 ·
.. 6), i is an averaging period index, and n is a frame index.

The average LSF parameter vector f ^{mean for} frame n is preferably the same quantization table that the speech coder uses to quantize the unequalized LSF parameter vector in the normal speech coding mode. Is quantized using However, a quantization algorithm modified so as to be suitable for quantizing comfort noise is used. LSF to quantize
The prediction residual is obtained according to the following equation.

[0154]

(Equation 21)

[0155]

[Outside 5]

[0156]

[Outside 6]

[0157]

(Equation 22)

[0158]

[Outside 7]

[0159]

[Outside 8]

Further, the random excitation signal gain for each subframe is calculated as follows based on the LP residual signal energy of the subframe.

[0161]

(Equation 23)

Here, g _cn (j) is the calculated value of the random excitation signal gain of subframe j, r (l) is the l-th LP residual signal sample of subframe j, and l is the sample index (l = 0... 39). In Equation (10), 1.286 is used as a scaling factor to match the level of comfort noise with the level of background noise encoded by the speech codec. However, the invention is not limited to this particular value of the scaling factor.

Since the subframe excitation signal (pseudo noise) during the comfort noise generation period includes 10 samples taking values other than zero between −1 and +1, the calculated value of the energy of the LP residual signal is calculated as follows. By dividing by 10, the energy per random excitation signal pulse can be determined.
The calculated value of the random excitation signal gain is averaged, and when an updated CN parameter is needed, SP = "0"
Are updated in the first subframe of each frame n marked with.

[0164]

(Equation 24)

Here, g _cn (n) (1) is the calculated value of the random excitation signal gain in the first subframe of frame n, and g _cn (n−i) (j) is one past frame (i =
1... 6) are calculated values of the random excitation signal gain in subframe j. Since the random excitation signal gain of only the first sub-frame of the current frame is used in the averaging process, once the processing of the first sub-frame of the current frame is completed, an updated set of comfort noise parameters is obtained. It becomes possible to transmit.

[0166]

[Outside 9]

In calculating the RESC parameter, LP
Since the residual r (n) deviates somewhat from the flat spectrum characteristic, if the comfort noise is synthesized on the receiving side using a random excitation signal having a flat spectrum, the quality of the comfort noise deteriorates (background noise). And the comfort noise). To better match the spectral characteristics, during the comfort noise averaging period, L
A second-order LP analysis is further performed on the P residual signal, and the obtained average LP coefficient is included in the comfort noise parameter message and transmitted to the receiving side. Make it happen. This method is called random excitation spectrum control (RESC). The LP coefficient obtained by this method is represented by the RESC parameter Λ.

The LP residual signal r (n) of each subframe of the frame is concatenated so that the autocorrelation r _res (k) (k = 0... 2) of the LP residual signal of the 20 ms frame is It is calculated by the following equation.

[0169]

(Equation 25)

After calculating the autocorrelation using the above equation,
The obtained autocorrelation is normalized to obtain a normalized autocorrelation r ′
_{Find res} (k).

The processing of the first subframe of the current frame is performed by averaging the newest frame during the comfort noise averaging period using the autocorrelation obtained from only the first subframe. Is completed, it is possible to prepare updated comfort noise parameters to be sent next.

When it is necessary to update the comfort noise parameter, the average of the calculated values of the normalized autocorrelation is calculated using the following equation, and the first value of each frame n with SP = "0" is added. Update in subset.

[0173]

(Equation 26)

Here, r ′ _res (n) (1) is the normalized autocorrelation in the first subframe of frame n,
r ′ _res ( _ni ) is a past frame (i = 1...)
6) Normalized autocorrelation of one frame, where n is a frame index.

[0175]

[Outside 10]

[0176]

[Outside 11]

[0177]

[Outside 12]

The comfort noise coding algorithm generates 38 bits of data as shown in Table 2 for each comfort noise parameter message. In addition, these 38
Bit data is represented by a vector cn [0 ... 37]. The comfort noise bit data cn [0 ... 37]
The signals are supplied to the FACCH channel encoder in the order shown in Table 2 (without any special bits being aligned).

[0179]

[Table 2]

Regardless of the content (voice, comfort noise parameter message, other FACCH message, or no signal), the radio receiver of the base station 30 performs various processing functions according to the received three flags. Each of the designated traffic frames is sequentially supplied to the receiving side discontinuous transmission handler. The three flags are a bad speech frame indication (BFI) flag, a bad comfort noise parameter indication (BFI_CN) flag,
The comfort noise update flag (CNU), which is described below in Table 3 and the like. These flags are used to distinguish traffic frames by their purpose. This division allows the receiving discontinuous transmission handler to easily know how to process the received frame, as shown in Table 3.

[0181]

[Table 3]

The binary BFI and BFI_CN flags allow the traffic frame to contain meaningful information bits (BFI flag = “0” and B
FI_CN flag = “1”) or does not include a meaningful information bit (BFI flag = “1”)
And whether the BFI_CN flag is “1” or the BFI flag is “0” and the BFI_CN flag is “0”. In the present invention, since the FACCH frame is not considered to contain significant bits, except when it contains a comfort noise parameter message, the BFI SP flag = "1" and B
FICN flag = "1".

The binary CNU flag marks CNU = "1" for traffic frames aligned according to the transmission channel quality information sent by the FACCH.

The discontinuous transmission handler on the receiving side relates to the entire discontinuous transmission operation on the receiving side. The discontinuous transmission operation on the receiving side is performed as follows. If the speech frame is determined to be a good speech frame, the discontinuous transmission handler supplies the speech frame directly to the speech decoder.
When it is determined that the voice frame or the comfort noise parameter message is missing, the replacement and muting processing is performed. If an appropriate comfort noise parameter message frame is detected, the comfort noise is detected using this message to detect the next comfort noise parameter message (CNU = “1”) or the next good speech frame. Continue until done. While performing the above processing, the receiving-side discontinuous transmission handler ignores any frame supplied from the wireless receiver if the frame cannot be used. Also, the following operations are optionally possible. That is, when the first missing comfort noise parameter message is detected, the parameter is replaced with the parameter of the last suitable comfort noise parameter message that was detected last, and a normal Perform processing. If the second missing comfort noise parameter message is detected, a muting process is performed.

In the LP parameter averaging process and the decoding process, when the decoder receives a speech frame, the LP parameters of the latest six speech frames are stored in the memory. The decoder counts the number of frames received since the encoder last updated the comfort noise parameter and provided it to the wireless transmitter. Based on this count, the decoder determines whether there is a hangover period at the end of the speech burst (when the first comfort noise parameter message after the speech burst is received, If at least 30 frames have elapsed since the parameter was updated, it may be considered that there was a hangover period at the end of the voice burst).

[0186]

[Outside 13]

The averaging process for obtaining the above reference parameters is performed as follows.

Upon receiving a voice frame, the LSF parameters are decoded and the result is stored in memory. If a comfort noise parameter message is received and it is determined that there is a hangover period at the end of the speech burst, the stored LSF parameters are rewritten as in the speech encoder as follows: Average.

[0189]

[Equation 27]

[0190]

[Outside 14]

[0191]

[Outside 15]

[0192]

[Equation 28]

[0193]

[Outside 16]

In each sub-frame, the fixed codebook excitation signal vector containing four non-zero pulses of a normal speech decoder is a random excitation containing ten non-zero pulses during speech interruption. Replaced with the signal vector. The pulse position of the random excitation signal and its sign are locally generated using a uniformly distributed pseudorandom number. The excitation signal pulse in the random excitation vector can take +1 and -1 as its value. The random excitation signal generation algorithm is executed according to the following pseudo code.

For (i = 0; i <40; i ++) code (i) = 0; for (i = 0; i <10; i ++) ｛j = random (4); idx = j * 10 + i; if ( random (2) == 1) code (idx) = 1; else code (idx) = − 1; ｃｏ where code [0... 39] is a fixed codebook excitation signal buffer, and random ( k)
Is a pseudorandom integer uniformly distributed in the range [0... K-1].

The received RESC parameter index is decoded and the received RESC parameter λ (i) (i =
1, 2) are obtained. The generated random excitation signal is filtered by a RESC synthesis filter having a characteristic represented by the following equation.

[0197]

(Equation 29)

Preferably, the RESC synthesis filter is implemented using a lattice filter. After passing through the RESC synthesis filter, the random excitation signal is subjected to scaling processing and then filtered by the LP synthesis filter.

The generation of comfort noise is performed using a method in which the speech decoding algorithm is changed as follows. That is, the value of the fixed codebook gain is replaced with the value of the random excitation gain obtained by the comfort noise parameter message, and the fixed codebook excitation signal is replaced with the locally generated random excitation signal, as described above. . Also, the random excitation signal is filtered by the RESC synthesis filter as described above. The value of the adaptive codebook gain for each subframe is set to zero. The pitch delay value of each subframe is set to, for example, 60. As the LP filter parameter, the parameter received by the comfort noise parameter message is used. The predictor memory used for normal LP parameter quantization and fixed codebook gain quantization algorithm is reset when SP flag = "0" and when the speech is restarted again, the quantizer is in its initial state. The operation can be started again from. When the speech decoder performs a normal operation using the above parameters, comfort noise is synthesized. Each time a reasonable comfort noise parameter message is received, the comfort noise parameters (random excitation signal gain, RESC parameters, LP filter parameters) are updated as described above. In updating the comfort noise, the above parameters are interpolated during the comfort noise update period to make the change smoother. Unusable frames received when the receiving discontinuous transmission handler is experiencing comfort noise are referred to as missing comfort noise parameter messages. If such a frame is received, a comfort noise message is inferred (comfort noise update flag CNU = "1").

If such a loss occurs in a single comfort noise parameter message, that parameter is replaced by the parameter of the last valid comfort noise parameter message, and Processing is performed. Further, when the comfort noise parameter message is lost, the comfort noise is muted, the output level is gradually reduced (−3 dB / frame), and finally the output of the decoder is set to zero. The muting is performed by reducing the random excitation signal gain to 0 at a constant rate of -3 dB per frame. This value is maintained as it is when further loss occurs.

As described above, specific values such as the frame length and the number of frames are assumed, and the specific message type (F
ACCH), the number of frames, the length of a frame, the length of a hangover period, the length of an averaging period, the type of a message, and the like are specified in a specific digital mobile communication system specification. Various changes can be made as required. In the above description, FIGS.
The present invention has been described with reference to circuit block diagrams such as (b), 3 (a), 3 (b), 4, 5, 10, etc.
It can also be implemented using a suitably programmed digital data processor (eg, controller 18 of FIG. 12 is an example) that is included as part of zero. As one of the examples, for example, the selector 307,
Although 319 and 410 are shown as switches in FIGS. 4 and 5, these selectors can also be implemented completely in software.

In some systems, when generating comfort noise, the comfort noise parameter message (SID frame) used for transmitting the RESC parameter from the transmission side to the reception side may include a spare bit. is there. In such a case, the RESC filter of the present invention can be replaced with a synthesis filter having fixed coefficients. The fixed filter coefficients are optimized such that the frequency response characteristics of the synthesis filter are the average frequency response of a normal RESC filter determined by the value to which the coefficients are transmitted. Alternatively, the filter coefficients may be selected such that the filter has a frequency response characteristic suitable for obtaining perceptually favorable comfort noise.

The present invention has been described with reference to specific embodiments. However, the present invention is not limited to these embodiments, and various modifications can be made without departing from the scope and spirit of the present invention. It will be obvious.

[Brief description of the drawings]

FIG. 1a is a block diagram showing a conventional circuit for generating a comfort noise parameter on the transmitting side.

FIG. 1b is a block diagram showing a decoder conventionally used to generate comfort noise on the receiving side.

FIG. 1c shows the spectrum of the signal in some parts of the conventional decoder of FIG. 1 (b).

FIG. 1d shows the averaging block of FIG. 1 (a) in more detail.

FIG. 2a is a block diagram illustrating a circuit for generating a comfort noise parameter at a transmitting side according to the present invention;

FIG. 2b is a block diagram illustrating a circuit for generating comfort noise on the receiving side according to the present invention.

FIG. 2c is a diagram showing a spectrum in the decoder of FIG. 2 (b).

FIG. 3a is a block diagram illustrating a circuit for generating a comfort noise parameter at a transmitting side according to a second embodiment of the present invention;

FIG. 3b is a block diagram illustrating a decoder at a receiving side according to a second embodiment of the present invention;

FIG. 4 is a block diagram showing a circuit for calculating a comfort noise parameter at a transmitting side of a discontinuous transmission digital communication system according to an embodiment of the present invention;

FIG. 5 is a block diagram showing a circuit for calculating a comfort noise parameter at a transmission side of a discontinuous transmission digital communication system according to another embodiment of the present invention;

FIG. 6 is a block diagram of a conventional speech encoder.

FIG. 7 is a timing chart showing an example of the output of the conventional speech encoder of FIG. 6;

8 is a timing chart showing another example of the output of the conventional speech encoder of FIG.

FIG. 9 is a block diagram showing a conventional audio decoder.

FIG. 10 is a block diagram illustrating a speech decoder according to an embodiment of the present invention.

FIG. 11a shows a typical frequency response of a RESC filter.

FIG. 11b shows a typical frequency response of a RESC filter.

FIG. 11c shows a typical frequency response of a RESC filter.

FIG. 11d shows a typical frequency response of a RESC filter.

FIG. 11e shows a typical frequency response of a RESC filter.

FIG. 11f shows a typical frequency response of a RESC filter.

FIG. 11g shows a typical frequency response of a RESC filter.

FIG. 12 is a diagram illustrating a mobile station suitable for implementing the present invention.

FIG. 13 illustrates a mobile terminal suitable for implementing the present invention for wireless communication with a ground station.

FIG. 14 is a diagram showing a normal hangover procedure;
elapsed represents the number of frames after the comfort noise (CN) parameter was last updated, and hangover occurs only when Nelapsed is 24 or more.

FIG. 15 is a timing chart showing processing of a short voice burst when Nelapsed is smaller than 24;

[Explanation of symbols]

 REFERENCE SIGNS LIST 100 audio signal 101 linear predictive coding (LPC) analysis block 103 inverse filter 104 residual signal 105 excitation gain calculation block 107, 108 averaging block 110 random excitation signal (RE) generator 112 voice synthesis filter 200 random excitation linear prediction code Analysis block 202 Random excitation control inverse filter

 ──────────────────────────────────────────────────の Continuing on the front page (72) Inventor Pekka Kapanen Finland, 33720 Tampere, Naitteliyankatu 27 A14 (72) Inventor Besa Ruoppira Finland, 33720 Tampere, Birupland Polk 20 Be (72) Inventor Jani Rotola-Pukkira Finland, Tampere, Finn-33820, Mannicon Katsu 3A 22

Claims (50)

[Claims] 1. A method for generating comfort noise (CN) in a digital mobile terminal using a discontinuous transmission scheme, the method comprising: calculating a random excitation spectrum control (RESC) parameter when speech is cut off; Transmitting the RESC parameter to the receiver together with other CN parameters; receiving the RESC parameter; shaping the spectrum content of the excitation signal using the received RESC parameter; Inputting. 2. The method of claim 1, wherein calculating the RESC parameter comprises analyzing the residual signal at a speech coder. 3. The method of claim 2, wherein the speech encoder performs an LPC analysis process, wherein the analyzing is of a lower order than the LPC analysis process. 4. The speech coder performs an LPC analysis process of an order larger than the second order, and the analyzing includes:
3. The method of claim 2, wherein the method is performed by second or second order LPC analysis. 5. The step of calculating a RESC parameter comprises: analyzing a residual signal in a speech encoder to generate a spectrum parameter; and averaging the spectrum parameter over a plurality of frames.
The method of claim 1, comprising generating a parameter. 6. The method of claim 5, wherein the number of said plurality of frames is about 10 or more. 7. The step of calculating the RESC parameter comprises: inputting an LPC residual signal from a speech encoder inverse filter to a RESC inverse filter H _RESC (z);
The method of claim 1, comprising generating a residual signal whose spectrum is controlled to have a generally flat spectrum as compared to the spectrum of the residual signal. 8. The RESC inverse filter H _RESC (z) is given by the following equation: (However, b (i) is a filter coefficient (i = 1,...,
The method according to claim 7, wherein the filter is an all-zero filter represented by R). 9. The method of claim 7, further comprising determining an excitation signal gain from said spectrally flattened residual signal. 10. The shaping step includes the steps of: forming an excitation signal by generating a white noise excitation signal sequence; and scaling the generated white noise sequence to generate a scaled noise sequence. The method of claim 1, comprising: processing the scaled noise sequence with a RESC filter to generate an excitation signal having a desired spectrum. 11. The step of calculating the RESC parameter comprises: converting the LPC residual signal from a speech coder inverse filter using the following equation: (However, b (i) is a filter coefficient (i = 1,...,
R) having an all-zero filter characteristic expressed by
_{Inputting the} SC inverse filter H _RESC (z) to generate a residual signal whose spectrum is controlled to have a spectrum that is generally flatter than the spectrum of the LPC residual signal. Generating an excitation signal by generating a white noise excitation signal sequence; scaling the generated white noise sequence to generate a scaled noise sequence; Processing by a filter to generate an excitation signal having a desired spectrum, wherein the RESC filter operates in the inverse of the operation of the RESC inverse filter, ie, the following equation: The method of claim 1, wherein the operation represented by 12. A filter coefficient b (i) (i = 1,...)
, R) defining the RESC parameter r _mean (i)
12. (i = 1,..., R) are transmitted as part of the CN parameter, and are used by the RESC filter to weight the excitation signal spectrum of the synthesis filter.
The described method. 13. An apparatus for generating comfort noise in a system including a digital mobile terminal performing discontinuous communication with a communication network, wherein the digital mobile terminal performs random excitation spectrum control when speech is interrupted. (RESC) parameter is calculated, and this RESC parameter is
Means for transmitting to the receiver of the communication network together with the parameters, and means for shaping the spectral content of the synthesized signal before inputting the excitation signal to the synthesis filter using the received RESC parameters in the communication network. Equipment including. 14. The apparatus according to claim 13, wherein said calculating means analyzes the residual signal in a speech coder. 15. The method of claim 14, wherein the speech encoder performs an LPC analysis process, wherein the analysis is of a lower order than the LPC analysis process. 16. The speech coder performs an LPC analysis process on an order greater than the second order, wherein the analysis is a first order or a second order.
15. The method of claim 14, wherein the method is performed by the following LPC analysis. 17. The RES by calculating the residual signal in a speech encoder and averaging spectral parameters over a plurality of frames.
14. The apparatus according to claim 13, comprising means for generating a C parameter. 18. The method according to claim 18, wherein the number of the plurality of frames is about ten.
18. The device according to claim 17, which is above. 19. The calculating means inputs the LPC residual signal output from the speech encoder inverse filter to a RESC inverse filter H _RESC (z), and _generates a spectrum that is generally flatter than the spectrum of the LPC residual signal. 14. The apparatus of claim 13, wherein the apparatus is adapted to generate a residual signal whose spectrum is controlled to have. 20. The RESC inverse filter H _RESC (z)
Is the following equation: (However, b (i) is a filter coefficient (i = 1,...,
R)).
The described device. 21. The apparatus of claim 19, further comprising: means for determining an excitation signal gain from said flattened residual signal. 22. A means for forming an excitation signal by generating a white noise excitation signal sequence, wherein said shaping means generates a excitation signal by generating a white noise excitation signal sequence, and means for generating a scaled noise sequence by scaling the generated white noise sequence. Means for processing the scaled noise sequence with a RESC filter to generate an excitation signal having a desired spectrum. 23. The calculating means: converting the LPC residual signal from a speech coder inverse filter by the following equation: (However, b (i) is a filter coefficient (i = 1 ... R)
Which has an all-zero filter characteristic represented by
_Means for inputting to an inverse filter H _RESC (z) to generate a residual signal whose spectrum is controlled to have a spectrum that is generally flatter than the spectrum of the LPC residual signal, wherein the shaping element stage has a white color. Means for forming an excitation signal by the generation of a noise excitation signal sequence; means for scaling the generated white noise sequence to generate a scaled noise sequence; Means for processing to generate an excitation signal having a desired spectrum, wherein said RESC filter operates in a manner opposite to that of a RESC inverse filter, ie, the following equation: 14. The apparatus according to claim 13, wherein the apparatus performs an operation represented by: 24. A filter coefficient b (i) (i = 1,...)
, R) defining the RESC parameter r _mean (i)
24. The apparatus of claim 23, wherein (i = 1 ... R) is transmitted as part of the CN parameter, using which the RESC filter weights the spectrum of the excitation signal of the synthesis filter. 25. A method for generating comfort noise (CN) in a digital mobile terminal using discontinuous transmission, comprising the steps of: buffering a set of speech coding parameters when speech is interrupted; In the averaging period, replacing the speech coding parameters in the set that do not well represent the background noise with the speech coding parameters that favorably represent the background noise, and averaging the set of the speech coding parameters. Seeking. 26. The method according to claim 26, wherein the permuting step included in the step includes: measuring a distance between speech coding parameters between individual frames in the averaging period; Identifying a speech coding parameter having a maximum distance from the parameter; and if the distance is greater than a predetermined threshold, averaging the identified speech coding parameter with the averaging. Replacing the speech coding parameter with the smallest measured distance to other speech coding parameters within the time period. 27. The permuting step included in the step includes: measuring a distance between speech coding parameters between individual frames in the averaging period; and performing another step in the averaging period. Identifying a speech coding parameter whose distance from the parameter is the largest; and if the distance is greater than a predetermined threshold, the identified speech coding parameter is replaced by the median value. And substituting with the speech coding parameters. 28. The method according to claim 28, wherein the step of _obtaining an average comprises: an average excitation signal gain g _mean and an average short-term spectral coefficient f _mean.
Calculating (i). 29. The permuting step included in the step includes forming a set of buffered excitation signal gain values during an averaging period, and forming the set of buffered excitation signal gain values. Arranging, and among the excitation signal gain values whose difference from the median of the set of excitation signal gain values exceeds a predetermined threshold, the L excitation signal gain values having the largest difference from the median value Replace with the median. Median replacement step. 30. The length of the averaging period N is odd, and the median of the aligned set of excitation signal gain values is:
30. The method of claim 29, wherein the value is the value of the ((N + 1) / 2) th element of the set. 31. A line spectrum pair (LSP) coefficient f (k) (k =
1,..., M), and the LSP coefficient f of the ith frame of the averaging period
L of the j-th frame in the averaging period of _i (k)
Determining a spectral distance for the SP coefficient f _j (k). 32. The step of determining the spectral distance is performed by the following equation: (Where M is the order of the LPC model, f _i (k) is the k-th LSP of the i-th frame in the averaging period
32. The method according to claim 31, wherein the method is performed according to (parameter). 33. In an averaging period of length N, the LSP coefficients f _i (k) of the i-th frame are calculated for all other frames (j = 1,..., N,
The method of claim 31, further comprising determining a spectral distance ΔS _i for LSP coefficients of j i). 34. The step of determining the spectral distance is performed by the following equation: The sum of the spectral distances ΔR _ij is calculated for all i = 1,
34. The method of claim 33, wherein the method is performed by determining for N. 35. After determining a spectral distance ΔS _i for each LSP vector f _{i in} an averaging period, aligning the spectral distances according to the value _; and determining the averaging period (i = 1, 2,... A vector f _i having the smallest distance ΔS _i in N) is defined as a center vector f _med of the averaging period, and the distance is represented by ΔS _med ; P (0 ≦ P ≦ N−1) LSPs; the method of the further claim 33 the step of replacing the median of vector f _i with the central vector f _med. 36. The method of claim 26, wherein said identifying and replacing steps are performed independently for each of the excitation signal gain value g and the line spectrum pair (LSP) vector f _i . 37. The method of claim 26, wherein said identifying and replacing steps are performed simultaneously for both an excitation signal gain value g and a line spectrum pair (LSP) vector f _i . 38. If it is determined that the speech coding parameter of a frame should be replaced by the median value of the parameter, both the excitation signal gain value g and the LSP vector f _i of the frame are replaced by the median value. 38. The method of claim 37, comprising the step of replacing with a parameter corresponding to each of the frames containing the parameter. 39. The distance ΔT _ij between the i-th frame and the j-th frame in the averaging period is calculated by the following equation: (Where, M is the order of the LPC model, f _i (k) is the k-th LSP parameter of the i-th frame in the averaging period, g _i is the excitation signal gain parameters of the i-th frame) is determined in accordance with 39. The method of claim 38. 40. The distance ΔS of the speech coding parameters of frame i in an averaging period of length N to all other frames j = 1,..., N (j ≠ i)
the _i, all i = 1, ···, for N, the following formula: [number 10] 40. The method of claim 39, further comprising the step of determining according to: 41. After determining the distance ΔS _i for each frame within the averaging period, further arranging the distances in the order of their values; N) taking the frame having the smallest distance ΔS _i as the center frame of the averaging period, and performing the steps of: ΔS _{med representing} the distance, and g _med and f _med representing the speech coding parameters. 40. The method of claim 40. 42. An averaging period (i = 1, 2,...,
The median substitution process for the speech coding parameters of the frames in N) is performed by converting the parameters g _i and f _i of L (0 ≦ L ≦ N−1) frames into the parameters g of the center frame.
_{42. Performed} by substituting with _med and f _med.
The described method. 43. The difference between an individual distance and the median distance,
_42. The method of claim 41, wherein each distance is represented by a quotient ΔS _i / ΔS _med divided by a median distance. 44. The difference between each distance and the median distance
36. The method of claim 35, wherein each distance is represented by a quotient ΔS _i / ΔS _med divided by a median distance. 45. An apparatus for generating comfort noise in a system including a digital mobile terminal performing discontinuous communication with a communication network, wherein a set of voice coding parameters is averaged during a time when voice is interrupted. And a data processing unit for replacing a speech coding parameter of the set that does not well represent the background noise with a speech coding parameter that favorably represents the background noise, wherein the data processing means comprises: An apparatus for calculating an average value of a set of coding parameters and transmitting the obtained set of average values of speech coding parameters to the communication network. 46. The data processing means may include replacing the speech coding parameters of the set by aligning the set of speech coding parameters and reciprocating speech coding parameters between individual frames within an averaging period. Measuring the distance and identifying the speech coding parameter whose distance from other parameters is the largest within the averaging period, if the maximum distance exceeds a predetermined threshold, 46. The apparatus of claim 45, wherein the speech coding parameter identified as having a maximum distance is replaced with a speech coding parameter whose distance from other speech coding parameters within the averaged distance is measured to be a minimum. . 47. The data processing means may include replacing the speech coding parameters of the set with each other by aligning the set of speech coding parameters, and reciprocating the speech coding parameters between the individual frames within the averaging period. Measuring the distance and identifying the speech coding parameter whose distance from other parameters is the largest within the averaging period, if the maximum distance exceeds a predetermined threshold, 46. The apparatus of claim 45, wherein the speech coding parameters identified as having a maximum distance are replaced with speech coding parameters having a median. 48. The data processing means comprises: an excitation signal gain value g of a speech coding parameter and a line spectrum pair (L
SP) according to claim 45, wherein the replacement identification independently to the vector f _i. 49. The data processing means, comprising: an excitation signal gain value g of a speech coding parameter and a line spectrum pair (L
SP) vector f _i according to claim 45, wherein the replacement both simultaneously identified for. 50. A method for generating comfort noise (CN) in a digital mobile terminal using a discontinuous transmission scheme, the method comprising: transmitting CN parameters to a receiver when speech is interrupted; Shaping the spectral content of the excitation signal into the following steps: forming an excitation signal from the white noise excitation signal sequence; scaling the white noise excitation signal sequence to generate a scaled noise signal sequence; Whether the generated noise signal sequence has a desired comfort noise characteristic or a frequency response characteristic similar to that of a random excitation spectrum control (RESC) filter having transmitted coefficients. Processed by a synthesis filter with fixed coefficients optimized to be at least one of The method carried out by the steps comprising a step.