JPH09149104A - Method for generating pseudo background noise - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide the method for generating background noise by which a pseudo background noise without a sense of incongruity for a recipient is generated independently of background noise information sent intermittently. SOLUTION: In the method for generating background noise for a mobile communication equipment in a non-speech state, a parameter smoothing section 4 and a parameter storage section 5 in cooperation with each other apply smoothing processing to a reflection coefficient of each of prescribed numbers of consecutive frames before coding in one and the same silence period. Then the reflection coefficient subjected to smoothing processing is replaced with the reflection coefficient before the smoothing processing and the replaced reflection coefficient is coded in a mobile station.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method of generating pseudo background noise to be inserted in a silent section during a call used in a mobile communication device, and more specifically, to reduce power consumption of a mobile station. The present invention relates to a method of generating pseudo background noise when controlling a transmission output of voice information in an off state in a silent section during a call.

[0002]

2. Description of the Related Art In order to extend the life of a battery in a mobile station (portable terminal) of a mobile communication device, a so-called VOX (Voice O
peratede Transmitter) is known as a method of transmitting audio signals called control.

In VOX control, when a mobile station detects silence during a call, a postamble signal (hereinafter referred to as P
By sending an OST signal), the audio signal output of the next frame is turned off. In addition, when a voice is detected in the silent state, a preamble signal (hereinafter, also referred to as a PRE signal) is transmitted to notify that the transmission output of the audio signal is turned on.

On the other hand, in order to enable the generation of pseudo background noise on the base station side during a silent section, the POST signal contains information on the mobile station background noise. The information on the mobile station background noise included in the POST signal is a signal obtained by encoding the mobile station background noise according to the normal voice coding standard (standard rule of the Radio System Development Center of the Foundation). The mobile station periodically performs POS during the silent period (maximum period is once per second).
By transmitting the T signal, it is possible to update the pseudo background noise generated on the base station side.

During the silent period, the mobile station has the above-mentioned maximum period,
The background noise information is transmitted only once per second. Therefore, if the voice information is transmitted 50 times per second during the sound period, the base station can obtain the background noise information only once per second during the silent period, and this amount of information is
It corresponds to only 2% of the voice information during the voiced period. However, it is impossible to reproduce the actual background noise with this 2% information. So at the base station
There is no choice but to insert pseudo background noise by using the information for%.

From the viewpoint that the pseudo background noise to be inserted is easy to hear, the following (a) to
It is desired to satisfy the point (c). a) Pseudo-background noise should not be offensive to the recipient. b) Smooth updating at the time of entering a silent period from a sound period and at the time of updating pseudo background noise of each cycle in the same silent period. c) There is not much difference in the sound quality of the pseudo background noise generated in each cycle during the same silent period.

In mobile communication, a digital voice signal obtained by converting an analog voice signal into a PCM is converted into a VSELP code (Vector Sum Excitation (residual) Linear Prediction Code (Vector-Sum E) based on the standard regulations of the aforementioned Radio System Development Center.
xcited Linear Predictive Coding)) and is transmitted. In order to perform this VSELP encoding, parameters of frame energy and reflection coefficient are calculated based on the digital audio signal, the calculated parameters are quantized, and the codewords corresponding to the quantized values are arranged in the order based on the above rule. Then, a transmission code is formed and transmitted. In this case,
As for the reflection coefficient, the reflection coefficient parameter is -1.
The value of the reflection coefficient is quantized by referring to the quantization table defined by the above-mentioned regulation. The calculated frame energy parameter is also quantized for the frame energy. In this specification, the quantized value of the parameter of the reflection coefficient is referred to as the reflection coefficient value before encoding (or the reflection coefficient value before encoding), and the code word corresponding to the quantized value of the reflection coefficient is encoded. The reflection coefficient value (encoded reflection coefficient value) is referred to, and the codeword corresponding to the value of the quantized frame energy parameter is referred to as an encoded frame energy value (or encoded frame energy value). The background noise information including the coded frame energy value and the coded reflection coefficient value will be described herein as included in the postamble signal.

Regarding the insertion of background noise in a silent section, for example, a method described in Japanese Patent Laid-Open No. 5-122165 is known. The configurations of the transmitting side and the receiving side of the mobile station to which this method is applied are as shown in FIGS. 9 (a) and 9 (b). On the transmission side shown in FIG. 9 (a),
When the voice changes from voiced to silence, the postamble generator 45 sends the postamble signal to the data switching unit 4.
7 to the transmitting unit 47. The encoded frame power (frame energy value) of the voice output from the high-efficiency voice encoder 42 following the postamble signal sent from the data switching unit 47 to the transmission unit, the prediction coefficient, etc. (the reflection coefficient is predicted The code processed to send the coefficient to the subsequent process) is sent.

On the receiving side shown in FIG. 9B, the background including the coding frame power and the prediction coefficient of the background noise speech transmitted from the high-efficiency speech encoder 42 on the transmitting side in the silent section. The noise information is captured and stored in the storage unit 63. After this background noise information is read from the storage unit 63, it is input to the background noise synthesis unit 65 together with the residual signal generated from the random residual generation unit 64, and is synthesized there to generate background noise.

[0010]

However, in the above-described conventional mobile communication device, the method of generating the pseudo background noise is such that the background noise information from the high-efficiency speech coder is random as it is on the transmitting side and the receiving side. To combine with the residuals,
It is unknown whether or not it produces noise that is easy to hear.

The reason is that the frame energy value of the voice determines the sound pressure level of the voice, and the reflection coefficient value determines the height (frequency) of the voice. Based on these, the sound pressure level and the height of the pseudo background noise are determined. Can be decided. However, since background noise information is transmitted only once per second at maximum during a silent period, if there is a large variation between the frame energies and the prediction coefficients at each reception, the magnitude and frequency of the pseudo background noise to be generated will vary. The variation is also large. If the noise with a large variation changes at maximum every 1 second, it is considered unnatural and difficult to hear.

Further, generating a random residual each time background noise is generated increases the processing on the base station side, and is not very efficient.

It is an object of the present invention to provide a pseudo background noise generation method capable of generating pseudo background noise that does not cause a discomfort to the receiver regardless of the background noise information transmitted intermittently.

[0014]

The pseudo background noise generation method of the present invention is a pseudo background noise generation method for a mobile communication device during a silent period, in which a predetermined number of consecutive pre-encodings in the same silent section are encoded. The mobile station performs smoothing on the reflection coefficient value of each frame, replaces the smoothed reflection coefficient value with the reflection coefficient value before smoothing, and encodes the replaced reflection coefficient value. Is characterized by.

According to the pseudo background noise generation method of the present invention,
The reflection coefficient value of each of a predetermined number of consecutive frames before encoding in the same silent section is smoothed, and the smoothed reflection coefficient value is replaced with the reflection coefficient value before smoothing, and replaced. The reflected reflection coefficient value is encoded and transmitted. As a result, the pseudo background noise generated based on the coded reflection coefficient value does not change suddenly and does not give a feeling of strangeness to the receiver regardless of the time interval of the transmitted reflection coefficient value.

According to the pseudo background noise generating method of the present invention, the reflection coefficient value before encoding which is smoothed is the reflection coefficient value of each of a predetermined number of frames which continue from the end of the hangover interval to the end of the hangover interval. Is characterized in that.

According to the pseudo background noise generation method of the present invention,
In the hangover section, since the pre-encoding reflection coefficient values of a predetermined number of frames that continue back in the hangover section from the end of the hangover section are smoothed, reception is performed even at the initial generation of pseudo background noise. It is possible to obtain a pseudo background noise that does not cause discomfort to the person.

The pseudo background noise generation method of the present invention is the pseudo background noise generation method during a silent period in a mobile communication device, which is an average value of coding energy values of a predetermined number of consecutive frames in the same silent section. Is calculated in the mobile station, the coding energy value of the average value subjected to the calculation processing is corrected by the coding energy value transmitted immediately before, and the corrected coding energy value is transmitted from the mobile station. It is characterized in that it is a coded energy value.

According to the pseudo background noise generation method of the present invention,
An average value of the coding energy values of a continuous predetermined number of frames in the same silent section is calculated in the mobile station, and the coding energy value of the calculated average value is transmitted immediately before. And the corrected encoded energy value is transmitted from the mobile station. Therefore, the pseudo background noise generated based on this coded energy value does not change its size extremely, and the size of the pseudo background noise is gently changed, so that the pseudo background noise which is easy for the receiver to hear is obtained. To be

The pseudo background noise generating method of the present invention is a pseudo background noise generating method in a mobile communication device during a non-communication period, in which a received coding energy value, a coded reflection coefficient value, and a residual error of white noise which is provided in advance. It is characterized in that a pseudo background noise signal is generated on the base station side based on the code.

According to the pseudo background noise generation method of the present invention,
A pseudo background noise signal is generated on the base station side based on the received coded energy value, the coded reflection coefficient value, and the white noise residual code provided in advance. In this generation, since it is generated based on the white noise residual code provided in advance, the amount of processing on the base station side can be reduced.

In the pseudo background noise generating method of the present invention, the residual code of white noise is stored in advance as a file, and the stored residual code of white noise is sequentially and repeatedly read during a silent period to obtain a pseudo background noise signal. It is characterized by generating.

According to the pseudo background noise generation method of the present invention,
The white noise residual code is stored in advance as a file, and a pseudo background noise signal can be generated by repeatedly reading the stored white noise residual code sequentially during a silent period. The amount of processing on the base station side will be small.

[0024]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below. 1 and 2 are a schematic configuration diagram on the mobile station side and a schematic configuration diagram on the base station side of a mobile communication device to which the pseudo background noise signal generation method of the present invention is applied.
The parts related to the method of the present invention are mainly shown in both schematic configuration diagrams.

The mobile station side shown in FIG. 1 will be described. The digital audio signal converted into digital data is supplied to the encoding unit 1, and the digital audio signal is encoded by the encoding unit 2. Further, the digital audio signal is supplied to the sound / silence detection unit 3, and when the sound signal / silence detection unit 3 detects that the sound signal is in the sound state, that is, the sound / silence detection. If the output of the unit 3 is “voiced → voiced”, the output voice code from the encoding unit 2 is transmitted via the transmission line encoder 6.

When the sound / silence detection unit 3 detects that the voice signal has changed from the sound state to the silence state, that is, when the output of the sound / silence detection unit 3 is “sound → silence”, a hangover occurs. The section is started, and the voice code is transmitted via the transmission line encoder 6. When the voice / silence detection unit 3 detects that the voice signal is kept silent, that is, when the output of the voice / silence detection unit 3 is “silence → silence”, there is a period between hangovers and a periodic period. To POST
Stop voice code transmission except when transmitting signals.

At this time, a short burst signal of the minimum information required to establish synchronization is transmitted. In the (SN) frames (S ≧ N) after the start of the hangover period (length S frame), the voice code is sent to the transmission line encoder 6 as it is, but in the remaining N frames, the reflection code before the voice signal is encoded. The numerical value is smoothed in cooperation with the parameter smoothing unit 4 and the parameter storage unit 5, and the smoothed reflection coefficient value is replaced with the reflection coefficient value before smoothing. The replaced reflection coefficient value is encoded and then sent to the transmission line encoder 6 together with other codes.

On the other hand, when the POST signal is transmitted periodically after the end of the hangover period, the frame energy value and the reflection coefficient value of the voice signal are smoothed and averaged by the cooperation of the parameter smoothing unit 4 and the parameter storage unit 5. After processing, it is added to the POST signal together with other codes and sent to the transmission line encoder 6.

Further, the frame energy and the reflection coefficient value output from the parameter smoothing unit 4 are stored in the parameter storage unit 5 for the above-described smoothing and averaging processing of the frame energy value and the reflection coefficient value for the next transmission. send. If the output of the sound / silence detection unit 3 is “silence → sound”, a PRE signal is sent to the transmission path encoder 6 to give advance notice that voice signal transmission is started.

The operation of the multiplexer 61 will be described below. When "voiced → voiced" is detected based on the audio signal, the multiplexer 61 detects the sign by the detected signal "voiced → voiced". The speech code encoded by the conversion unit 2 is channel-encoded by the channel encoder 6 and transmitted. When it is detected that "soundless → silent" based on the voice signal, the multiplexer 61 is switched by the detection signal "soundless → silent", and the smoothed coded reflection coefficient value and the averaged code are used. The converted frame energy value and other codes are channel-coded in the channel encoder 6 and transmitted. When ‘silence → sound 〃’ is detected based on the voice signal, the multiplexer 61 is switched by the detection signal of ‘silence → sound 〃’ and the voice code encoded by the encoder 2 is transmitted through the transmission line. The encoder 6 encodes the transmission path and sends it.

More detailed description will be given later based on the operation flow of the mobile station transmitting side in the silent section shown in FIGS. 3 and 4.

Next, the base station side shown in FIG. 2 will be described. The received signal is supplied to the separation circuit 9, and the separation circuit 9 extracts a voice code and voiced / unvoiced information. Separation circuit 9
If there is an output of only the audio signal from, the switching unit 14 outputs the audio code. If the separation circuit 9 outputs the POST signal, the voice code is stored in the storage unit 10, and the pseudo background noise is generated in the silent section until the next POST signal or PRE signal comes.

To generate pseudo background noise, the reflection coefficient value calculated by the reflection coefficient value calculation unit 11 is calculated, and at the same time, the residual code file reading unit having a memory in which the white noise residual code is stored in advance as a file. The residual code is read by 12 and sent to the pseudo background noise synthesizer 13 together with other codes. The silent section switching unit 14 outputs the background noise code from the pseudo background noise synthesis unit 13. If the PRE signal is output from the separation circuit 9, the silent section is ended, and the switching unit 14 outputs a voice code.

More detailed description will be given later based on the operation flow on the base station side in the silent section shown in FIG.

Next, the operation on the mobile station side will be described with reference to FIGS.
It will be described based on the flow church shown in.

At the mobile station side, it is checked whether there is sound or no sound (step S1), and the voice signal is encoded and transmitted until it is judged that there is no sound (step S1).
2). If it is determined that there is no sound in step S1, the hangover section is started (step S3),
The encoded frame energy value and the reflection coefficient value before encoding are stored (step S4). Normally, when a silent state is detected from a voiced state, the voice code transmission is not immediately cut off, but a few-frame silence signal is directly encoded as a voice signal so that the ending is not cut off in the voiced state. Sent. This section is called a hangover section. The length of the hangover section is S frames.
Further, it is assumed that the POST signal is transmitted every 50 frames during the silent section.

After step S4, it is checked whether or not (SN) frames have passed (step S5). If it is determined that the frame has not passed, the silent signal is directly encoded as a voice signal and transmitted. (Step S
6). N is a natural number and N ≦ S. If it is determined in step S5 that (SN) frames have elapsed, the pre-encoding reflection coefficient value stored in step S4 is read and smoothed (step S7).
The smoothed reflection coefficient value before encoding is replaced with the reflection coefficient value before smoothing and before encoding, and the replaced reflection coefficient value is encoded and then transmitted together with other codes ( Step S8). After step S8, the process is repeated from step S4 until S frames have elapsed (step S
9).

When it is determined in step S9 that the S frame has elapsed, it is then checked whether or not silence has continued even after the hangover interval has elapsed (step S10), and it is determined that silence has not continued. If so, the voice signal is encoded and transmitted following step S10 (step S11). When it is determined in step S10 that silence is continuing, smoothing processing of the coded frame energy value and smoothing processing of the reflection coefficient value before coding are performed (step S12), and the smoothed code The smoothed frame energy value is replaced with the coded frame energy value before smoothing, the smoothed reflection coefficient value before coding is replaced with the reflection coefficient value before smoothing, and the smoothed and replaced code The reflection coefficient value before conversion is encoded (step S13).

The voice signal encoded in step S13 is transmitted together with the first postamble signal (step S14), and it is checked whether or not the next frame is also silence (step S15). The preamble signal is transmitted (step S16), and the process is executed from step S1.

When it is determined that there is no sound in step S15, it is checked whether or not (50-M) frames have passed since the immediately preceding postamble signal was transmitted (step S17). M is a natural number and M ≦ S (S
Is the number of frames in the hangover section). Step S
When it is determined in (17) that (50-M) frames have not elapsed, a short burst signal is transmitted following step S17 for synchronization, and the processing is executed again from step S15 (step S18).

When it is determined in step S17 that (50-M) frames have elapsed, the encoded frame energy value and the reflection coefficient value before encoding are stored (step S19). Following step S19, it is checked whether or not 50 frames have passed after the immediately preceding postamble signal was transmitted (step S20).
If it is determined that 0 frames have not elapsed, then a short burst signal is transmitted following step S20 for synchronization, and the processing is executed again from step S15 (step S21).

When it is determined in step S20 that 50 frames have passed, the coding frame energy value smoothing process and the pre-coding reflection coefficient value smoothing process are performed (step S22). The coded frame energy value is replaced with the coded frame energy value before smoothing, and the smoothed reflection coefficient value before coding is replaced with the reflection coefficient value before smoothing, smoothed and replaced The reflection coefficient value before encoding is encoded (step S
23), the signal coded in step S23 is transmitted together with the other code together with the next postamble signal and subsequently executed again from step S15 (step S24).

Next, the smoothing and averaging processing will be described in detail. First, prior to the description of the smoothing and averaging processing, a VSELP code used in mobile communication for voice encoding and decoding will be described.

One VSELP code has audio information of one audio frame. The length of one audio frame is 160 samples (20 msec), and there are 50 frames per second.
Furthermore, one audio frame has four subframes (5 msec).
Divided into Table 1 below is a parameter code list of VSELP codes, where R is the frame energy and soft
in is a soft interpolation bit, ri is a reflection coefficient, Lj is a lag of the jth subframe, Ij is a residual of the jth subframe, and gj is [GS, PO] of the jth subframe. .

[0045]

[Table 1]

Hereinafter, k is the background noise update cycle number in the same-silent interval, and t is the number of frames of background noise during each cycle. Equation (1) is the code of the background noise that has been smoothed by the mobile station transmitting the k-th period during the same-silent interval,
It is shown by equation (3). Expression (2) is the sign of the background noise of the t-th frame generated after receiving the k-th cycle in the base station during the same-silent section, and is expressed by Expression (4).

[0047]

(Equation 1)

[0048]

(Equation 2)

[0049]

(Equation 3)

[0050]

(Equation 4)

Here, i = 1 to 10 and j = 0 to 3. k and t differ depending on the length of the silent section and the length of the transmission cycle of the mobile station. When the silent period is 5 seconds and the transmission cycle of the mobile station is 1 second, k = 1 to 5 and t = 1 to 50. FIG. 6 shows an operation sequence of an example of the transmission side of the mobile station. In the following, the transmission of the k-th cycle will also be referred to as the k-th post signal transmission.

The example shown in FIG. 6 has a smooth section of the previous N frames where the voiced section ends and the hangover section of the S frame has elapsed, and a silent section continues after the hangover section and continues to the hangover section. 1st PO on the frame
The case where the ST signal is transmitted, the second POST signal is transmitted at the 50th frame after the first POST signal is transmitted, and the same period is entered after the kth POST signal is transmitted is illustrated below. .

Next, the smoothing of the reflection coefficient value before encoding and the correction of the encoded reflection coefficient value in the hangover section on the mobile station side will be described. The smoothing of the reflection coefficient value before coding and the correction of the coded reflection coefficient value in the hangover section on the mobile station side are smoothing and correction in the section a of FIG. In order to smoothly continue the frequency distribution of the voice in the hangover section and the frequency distribution of the pseudo background noise in the silent section, the reflection coefficient value before encoding is smoothed in the hangover section.

After the start of the hangover period (SN), the silence signal of the frame is transmitted as a voice signal, and the remaining N
For (N ≦ S) frames, the reflection coefficient value before encoding is smoothed as shown in the following equation (5), and the smoothed reflection coefficient value before encoding is smoothed. It is replaced with the reflection coefficient value before being processed.

[0055]

(Equation 5)

In the equation (5), n = (N-1) to 0, i
= 1 to 10. ri (S-n) is the i-th reflection coefficient value before the coding of the (S-n) th frame, and ri (S-n-
1) is the i-th reflection coefficient value before encoding of the frame one frame before (S-n).

In the smoothing by the above-mentioned equation (5), the reflection coefficient value of two consecutive frames is smoothed, and the smoothed reflection coefficient value is replaced with the reflection coefficient value before smoothing. These processes are processes of reflection coefficient values before encoding, and the smoothed and replaced reflection coefficient values are encoded (N-
1) Frames are transmitted. In this case, smoothing may be performed on two or more consecutive reflection coefficient values before encoding.

Next, for the last frame ri (S) of the hangover section, the reflection coefficient value before encoding is similarly smoothed, and the smoothed reflection coefficient value before encoding is before smoothing. Substituted for the reflection coefficient, the smoothed and substituted reflection coefficient value is encoded. Only the first reflection coefficient value r1 (S) in the coded reflection coefficient values is subjected to the processing represented by the following expression (6).

[0059]

(Equation 6)

That is, the first of the encoded reflection coefficient values
Only for the reflection coefficient value r1 (S), rl (S) = rl (S)
The correction of ± 1 or ± 2 is performed, the encoded reflection coefficient values other than the encoded first reflection coefficient value are not corrected, and the corrected first reflection coefficient value and the uncorrected second reflection coefficient value are not corrected. ~ The tenth reflection coefficient value is transmitted.

Ri (S-n) in the equation (5) is a reflection coefficient value before encoding and is a decimal value (-1 <ri (S-n) <1).
However, rl (S) in the equation (6) is the encoded first reflection coefficient value and is (0 ≦ rl (S) ≦ 31). Here, since the first reflection coefficient value is coded into 5 bits, it is limited to 0 ≦ rl (S) ≦ 31. Furthermore, the reason for the processing of equation (6) will be briefly described. The reflection coefficient value has frequency information of voice. In particular, the first reflection coefficient value has the largest amount of information. The smaller the codeword value, the lower the voice frequency, and the larger the codeword value, the higher the voice frequency. Since it is difficult to hear if the frequency is too low or too high, it is difficult to hear. Therefore, the signal when transmitting the POST signal for each time in the silent period described below and the encoded first reflection of the last frame of the hangover period described here. The coefficient value is corrected as shown in equation (6).

Next, the smoothing of the coded frame energy value of the mobile station in the silent section will be described.

Smoothing of the coded frame energy value of the mobile station in the silent section is performed by, for example, section d in FIG.
3 is an averaging of the coded frame energy values in the section illustrated in FIG. During the silent section, the mobile station kth PO
Let mk be the audio frame number when transmitting the ST signal. The coded frame energy value to be transmitted together with the k-th POST signal is displayed by equation (7). In order to calculate the coded frame energy value displayed by the equation (7), first, m
Average value {R of the encoded frame energy values from the kth frame to the Mth frame (mk-M-1)
(K) AVR} is calculated by the equation (8).

[0064]

(Equation 7)

[0065]

(Equation 8)

When the kth POST signal is transmitted, the average value {R (k) AVR} of the coded frame energy values is further converted into the {average value R (1 ) AVR} and take the average, and the average value is {R '
(K) AVR}. The value in this case is calculated by the equation (9). The average value of the coded frame energy values when transmitting the first POST signal is {R '(1) AVR =
(2 · R (1) AVR / 2) = R (1) AVR}.

[0067]

(Equation 9)

Average value of encoded frame energy values {R '
If the value after subtracting p (k) shown in Table 2 from (k) AVR} is displayed by the expression (10), the value is obtained by the calculation of the expression (11). If the value obtained by the equation (11) is less than 0, the value of the equation (11) is set to 0.

[0069]

[Table 2]

[0070]

(Equation 10)

[0071]

[Equation 11]

Subsequently, the previously transmitted coded frame energy value shown in equation (12) is used to correct the coded frame energy value to be transmitted next shown in equation (13).

[0073]

(Equation 12)

[0074]

(Equation 13)

In order to correct the encoded frame energy value to be transmitted, first, the difference between the value of the equation (12) and the value of the equation (13) is obtained by the equation (14).

[0076]

[Equation 14]

Difference ΔR calculated by the equation (14)
When the absolute value of (k) is larger than 1, the energy value to be transmitted, which is shown in the equation (7), is obtained by the equation (15).

[0078]

(Equation 15)

Difference ΔR obtained by the calculation of equation (14)
When the absolute value of (k) does not exceed 1, the transmission energy value shown in the above equation (7) is obtained by the equation (16).

[0080]

(Equation 16)

The first POST is calculated by the equation (14).
When obtaining the coded frame energy value to be transmitted with the signal, that is, when calculating ΔR (1) when (k = 1), the code at that time is set as the initial value of the coded frame energy value shown in equation (17). Frame energy value {=
The average value R (k) AVR} is used. That is, the value of the equation (18) is used.

[0082]

[Equation 17]

[0083]

(Equation 18)

The average value of the coded frame energy values calculated as described above is replaced with the coded frame energy value in the coded voice signal data, and the coded voice signal data thus replaced is transmitted.

The smoothing processing of the above frame energy value is all processing of the coded frame energy value. As is clear from the above equations (9), (11), (14), (15), and (16), in this example, the average value of the encoded coded frame energy values is calculated. The calculated encoded frame energy value of the average value is corrected by the encoded frame energy value at the time of the immediately preceding transmission, and the energy value transmitted each time during the same-silent section has less variation. , It will be gradually attenuated, making it easier to hear the pseudo background noise. In a normal state, the average value of the coded frame energy values calculated and transmitted at the beginning of the silent section is not exceeded,
And the average value of the coded frame energy values transmitted immediately before is not exceeded. By the above processing, the energy value transmitted each time during the same-silent section has a small variation and is gradually attenuated, so that the pseudo background noise becomes easy to hear.

Next, the smoothing of the reflection coefficient value of the mobile station in the silent section after the end of the hangover section will be described.

In this case, for example, the smoothing process of the reflection coefficient value in the sections b and c in FIG. 7 is performed. In the same manner as the processing of the coded frame energy value of the mobile station in the silent section, Mobile station kth POST
The audio frame number when transmitting a signal is mk.

In order to obtain the pre-encoding reflection coefficient value displayed by the following equation (19) when transmitting the kth POST signal, first, the m frames are traced back to the M frames (m
The average value of the reflection coefficient values ri (i = 1 to 10) before encoding, which is represented by the expression (20) up to k−M−1) frames, is calculated as (21).
Determined by the formula. Here, M is a natural number of M ≦ S.

[0089]

[Equation 19]

[0090]

(Equation 20)

[0091]

(Equation 21)

Average value {ri of reflection coefficient values before encoding
(K) AVR} and the reflection coefficient value before encoding shown in the equation (22) transmitted last time, and the k-th P
The transmission value shown in the equation (19) when transmitting together with the OST signal is obtained.

[0093]

(Equation 22)

[0094]

(Equation 23)

The initial value of the equation (22) used when calculating the pre-encoding reflection coefficient value shown in the equation (24) which is transmitted with the first POST signal by the equation (23) is the last code of the hangover interval. Reflection coefficient value ri (m1-
Use 1). That is, it is as in the equation (25).

[0096]

(Equation 24)

[0097]

(Equation 25)

A method of determining the coefficients a and b in the equation (23) (a and b are predetermined positive integers) will be described. If the coefficient b is set to be larger than the coefficient a to some extent, the previous transmission value is emphasized, so that the value of the equation (23), that is, the dispersion of the reflection coefficient value before encoding becomes small. The reflection coefficient value before encoding smoothed by the above processing is replaced with the reflection coefficient value before encoding.

Next, only the value of the coded first reflection coefficient value represented by the equation (27) in the coded reflection coefficient values displayed by the equation (26) is converted into the following equation (28). Correct as shown.

[0100]

(Equation 26)

[0101]

[Equation 27]

[0102]

[Equation 28]

The reflection coefficient value (ri (mk-
h)) and the reflection coefficient value in the equation (23) (reflection coefficient value shown in the following equation (29)) is a reflection coefficient value before encoding and is a decimal value, but in the equation (28), Expression (30) is an encoded reflection coefficient value, which is a value within the range shown in expression (31).
This range is because the first reflection coefficient value is encoded into 5 bits.

[0104]

(Equation 29)

[0105]

[Equation 30]

[0106]

(Equation 31)

Here, the processing of the reflection coefficient value in the silent section after the hangover period has passed will be summarized. As for the reflection coefficient value transmitted together with the k-th POST signal, first, the reflection coefficient value before encoding of each frame M frames before the k-th POST signal is averaged. Similar to the averaged reflection coefficient value before encoding,
−1) From the POST signal, the reflection coefficient values before coding of each frame M frames before are averaged. The averaged reflection coefficient values before both encodings are weighted and averaged. The weighted average reflection coefficient value before encoding (for example, the k-th PO
(For the reflection coefficient value transmitted with the ST signal)
Reflection coefficient value before being encoded (eg, k-th POS
(For the reflection coefficient value transmitted with the T signal) and the replaced reflection coefficient value is encoded.

The coded first reflection coefficient value in the coded reflection coefficient value is replaced with the coded reflection coefficient value before being corrected by the correction by the equation (28). No correction process is performed on the encoded second to tenth reflection coefficient values. This is because the first reflection coefficient value occupies the largest weight among the other reflection coefficient values.

The first POST signal is transmitted in the first frame, that is, (S + 1) frame, following the end of the hangover period. In this case, M frames are averaged including the reflection coefficient values of the frames in the overhang period before being encoded. When smoothing (2
As shown in the equation (5), the last reflection coefficient value of the hangover section is used as the initial value.

Next, generation of pseudo background noise in the base station will be described with reference to the flowchart of FIG.

When the base station receives the signal, it is checked whether or not the voice signal is received (step S3).
0), if it is determined to be a voice signal, step S30
After the audio signal is output, the process is executed again from step S30 (step S31). If it is determined in step S30 that the signal is not a voice signal, the type of signal is determined (step S32). When it is determined that the signal is the PRE signal in step S32, a voice signal is output, and then the process is repeated from step S30 (step S33).

When it is determined in step S32 that the signal is a POST signal, the audio signal is stored (step S34) and the reflection coefficient value is obtained (step S3).
5). When it is determined in step S32 that the signal is a short burst signal, step S34 is skipped and step S35 is executed.

Subsequent to step S35, the white noise residual code previously stored in the memory as a file is read (step S36), the pseudo background noise signal is synthesized (step S37), and the pseudo background noise signal is obtained. Is output, the process is executed again from step S30 (step S3).
8).

Next, generation of pseudo background noise in the base station will be described in detail. After receiving the kth POST signal, the base station receives the next POST signal or PR during the silent section.
During the period until the E signal arrives, the VSELP code (equation (3)) shown by the equation (32) of the background noise input together with the POST signal and the white noise residual code prepared as a file in the memory in advance are displayed. Using pseudo background noise (3
The VSELP code (= equation (4)) displayed by the equation 3) is combined.

[0115]

(Equation 32)

[0116]

[Equation 33]

In order to give the pseudo background noise the characteristic of the actual background noise, the energy value displayed by the formula (36) of the received background noise is used as the frame energy value displayed by the formula (34). Further, the reflection coefficient value displayed by the equation (35) is the reflection coefficient value received by the equation (37) and the reflection coefficient value of the last frame calculated after the last reception by the base station (48) Calculated using and. In order to give the pseudo background noise randomness, the residual code displayed by equation (38) uses the residual code of the VSELP code displayed by equation (39) generated by white noise. (4
The gain displayed by the equation (0) is a constant, and the lag displayed by the equation (41) is zero.

[0118]

(Equation 34)

[0119]

(Equation 35)

[0120]

[Equation 36]

[0121]

(37)

[0122]

(38)

[0123]

[Equation 39]

[0124]

(Equation 40)

[0125]

[Equation 41]

If the above relationship is shown, equations (42) and (4
Expressions 4) to (47) are obtained. The value of sofin is set to "1" as shown in equation (43).

[0127]

(Equation 42)

[0128]

[Equation 43]

[0129]

[Equation 44]

[0130]

[Equation 45]

[0131]

[Equation 46]

[0132]

[Equation 47]

(T / k) in the above equations indicates the t-th pseudo background noise frame generated after the k-th reception by the base station. The expression (48) in the expression (44) is the reflection coefficient value of the last frame calculated after the last reception by the base station.
The reason (50-t) is displayed in the equation (44) is that the base station needs to generate pseudo noise of 50 frames when the transmission cycle of the mobile station is once per second. After the base station receives the first POST signal,
The initial value displayed by the expression (50) used when calculating the expression 9) is the reflection coefficient value of the last frame of the hangover section. That is, it is as shown in Expression (51).

[0134]

[Equation 48]

[0135]

[Equation 49]

[0136]

[Equation 50]

[0137]

(Equation 51)

U in the equation (47) is defined in the standard of the Radio System Development Center of the foundation [GS, P
O] This is the index number of the codebook. The method of determining U will be described below. The gain γ for determining the loudness of the weighted synthesizer specified in the standard of the Radio Wave System Development Center is calculated by the equation (52). GS and PO in the equation (52) are [GS, PO].
It is the component of the vector selected from the codebook.

[0139]

[Equation 52]

[GS, PO] If Tu shown in equation (53) is calculated for index numbers (u = 0 to 127) of the codebook, each index number u affects the magnitude of the gain γ. I understand. When the gain γ is maximized with respect to pseudo background noise, u (= 11) that maximizes Tu.
You can choose 3). When the gain γ is reduced, Tu is reduced. For example, by selecting u (= 75), the volume of the pseudo background noise sound can be set to a desired volume.

[0141]

(Equation 53)

The residual code of the VSELP code of white noise (Equation (54) below) is prepared in advance for 5 seconds (250 frames) in the residual code file of white noise.
When the pseudo background noise is generated, the white noise residual code file is sequentially read from the beginning, and when it is completed, the white noise residual code file is read from the beginning again.

[0143]

(Equation 54)

[0144]

As described above, according to the method of the present invention, the actual background noise is utilized to average and smooth the discontinuous voice code parameters, thereby reproducing the voice without any discomfort. There is an effect that can be. Further, according to the method of the present invention, by performing the smoothing process in the hangover section, it is possible to avoid the disappearance of the ending and to change the voice to the pseudo background noise in a natural state when switching to the pseudo background noise. . Furthermore, by performing the averaging process of the frame energy value and the reflection coefficient value of the mobile station, the smoothing process, and the smoothing process of the reflection coefficient value of the base station, the loudness and the high level of the pseudo background noise in all the silent sections are measured. Is less likely to be affected by the variation of the background noise signal at the time of transmission in each cycle, and it is possible to obtain an effect that it is possible to reproduce sound without discomfort.

[Brief description of the drawings]

FIG. 1 is a block diagram of an embodiment showing a schematic configuration of a transmission circuit of a mobile station to which the method of the present invention is applied.

FIG. 2 is a block diagram of an embodiment showing a schematic configuration of a receiving circuit of a base station to which the method of the present invention is applied.

FIG. 3 is a flowchart for explaining the operation of the present invention.

FIG. 4 is a flowchart for explaining the operation of the present invention.

FIG. 5 is a flowchart for explaining the operation of the present invention.

FIG. 6 is a schematic explanatory diagram of a voiced section and a silent section used for explaining the operation of the present invention.

FIG. 7 is a schematic explanatory diagram for explaining smoothing of a reflection coefficient value in the present invention.

FIG. 8 is a schematic explanatory diagram for explaining averaging of coded frame energy values in the present invention.

FIG. 9 is a block diagram showing a configuration of a transmission circuit and a reception circuit on a mobile station side of a conventional example.

[Explanation of symbols]

 2 Encoding section 3 Voice / silence detection section 4 Parameter smoothing section 5 Parameter storage section 6 Transmission line encoder 9 Separation circuit 10 Parameter storage section 11 Reflection coefficient value calculation section 12 Residual signal file reading section 13 Pseudo noise synthesis section 14 Switching unit 61 Multiplexer

Claims (5)

[Claims] 1. A pseudo background noise generating method for a mobile communication device during a silent call, wherein a predetermined predetermined number of consecutive reflection coefficient values before coding in the same silent section are smoothed and smoothed. A pseudo background noise generation method, comprising: replacing a processed reflection coefficient value with a reflection coefficient value before smoothing processing and encoding the replaced reflection coefficient value in a mobile station. 2. The pseudo background noise generation method according to claim 1, wherein the pre-encoding reflection coefficient value to be smoothed is a predetermined number of frames that are continuous from the end of the hangover interval back in the hangover interval. Pseudo background noise generation method, characterized in that it is the reflection coefficient value of. 3. A pseudo background noise generating method for a mobile communication device during a silent period, wherein the mobile station arithmetically processes the average value of the encoded energy values of a predetermined number of consecutive frames in the same silent section, The coding energy value of the average value subjected to the arithmetic processing is corrected by the coding energy value transmitted immediately before, and the corrected coding energy value is set as the coding energy value for transmission from the mobile station. Characteristic pseudo background noise generation method. 4. A pseudo background noise generating method for a mobile communication device during a non-communication period, based on a received coded energy value, a coded reflection coefficient value and a preset white noise residual code. A pseudo background noise generation method characterized by generating a pseudo background noise signal in. 5. The pseudo background noise generation method according to claim 4, wherein the residual code of the white noise is stored in advance as a file, and the stored residual code of the white noise is read repeatedly in sequence during a silent period and simulated. A method for generating pseudo background noise, which comprises generating a background noise signal.