KR0180824B1 - Adaptive differential pulse code modulation - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

Speech activity detection method using differential energy and output code range measurement, and variable rate adaptive differential pulse code modulation device using the same.

2. The technical problem to be solved by the invention

The 32k adaptive differential pulse coded modulator and the variable rate adaptive differential pulse coded modulator have a problem in that they cannot use the bandwidth of the channel efficiently.

3. Summary of Solution to Invention

It detects the voice activity and transmits the voice data at the low data rate in the part where the voice activity is low and at the high data rate in the part where the voice activity is high, thereby effectively using the channel transmission band and reducing the amount of interference by reducing the transmission power. To provide a voice activity detection method for increasing the capacity of a code division multiple access (CDMA) system and a variable rate adaptive differential pulse code modulation device using the same.

4. Important uses of the invention

Used to increase the capacity of code division multiple access (CDMA) systems.

Description

Speech activity detection method and variable rate adaptive differential pulse code modulation device using the same

1 is a flow chart for a voice activity detection (VAD) method using differential energy according to an embodiment of the present invention.

2 is a schematic diagram of a variable rate adaptive differential pulse code modulation (ADPCM) apparatus according to an embodiment of the present invention to which a voice activity detection (VAD) method using differential energy is applied.

3 is a flowchart of a voice activity detection (VAD) method by output code range measurement according to another embodiment of the present invention.

4 is a schematic diagram of a variable rate adaptive differential pulse code modulation device (ADPCM) according to another embodiment of the present invention to which a voice activity detection (VAD) method based on output code range measurement is applied.

5 is an explanatory diagram of an output code range of a 32k adaptive differential pulse code modulation device (ADPCM) and determination of a data rate accordingly.

6 is an explanatory diagram of an output code range in a frame according to each transmission rate and a corresponding transmission code.

7 is a flowchart of a voice activity detection (VAD) method using differential energy and output code range measurement according to another embodiment of the present invention.

8 is a schematic diagram of a variable rate adaptive differential pulse code modulation (ADPCM) apparatus according to another embodiment of the present invention to which a voice activity detection (VAD) method using differential energy and output code range measurement is applied.

* Explanation of symbols for main parts of the drawings

21: buffer

22: voice activity detection unit using differential energy

23: variable rate adaptive differential pulse code modulator

41: voice divider 42: 32k adaptive differential pulse code modulator

43: frame storage unit and decimal conversion unit

44: comparator 45: transmission rate determining unit

The present invention relates to a voice activity detection (VAD) method by measuring differential energy and output code range, and an adaptive differential pulse code modulation (ADPCM) using the same.

The 32k Adaptive Differential Pulse Code Modulation (ADPCM), recommended by the International Telecommunication Union (ITU-T), delivers data at a fixed data rate, so that data is transmitted at a high data rate of 32 kbps even in areas where voice activity is low. Therefore, there is a disadvantage in that the bandwidth of the channel is not used more efficiently than the method of adaptively transmitting the transmission rate.

On the other hand, in the voice activity detection (VAD) method of the QCLP (Qualcomm Code Excited Linear Prediction) used in the North American Code Division Multiple Access Mobile Communication System (IS-95), the data rate for one frame is the voice activity. It is decided according to. In other words, the data rate to be processed is determined by comparing the threshold value with the energy in the frame. The energy R _i (0) of each frame is obtained from the first autocorrelation coefficient of the i th frame, and the threshold is obtained from the i th background noise level B _i . R _i (0) is compared with three thresholds T1 (B _i ), T2 (B _i ), T2 (B _i ). If R _i (0) is greater than the three thresholds, data rate 1 is selected, and if R _i (0) is greater than two of the three thresholds, a half rate is selected, and R _i ( If 0) is greater than one threshold, a rate of 1/4 is selected; if R _i (0) is less than any of the three thresholds, a rate of 1/8 is selected.

The thresholds used to determine the data rate are updated every frame before the data rate is determined. First, the background noise B _i is calculated by the following equation.

_{B i = min [R i-} 1 (0), 160000, max (1.00547B i-1, B i-1 +1)]

At initialization, the background noise B ₁ for the first frame is set to 160,000. The three thresholds are calculated as follows.

T ₁ (B _i ) =-(5.544613 × 10 ^-6 ) B ² _i +4.047 152B _i +362

T ₂ (B _i ) =-(1.529733 × 10 ^-5 ) B ² _i + 8.750045B _i +1136

T ₃ (B _i ) =-(3.957050 × 10 ^-5 ) B ² _i +18.89962 B _i +3347

32k Adaptive Differential Pulse Code Modulator (ADPCM) (G.721) and Variable Rate Adaptive Differential Pulse Code Modulator (G.726) recommended by the International Telecommunication Union (ITU-T) Since the voice is transmitted at a fixed rate, there is a problem in that the bandwidth of the channel is not used more efficiently than the method of adaptively transmitting the rate according to the voice activity.

In order to solve the above problems, the present invention senses voice activity and transmits voice data at a low data rate at a portion where voice activity is low and at a high data rate at a portion where voice activity is large, thereby efficiently using the channel transmission band as well. It is an object of the present invention to provide a voice activity detection method for increasing the capacity of a code division multiple access (CDMA) system by reducing the amount of interference by transmitting power and a variable rate adaptive differential pulse code modulation device using the same.

That is, in the present invention, two voice activity detection (VAD) methods (voice activity detection method using differential energy, voice activity detection method by output code range measurement) and the two voice activity detection methods are simultaneously applied to each other. We propose a voice activity detection method that compensates for the shortcomings, and apply each voice activity detection method to 32k Adaptive Differential Pulse Code Modulator (ADPCM) to operate at various data rates (32, 24, 26, 8kbps). A differential pulse code modulation device (ADPCM) is presented.

The apparatus of the present invention for achieving the above object, the voice using the differential energy to variably determine the transmission rate compared to the differential energy by setting the reference energy after calculating the differential energy by dividing the frame into frames by receiving a voice signal from the outside Activity detection means; Buffering means for buffering a voice signal input from the outside while determining a data rate in the voice activity detection means using the differential energy; And receiving information on the data rate to be operated from the voice activity detection means, setting a quantization level and a variable suitable for the data rate, encoding a voice signal input from the buffering means, and outputting a voice code to the outside at a corresponding data rate. Variable rate adaptive differential pulse code modulation means.

Another apparatus of the present invention for achieving the above object comprises: voice dividing means for dividing a voice signal input from the outside into a frame of a predetermined length; 32k adaptive differential pulse code modulation means for receiving the output of the speech dividing means and outputting an encoded speech code by setting a quantization level and a variable; Frame storing means and decimal converting means for converting the output code into a decimal number after storing an output code of said 32k adaptive differential pulse code modulation means in a buffer by a frame length; Comparison means for selecting a maximum value and a minimum value after receiving and comparing an output code converted to a decimal number within a frame from the frame storage means and the decimal conversion means; And a rate determining means for determining frame rate to which the maximum value inputted from the comparing means belongs and then transmitting frame data at a rate within the corresponding range.

Another apparatus of the present invention for achieving the above object comprises: voice dividing means for dividing a voice signal input from the outside into a frame of a predetermined length; A voice activity detection means using differential energy for receiving a voice signal from the voice dividing means, dividing it into frames, calculating differential energy, setting a reference energy, and changing a transmission rate in comparison with the differential energy; Buffering means for buffering a speech signal input from said speech dividing means while determining a data transmission rate in said speech activity detecting means using said differential energy; First comparison means for receiving an output (X) of the voice activity detection means using the differential energy and a predetermined threshold value (Y) and comparing the two values to output a control signal; Switching means for switching the output of said buffering means in accordance with a control signal input from said first comparing means; Delta modulating means for outputting a voice code (1 or 0) to the outside by modulating a delta modulation method when the output of the buffering means is input through the switching means; 32k adaptive differential pulse code modulation means for receiving an output of the buffering means through the switching means and outputting an encoded speech code by setting a quantization level and a variable; Frame storing means and decimal converting means for converting the output code into a decimal number after storing an output code of said 32k adaptive differential pulse code modulation means in a buffer by a frame length; Second comparing means for selecting a maximum value and a minimum value after receiving and comparing output codes converted to decimal numbers in one frame from the frame storage means and the decimal conversion means; And a transmission rate determining means for transmitting the frame data at a transmission rate in a corresponding range after determining a transmission rate to which the maximum value input from the second comparison means belongs.

A method of the present invention for achieving the above object comprises: a first step of calculating a differential energy (dENG) for each divided frame after dividing an audio signal into frames of a predetermined length; A second step of selecting a minimum value among a current differential energy dENG, a background noise maximum constant 100000, and a low-pass filtered reference energy MAX (1.00547B1, B1 + 1) and setting it as a reference energy B; And a third step of comparing the set reference energy value with the differential energy of the current frame to variably determine the data rate according to the ratio of the energy to the background noise of the current frame.

According to another aspect of the present invention, there is provided a method, comprising: a first step of dividing an audio signal into frames having a predetermined length, encoding each frame at a maximum data rate (32k ADPCM), and storing an output code in a buffer; A second step of retrieving an output code of each frame; And a third step of variably determining a transmission rate according to the number of bits of the output code shown in the search result of the second step.

According to another aspect of the present invention, there is provided a method including: a first step of dividing an audio signal into frames having a predetermined length and then calculating differential energy dENG for each divided frame; A second step of selecting a minimum value among a current differential energy dENG, a background noise maximum constant 100000, and a low-pass filtered reference energy MAX (1.00547B1, B1 + 1) and setting it as a reference energy B; A third step of transmitting frame data at an arbitrary transmission rate 4 (8 kbps) when a value obtained by multiplying a predetermined fourth threshold by the reference energy B is greater than the differential energy dENG; And a fourth step of encoding each frame at a maximum data rate (32k ADPCM), storing the output code in a buffer, retrieving the output code of each frame, and variably determining the data rate according to the number of bits of the output code.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a method for detecting a voice activity using differential energy according to an embodiment of the present invention, wherein a voice activity detection (VAD) for a variable rate adaptive differential pulse code modulation (ADPCM) proposed by the present invention is shown. A flowchart of the method.

The method of measuring voice activity of FIG. 1 basically measures the voice activity using differential energy of a voice signal. The reason for using the differential energy is that the adaptive differential pulse code modulation device (ADPCM) is designed to quantize the differential signal. A flowchart of a voice activity detection method using differential energy is sequentially described as follows.

First, the voice signal is divided into frames of a predetermined length (11). The length of the frame is advantageously shortened so that the characteristics of the voice signal do not change significantly, but the length of the frame should be maintained so that the overhead of voice activity information transmission is not too large. In the present invention, a frame length of 20 msec, which is widely adopted by Vocoder, is selected. In general, it is known that the characteristics of a voice signal do not change significantly during 20 msec.

The differential energy dENG is calculated for each divided frame (12). The differential signal is defined as the difference between the current sample and the past sample of the speech signal. The differential signal is squared and summed over the frame period to obtain the differential energy. That is, it is represented by the following formula.

Where dENG is the differential energy, L is the frame length, S (i) is the i-th sample in the frame, and i is the sample index in the frame.

Then, the reference energy (B) is obtained using the differential energy. The reference energy is a reference value when measuring voice activity. The reference energy is the minimum value of the current differential energy (dENG), the maximum background noise constant (100000), and the low-pass filtered reference energy (MAX (1.00547B1, B1 + 1)). Defined, this value is updated every frame (13). The reference energy thus obtained has a meaning of the differential energy of the background noise, which is a reference for measuring the voice activity of each frame.

Among the three factors required to obtain the reference energy, the maximum background noise constant and the regional filtered differential energy are briefly described as follows.

◎ Maximum background noise

If the background noise is too loud or the voice section lasts too long, the reference energy value may be too large to cause problems. Therefore, the upper limit of the reference energy needs to be set. The value is the maximum background noise constant. Was selected.

◎ Low-filtered reference energy (MAX (1.00547 × B1, B1 + 1))

In the case of a quiet environment, the differential energy dENG is very small, so the differential energy among the three elements becomes the minimum value, and the reference energy B becomes the differential energy dENG. However, when the voice interval starts, the differential energy value dENG increases abruptly. At this time, the element preventing the sudden change of the reference energy B is the low-pass filtered reference energy. Here, since B1 is the reference energy of the previous frame, the value of the low-pass filtered reference energy increases little by little but does not change rapidly. Therefore, the reference energy B increases little by little.

In addition, the differential energy (dENG) increases when the environment changes and the surrounding noise is severe. In this case, the low-pass filtered reference energy is selected as the current reference energy, so that the silent section is used as the voice section when the background noise is large. Misrecognition prevents the waste of allocating and encoding a large amount of bits.

The reason for choosing a higher value among 1.00547 × B1 and B1 + 1 when obtaining the low-pass filtered reference energy is to operate as 1.00547 × B1 when B1 is a large value and B1 + 1 when B1 is a small value. to be. This is because when B1 is a large value, if B1 + 1 is operated, 1 is too small compared to B1, and the increase rate is too slow. Likewise, when B1 is a small value, when B1 is small, the increase rate is too small. At the end of the negative interval, the differential energy value dENG decreases, so the reference energy B also decreases.

When the reference energy is obtained for each frame, this value is compared with the differential energy of the current frame, and the data rate is changed according to the ratio of the energy to the background noise of the current frame. The threshold is determined by experiment. Looking at this in detail.

That is, if the product of the fourth threshold and the reference energy (B) is greater than the differential energy (dENG), the operation rate is 4 (rate4: 8 kbps) (14), and the differential energy (dENG) is the fourth threshold and the reference energy (B). Is greater than the product of the fourth threshold and the reference energy (B), and operates at a rate of 3 (rate3: 16 kbps) (15), and the differential energy (dENG) is the third threshold and the reference energy (B). If it is larger than the value multiplied by and is smaller than the second threshold multiplied by the reference energy (B), it operates at a rate of 2 (rate2: 24kbps) (16). It works. Thereafter, the above operation is received for the next voice signal.

2 is a schematic block diagram of an ADPCM of a variable rate adaptive differential pulse code modulation apparatus according to an embodiment of the present invention to which a voice activity detection (VAD) method using differential energy is applied. A voice activity detection unit using differential energy, 23 denotes a variable rate adaptive differential pulse code modulation unit, respectively.

The buffer 21 buffers the voice signal input from the outside while the voice activity detector 22 using the differential energy determines the data rate.

The voice activity detection unit 22 using the differential energy receives a voice signal from the outside, divides it into frames, calculates differential energy, sets a reference energy, and compares the differential energy with the transmission rate.

The variable rate adaptive differential pulse code modulator 23 receives information about the rate of operation from the voice activity detector 22 to set a quantization level and a variable suitable for the rate and is input from the buffer 21. It encodes the voice signal and outputs the voice code to the outside at the corresponding rate.

When operating at 32 kbps, the number of quantization levels is quantized and outputted at 15 levels (4 bits), similar to the 32k Adaptive Differential Pulse Code Modulator (ADPCM). When the variable rate adaptive differential pulse code modulator 23 receives information from the voice activity detector 22 to operate at 24 kbps, the variable rate adaptive differential pulse code modulator 23 quantizes and outputs the input voice signal at a quantization level of 7 levels (3 bits). Similarly, upon receiving a signal to operate at 16 kbps from the voice activity detector 22, the variable rate adaptive differential pulse code modulator 23 quantizes and outputs the voice signal with a three-level (2-bit) quantizer. In addition, when receiving a signal to operate at 8 kbps from the voice activity detection unit 22, the variable rate adaptive differential pulse code modulation unit 23 samples the delta modulation method (DeltModulation) and compares the difference with the previous sample value. Outputs a signal of 1 or 0 using only the polarity corresponding to the positive and negative parts of the code).

3 is a flowchart illustrating a voice activity detection (VAD) method by output code range measurement according to another embodiment of the present invention.

First, the audio signal is divided into frames of an appropriate length (31), each frame is encoded at a maximum data rate (32k ADPCM) (32), the output code is stored in a buffer (33), and then the output code of each frame is retrieved ( 34).

As a result, if there is no output code of 2 bits or more among the output codes, the frame is transmitted at a transmission rate of 4 (8 kbps) (35). If there is no output code of more than 3 bits of the output code, the frame is transmitted at a transmission rate of 3 (16 kbps) (36). If there is no output code of 4 bits or more among the output codes, the frame is transmitted at transmission rate 2 (24 kbps) (37). Otherwise, if there is an output code of 4 bits or more among the output codes, the frame is transmitted at transmission rate 1 (32 kbps). Thereafter, the above operation is repeated for the next voice signal.

4 is a schematic configuration diagram of an ADPCM of a variable bit rate adaptive differential pulse code modulation device according to another embodiment of the present invention to which a voice activity detection (VAD) method by output code range measurement is applied. An installment, 42 denotes a 32k adaptive differential pulse code modulator, 43 denotes a frame storage and a decimal converter, 44 denotes a comparator, and 45 denotes a rate determiner.

FIG. 5 is an explanatory diagram of an output code range of the 32k adaptive differential pulse code modulation system (ADPCM) and its rate determination, and FIG. 6 is an explanatory diagram of an output code range in a frame for each transmission rate and a corresponding transmission code.

By dividing the input speech signal into frames of appropriate length and passing the 32k Adaptive Differential Pulse Code Modulator (ADPCM) to measure the range of encoded output codes, it is possible to operate at variable rates.

Looking at the configuration in detail as follows.

The voice divider 41 divides the voice signal input from the outside into frames of a suitable length.

Since the 32k adaptive differential pulse code modulator 42 uses a 15-level quantizer, it receives the output of the voice divider 41 and outputs a voice signal in units of 4 bits per sample, and has a sampling rate of 8k (sample / Second), so we output 32kbps of sampled data. The most significant bit of the output 4 bits represents a sign (i.e., 0 if + and 1 if-). The remaining 3 bits represent the magnitude of the voice signal. The range of the 4-bit output code output from the 32k adaptive differential pulse code modulator 42 is shown in FIG.

The frame storage unit and the decimal conversion unit 43 store the data output from the 32k adaptive differential pulse code modulation unit 42 in a buffer of a frame length, and then easily compare the output codes of 4 bits in a comparator later. Convert to decimal.

The comparator 44 receives output codes converted into a decimal number in one frame from the frame storage unit and the decimal converter 43, and compares them with each other, and selects the maximum value and the minimum value.

The rate determining unit 45 determines the range of the rate 1, rate 2, rate 3, and rate 4 shown in FIG. 5 after the maximum value input from the comparator 44 is in the range corresponding to the rate. Send a frame to

For example, if the maximum value is 0 (binary 0000) and the minimum value is -1 (binary 1111), only the least significant bit (0 or 1) needs to be transmitted. The decoder stage may sign extension to restore (1 → 1111 (decimal −1), 0 → 0000 (zero)).

In addition, when the maximum value is +1 (binary 0001) and the minimum value is -2 (binary 1110), only the lower two bits need to be transmitted, thereby operating at transmission rate 3 (16kbps). In other words, if the output code in a frame is +1 (0001), 0 (0000), -1 (1111), or -2 (1110), respectively, 01, 00, 11, and 10 are transmitted. Can be restored. On the other hand, the output code in the frame for each transmission rate and the actual transmission bit are summarized in FIG.

7 is a flowchart illustrating a voice activity detection (VAD) method using differential energy and output code range measurement according to another embodiment of the present invention.

The proposed algorithm operates in two steps.

The first step is to detect the silence section. This is the same as the determination process of transmission rate 4 in the transmission rate determination algorithm using differential energy. If it is determined as the transmission rate 4, it operates at 8kbps. Otherwise, it proceeds to the second step.

In the second step, the transmission rate 1 to the transmission rate 3 are determined by the transmission rate algorithm using the output code.

Looking at the specific operation is as follows.

First, the voice signal is divided into frames of a predetermined length (71). The length of the frame is advantageously shortened so that the characteristics of the voice signal do not change significantly, but the length of the frame should be maintained so that the overhead of voice activity information transmission is not too large. In the present invention, a frame length of 20 msec, which is widely adopted by Vocoder, is selected. In general, it is known that the characteristics of a voice signal do not change significantly during 20 msec.

The differential energy dENG is calculated for each divided frame (72). The differential signal is defined as the difference between the current sample and the past sample of the speech signal. The differential signal is squared and summed over the frame period to obtain the differential energy. That is, it is represented by the following formula.

Where dENG is the differential energy, L is the frame length, S (i) is the i-th sample in the frame, and i is the sample index in the frame.

Then, the reference energy (B) is obtained using the differential energy. The reference energy is a reference value when measuring voice activity. The reference energy is the minimum value of the current differential energy (dENG), the maximum background noise constant (100000), and the low-pass filtered reference energy (MAX (1.00547B1, B1 + 1)). This value is defined and updated every frame (73). The reference energy thus obtained has a meaning of the differential energy of the background noise, which is a reference for measuring the voice activity of each frame.

Thereafter, when the value multiplied by the reference energy B of the fourth threshold is greater than the differential energy dENG, the operation rate 4 (rate 4: 8 kbps) is performed (74).

After that, each frame is encoded at the maximum data rate (32k ADPCM), and the output code is stored in the buffer (75), and then the output code of each frame is retrieved (76).

As a result of the search, if there is no output code of 2 bits or more among the output codes, the frame is transmitted at transmission rate 4 (8 kbps) (77), and if there is no output code of 3 bits or more among the output codes, the frame is transmitted at transmission rate 3 (16 kbps) (78). If there is no output code of 4 bits or more among the output codes, the frame is transmitted at transmission rate 2 (24 kbps) (79). Otherwise, if there is an output code of 4 bits or more among the output codes, the frame is transmitted at transmission rate 1 (32 kbps). Thereafter, the above operation is repeated for the next voice signal.

8 is a schematic block diagram of an ADPCM variable rate adaptive differential pulse code modulation device according to another embodiment of the present invention to which a voice activity detection (VAD) method using differential energy and output code range measurement is applied. 81 is a voice splitter, 82 is a buffer, 83 is a voice activity detector using differential energy, 84 is a comparator, 85 is a switch, 86 is a delta modulator, 87 is a 32k adaptive differential pulse coder, 88 is a frame storage unit, The decimal converter 89 denotes a comparator and 90 denotes a data rate determining unit.

Looking at the configuration in detail as follows.

The voice divider 81 divides the voice signal input from the outside into a frame of a suitable length.

The buffer 82 buffers the voice signal input from the voice divider 81 while the voice activity detector 83 using the differential energy determines the data rate.

The voice activity detection unit 83 using differential energy receives a voice signal from the voice divider 81 and divides it into frames to calculate differential energy, and then sets reference energy to compare the differential energy with the differential energy.

The comparator 84 receives an output X of the voice activity detector 83 using the differential energy and a predetermined threshold Y, compares the two values, and outputs a control signal. That is, as a result of the comparison of the comparator 84, if X is greater than Y, the control signal is output so that the output of the buffer 82 is output to the delta modulator 86. If Y is greater than X, the output of the buffer 82 is 32k. A control signal is outputted to the adaptive differential pulse code modulator 87.

The switch 85 switches the output of the buffer 82 according to the control signal input from the comparator 84.

When the delta modulator 86 receives the output of the buffer 82 through the switch 85, the delta modulator 86 samples the input signal and determines the sign of the difference from the previous sample value. Send only 1 or 0 signal to outside.

Since the 32k adaptive differential pulse code modulator 87 uses a 15-level quantizer, it receives the output of the buffer 82 through the switch 85 and outputs an audio signal in units of 4 bits per sample. Since this is 8k (sample / second), 32kbps of sampled data is output. The most significant bit of the output 4 bits represents a sign (i.e., 0 if + and 1 if-). The remaining 3 bits represent the magnitude of the voice signal.

The frame storage unit and the decimal converter 88 store the data output from the 32k adaptive differential pulse code modulation unit 87 in a buffer of a frame length, and then easily compare the output codes of 4 bits in a comparator later. Convert to decimal.

The comparator 89 receives and compares the output codes converted into decimals in one frame from the frame storage unit and the decimal converter 88, and selects the maximum value and the minimum value.

The rate determining unit 90 determines whether the maximum value input from the comparator 89 falls within the range of the rate 1, the rate 2, the rate 3, and the rate 4, and then transmits the frame at the rate of the corresponding range.

The present invention configured and operated as described above has the following effects.

First, since the transmission rate is determined by the ratio between this value and the differential energy using the reference energy, the voice activity detection method using the differential energy can be easily adapted to the change of the background noise by applying the reference energy according to the magnitude of the background noise. There is an advantage.

Secondly, in the voice activity detection method based on the output code range, after retrieving the range of codes output by the encoding, the transmission rate is determined by the range of the maximum output code among the samples in the frame, so that the reproduction sound having the same sound quality as the maximum transmission rate is obtained. It can be effective.

Third, the voice activity detection method using differential energy and output code range usually operates at transmission rate 1 to transmission rate 3 in the voice section, so that it is possible to obtain reproducing sound with the same sound quality as when operating at maximum transmission rate. As the reference energy is increased and adjusted, the silent section operates at the transmission rate 4, thereby reducing the waste of bits used to transmit unnecessary background noise.

Fourth, the variable rate adaptive differential pulse code modulation apparatus applying the voice activity detection method using differential energy and output code range can reduce the average transmission rate. That is, since the transmission rate is determined according to the voice activity than the adaptive differential pulse code modulation (ADPCM) operating at 32k, the same quality of service as that of the 32k adaptive differential pulse code modulation (ADPCM) can be obtained at a lower transmission rate. The simulation results show an average data rate of 17kbps.

Fifth, the variable rate adaptive differential pulse code modulation apparatus applying the voice activity detection method using differential energy and output code range can increase the capacity of a code division multiple access (CDMA) system. That is, by applying the variable rate adaptive differential pulse code modulation (ADPCM) to the code division multiple access (CDMA) system, it is possible to increase the capacity by reducing the influence of interference between channels by transmitting the transmission power at a low transmission rate. .

The present invention described above is capable of various substitutions, modifications, and changes without departing from the technical spirit of the present invention for those skilled in the art to which the present invention pertains. It is not limited to the drawings shown.

Claims (11)

A voice activity detection means (22) using differential energy that receives a voice signal from the outside, divides it into frames, calculates differential energy, sets reference energy, and variably determines a transmission rate in comparison with the differential energy; Buffering means (21) for buffering a voice signal input from the outside while the data activity rate is determined by the voice activity detecting means (22) using the differential energy; And receiving information on the data rate to be operated from the voice activity detecting means 22, setting a quantization level and a variable corresponding to the data rate, and encoding a voice signal input from the buffering means 21 to a corresponding data rate. A variable rate adaptive differential pulse code modulation apparatus using a voice activity detection method comprising a variable rate adaptive differential pulse code modulation means (23) for outputting a voice code to the outside. 2. The variable rate adaptive differential pulse code modulation means (23) according to claim 1, wherein the variable rate adaptive differential pulse code modulation means (23) receives an input voice signal at a quantization level of 15 levels (4 bits) upon receiving information to operate at 32 kbps from the voice activity detection means (22). Quantized and outputted, and upon receiving information to operate at 24 kbps from the voice activity detecting means 22, the input voice signal is quantized and outputted at a quantization level of 7 levels (3 bits), and is output from the voice activity detecting unit 22. When receiving a signal to operate at 16kbps, the input audio signal is quantized and output at a three-level (2-bit) quantization level, and when receiving a signal to operate at 8kbps from the voice activity detector 22, a delta modulation method is used. A variable rate adaptive differential pulse code modulation apparatus using a voice activity detection method, characterized in that configured to output a signal of 1 or 0 by using. A voice activity detection method applied to a variable rate adaptive differential pulse code modulation device, comprising: firstly dividing an audio signal into frames having a predetermined length and then calculating differential energy dENG for each divided frame; ); A second step 13 of selecting a minimum value among a current differential energy dENG, a background noise maximum constant, and a low pass filtered reference energy and setting it as a reference energy B; And a third step (14 to 16) of comparing the set reference energy value with the differential energy of the current frame to variably determine the data rate according to the ratio of the energy to the background noise of the current frame. 4. The method of claim 3, wherein the differential energy calculation process of the first step (11, 12) is calculated by summing the difference between the current sample and the past sample of the speech signal over a frame period. The method according to claim 3 or 4, wherein the third steps (14 to 16) include a predetermined transmission rate 4 (8 kbps) if a value obtained by multiplying the predetermined fourth threshold value by the reference energy (B) is larger than the differential energy (dENG). A fourth step (14) of transmitting the frame data to the user; If the differential energy dENG is greater than the product of the predetermined fourth threshold and the reference energy B and less than the product of the predetermined third threshold and the reference energy B, the frame data at an arbitrary transmission rate 3 (16 kbps) is obtained. Transmitting a fifth step (15); If the differential energy dENG is greater than the value of the predetermined third threshold and the reference energy B and less than the value of the second threshold and the reference energy B, the frame data is transmitted at an arbitrary transmission rate 2 (24 kbps). A sixth step (16); And a seventh step of transmitting frame data at an arbitrary transmission rate 1 (32 kbps) when the differential energy is greater than the predetermined second threshold. Speech dividing means (41) for dividing a voice signal input from the outside into a frame having a predetermined length; a 32k adaptive derivative for outputting an encoded voice code by setting the quantization level and a variable by receiving the output of the speech dividing means (41) Pulse code modulation means 42; Frame storing means and decimal converting means (43) for storing the output code of said 32k adaptive differential pulse code modulation means (42) in a buffer for a frame length and then converting the output code into a decimal number; Comparison means (44) for selecting the maximum value and the minimum value after receiving and comparing the output code converted to the decimal number in one frame from the frame storage means and the decimal conversion means (43); And a rate determining means 45 for determining frame rate to which the maximum value input from the comparing means 44 belongs, and then transmitting frame data at a rate within the range corresponding to the variable rate adaptive differential pulse applying the voice activity detection method. Code modulator. A voice activity detection method applied to a variable rate adaptive differential pulse code modulation device, comprising: first dividing a voice signal into frames having a predetermined length and encoding each frame at a maximum data rate (32k ADPCM) to store an output code in a buffer; Steps 31 to 33; A second step 34 of retrieving an output code of each frame; And a third step (35 to 37) for variably determining a transmission rate according to the number of bits of the output code shown in the search result of the second step (34). 8. The method of claim 7, wherein the third step (35 to 37) comprises: a fourth step (35) of transmitting frame data at an arbitrary transmission rate 4 (8 kbps) if there is no output code of 2 bits or more; A fifth step 36 of transmitting frame data at an arbitrary transmission rate 3 (16 kbps) if there is no output code of 3 bits or more; A sixth step 37 of transmitting frame data at an arbitrary transmission rate 2 (24 kbps) if there is no output code of 4 bits or more; And a seventh step of transmitting frame data at an arbitrary transmission rate 1 (32 kbps) when there is an output code of 4 bits or more. Voice dividing means (81) for dividing an audio signal input from the outside into a frame having a predetermined length; A voice activity detection means (83) using differential energy that receives a voice signal from the voice dividing means (81), divides it into frames, calculates differential energy, sets reference energy, and varies transmission rate in comparison with differential energy; Buffering means (82) for buffering a voice signal input from said speech dividing means (81) while determining a data transmission rate in said speech activity detecting means (83) using said differential energy; First comparing means (84) for receiving an output (X) of the voice activity detecting means (83) using the differential energy and a predetermined threshold (Y), comparing the two values and outputting a control signal; Switching means (85) for switching the output of said buffering means (82) in accordance with a control signal input from said first comparing means (84); Delta modulating means (86) for outputting a voice code (1 or 0) to the outside by modulating the output of the buffering means (82) through the switching means (85) by a delta modulation; 32k adaptive differential pulse code modulation means (87) for receiving the output of the buffering means (82) through the switching means (85) and outputting an encoded speech code by setting a quantization level and a variable; Frame storage means and decimal conversion means (88) for storing the output code of said 32k adaptive differential pulse code modulation means (87) in a buffer for a frame length and then converting the output code into a decimal number; Second comparison means (89) for selecting a maximum value and a minimum value after inputting and comparing output codes converted to decimal numbers in one frame from the frame storage means and the decimal conversion means (88); And a rate determining means (90) for determining frame rate to which the maximum value input from the second comparing means (89) belongs, and then transmitting frame data at a rate within a corresponding range. Differential pulse code modulation device. A voice activity detection method applied to a variable rate adaptive differential pulse code modulation device, comprising: first step (71,72) of calculating a differential energy (dENG) for each divided frame after dividing a voice signal into frames having a predetermined length; ); A second step 73 of selecting a minimum value among a current differential energy dENG, a background noise maximum constant, and a low pass filtered reference energy and setting it as a reference energy B; A third step 74 of transmitting frame data at an arbitrary transmission rate 4 (8 kbps) when a value obtained by multiplying the predetermined fourth threshold by the reference energy B is greater than the differential energy dENG; And a fourth step (75 to 79) of encoding each frame at a maximum data rate (32k ADPCM), storing the output code in a buffer, searching for the output code of each frame, and variably determining the data rate according to the number of bits of the output code. Voice activity detection method comprising a. 11. The method of claim 10, wherein the step of variably determining the transmission rate according to the number of bits of the output code in the fourth step (75 to 79), if there is no output code of more than 2 bits, the frame data at an arbitrary transmission rate 4 (8kbps) A fifth step 77 of transmitting; A sixth step 78 of transmitting frame data at an arbitrary transmission rate 3 (16 kbps) if there is no output code of 3 bits or more; A seventh step 79 of transmitting frame data at an arbitrary transmission rate 2 (24 kbps) if there is no output code of 4 bits or more; And an eighth step of transmitting frame data at an arbitrary transmission rate 1 (32 kbps) when there is an output code of 4 bits or more.