KR20060008078A - A method and a apparatus of advanced low bit rate linear prediction coding with plp coefficient for mobile phone - Google Patents

Abstract

A voice coding apparatus and method of a mobile communications terminal can embody higher compressibility and ensure high sound quality, compared with the case of using a Linear Prediction (LP) coefficient, by performing a Linear Predictive Coding (LPC) using a Perceptual Linear Prediction (PLP) coefficient.

Description

Improved low-rate linear predictive coding apparatus and method for mobile terminals {A METHOD AND A APPARATUS OF ADVANCED LOW BIT RATE LINEAR PREDICTION CODING WITH PLP COEFFICIENT FOR MOBILE PHONE}

1 is a block diagram of a typical low-prediction linear predictive coded speech signal output model;

2 is a functional block diagram of a low-rate linear predictive coding apparatus of a conventional mobile terminal;

3 is a flow chart of a low-rate linear prediction coding method of a portable terminal according to the prior art;

4 is a functional block diagram of an improved low-rate linear predictive coding apparatus of the present invention portable terminal.

5 is a detailed functional diagram of the PLP counter of the mobile terminal of the present invention;

6 is a flowchart of an improved low-rate linear predictive coding method of a mobile terminal of the present invention.

** Explanation of symbols on the main parts of the drawing **

500: encryption unit 510: PLP counter 511: FFT unit

512: filter bank section 513: loudness section 514: matching section

515: inverse FFT part 516: phase processing part 517: frequency characteristic part

520: voiceless voice identification unit 530: voice breaking unit 540: pitch detection unit

550: coding unit

The present invention relates to low bit rate coding of a speech signal of a mobile terminal, and in particular, improved low bit rate linear prediction coding of a portable terminal for transmitting a low bit rate while coding a speech signal at a high compression rate using a PLP coefficient reflecting an auditory effect. An apparatus and method are provided.

The mobile communication system freely moves within the service area formed by the base station (RAN: RADIO ACCESS NETWORK) using a mobile terminal (MS: MOBILE STATION) and monitors and controls the mobile switching station (MSC: MOBILE SWITCHING CENTER). Through the communication path set by the switching (SWITCHING), it is an advanced wireless communication equipment to move anywhere and always carry the person directly by wireless connection to communicate with the desired party anytime and anywhere.

The mobile communication system including the mobile terminal (UE) is initially operated in a communication method using a voice level signal, and gradually increases the communication request and the amount of data to be transmitted, message data communication using numbers, letters, symbols, etc. Functions have been added, and current generation 3GPP mobile communication systems are being developed in such a manner as to provide multimedia-class communication in which video signals are included in the above-mentioned voice and text signals.

The mobile communication system uses a limited radio resource to occupy and occupy a plurality of subscribers at the same time, transmit a communication signal, and the channel is limited in bandwidth so that the channel can be used simultaneously. BIT RATE) is limited.

Since the data rate that can be transmitted through the allocated channel is limited as described above, a plurality of channels can be simultaneously assigned and communicated with each other. However, the communication users want to transmit a lot of data to each other.

As described above, a technique of transmitting a large amount of data using a limited data transmission rate of an allocated channel is a compression coding technique. The compression coding technique transmits a large amount of information with a smaller size as the compression ratio is higher.

There are many coding schemes for voice signal compression in the mobile terminal, and a compression coding scheme divided into data transmission rates requires a wide bandwidth or a channel having a high transmission rate, and has a high data rate higher than about 16 KBPS. BIT RATE coding method and a channel having a normal bandwidth or transmission rate, the medium bit rate coding method and a low bandwidth or transmission channel with a transmission rate ranging from 2.4 KBPS to 16 KBPS As a need for a low bit rate (LOW BIT RATE) coding scheme using a data rate ranging from 75 BPS to 2.4 KBPS.

In the coding of the speech signal, a sampling rate is generally 8 KBPS, and the high bit rate coding scheme is a speech coding scheme using general audio coding, a pulse code modulation (PCM) coding scheme, or an ADPCM scheme. (ADAPTIVE DELTA PULSE CODE MODULATION) coding scheme, and the medium bit rate coding scheme is based on CELP and corresponding VARIATION, LD-CELP coding scheme, CS-ACELP coding scheme, VSELP coding scheme, MELP coding scheme And a wideband speech coding method in which a high frequency component (8 KHz to 16 KHz) is coded in another manner as necessary.

The low bit rate coding scheme may include a LINEAR PREDICTIVE CODING (LPC) coding scheme, a RANDOM EXCITED LINEAR PREDICTIVE CODING (RELP) coding scheme, a FORMANTS VOCODER coding scheme, and a CEPSTRAL VOCODER coding scheme.

Since the high rate coding transmits a lot of data, the sound quality transmitted and received is best, and the sound quality is important. The medium rate coding is sound quality in general, and the low rate coding is not important for sound quality. Applicable to sufficient applications.

Since the data rate is fixedly limited in the channel allocated to the mobile terminal, the amount of information to be transmitted includes a voice signal, a data signal such as a text, and a video or image signal. In addition to using compression coding techniques, speech-grade signals adopt a low rate coding scheme that occupies less bandwidth.

Therefore, when voice is compressed and transmitted using a low bit rate coding scheme as described above, there is a problem in that the sound quality is bad. Therefore, there is a need to develop a technology for improving sound quality and improving compression efficiency in the low bit rate coding scheme. .

Hereinafter, a low bit rate voice coding apparatus and method of a portable terminal according to the prior art will be described with reference to the accompanying drawings.

Attached to explain the prior art, FIG. 1 is a functional block diagram of a typical low-prediction linear predictive coding voice signal output model, and FIG. 2 is a functional block diagram of a low-rate linear predictive coding apparatus of a mobile terminal according to the prior art. 3 is a flowchart of a low-rate linear predictive coding method of a mobile terminal according to the prior art.

Referring to FIG. 1, a general low-prediction linear predictive coded voice signal output model will be described. The voiced sound unit 10 receives a pitch period signal and impulses the period of the input pitch period signal. Generate and output VOICE SOUND by IMPULSE TRAIN, and the voiced sound output from the voiced sound unit 10 is applied to the GLOTTAL filter unit 20 to form a voice signal component. It is outputted with SHAPING.

Since the signal output from the global filter unit 20 is applied to the multiplier unit 30 and multiplied by a VOICE GAIN value, the signal is adjusted to an appropriate level and applied to the switch unit 60.

The unvoiced sound unit 40 generates a unvoiced sound signal based on random noise and outputs it to the corresponding multiplier unit 50, and the multiplier unit 50 generates an noise due to noise gain. Since the value is multiplied, it is adjusted to an appropriate level and applied to the switch unit 60.

Since the switch unit 60 selects and outputs a voiced signal and an unvoiced signal, the switch unit 60 outputs an excited signal (EXCITATION SIGNAL) Ug [n], and the excited signal Ug [N] is a vocal filter unit ( 70 is output as a signal Hvocal (z) processed by VOCAL TRACT.

The Hvocal (z) is a DISCRETE-TIME MODELING of the vocal filter unit 70, and performs linear prediction coding (LPC: LINEAR PREDICTION CODING) based on this.

That is, the unvoiced signal (UNVOICED SIGNAL) is not generated by being excited as RANDOM NOISE, but the LPC vocoder (VOCODER) is excited by RANDOM NOISE, and the period of the voiced sound part 10 becomes pitch (PITCH), and PITCH EXTRACTION Extract with ALGORITHM.

The signal Hvocal (z) of the vocal filter unit 70 is applied to the radiation filter unit 80 and radiated (Hrad (z)) so as to give the final speech signal S [n]. Output

With the above configuration, a general low-prediction linear predictive coded speech signal output model is described in detail in the document "DIGITAL PROCESSING OF SPEECH SIGNAL" ENGLEWOOD CLIFFS, NJ: PRENTICE HALL, published in 1978 by LR RABINER and RW SCHAFER. It is.

In the above model, the pitch period (PITCH PERIOD) is slightly different, but the function of outputting the speech level speech signal can be described as a circuit function.

Hereinafter, a low rate linear predictive coding apparatus of a portable terminal according to the prior art will be described with reference to FIG. 2.

When the speech signal S [n] is input to the encryption unit 100 of the portable terminal, a signal rx obtained by inputting from the autocorrelation unit 110 to compute an AUTOCORRELATION function FUNCTION. [n]) is output, and the signal rx [n] is applied to the LP coefficient unit 130 to calculate the coefficients an and the gain G, and output them to the coding unit 160.

In addition, since the speech signal S [n] input to the encryption unit 100 is applied to and analyzed by the voiced / voiceless identification unit 120, it generates a corresponding identification signal for distinguishing whether it is a voiced sound signal or an unvoiced sound signal. At the same time, the speech signal S [n] is outputted to the unblocking unit 140 to block the unvoiced sound signal, and passes only the voiced sound signal to the pitch detection unit 150. do.

The pitch detector 150 detects a pitch period signal P from the voiced sound signal input and outputs the pitch period signal P to the coding unit 160.

The coding unit 160 encodes the input coefficient (an), the gain (G), the pitch period (P), and the voiced and unvoiced identification signal by parameter coding (CODING) or a low data rate to control the control unit ( 300 is output to the wireless unit 400 by the corresponding process of the control unit 300, and transmits to the designated counterpart.

The low rate voice level signal received from the other party through the wireless unit 400 is applied to the decoding unit 200 by the corresponding analysis and control of the control unit 300, the decoding unit 200 is the decoding unit 210 Inputting from, separates and detects the corresponding parameter and outputs each detected parameter.

Among the parameters output from the decoding unit 210, a pitch period (P) parameter is applied to the voiced sound generator 220 to generate a voiced sound signal by the pitch period, the generated voiced sound is the switch unit 240 Output to

The switch unit 240 also inputs an unvoiced sound signal applied from the unvoiced sound generator 230, inputs a voiced / unvoiced identification signal parameter among the parameters output from the decoder 210, and the voiced sound generator 220. The voiced sound signal applied from) and the unvoiced sound signal applied from the unvoiced sound generator 230 are selectively output to the multiplier unit 250.

The multiplier unit 250 adjusts the voiced sound signal and the unvoiced sound signal input from the switch unit 240 to a predetermined level by a gain (G) parameter applied from the decoding unit 210 to the synthesizer unit 260. The synthesizer 260 demodulates the speech signal S [n] because the synthesizer 260 processes a signal input from the multiplier 250 by a coefficient applied from the decoder 210. Output

The low-rate linear predictive coding device according to the prior art of the above-described configuration has a problem that the configuration is complicated, bulky, manufacturing time and cost are high.

Hereinafter, with reference to the accompanying FIG. 3, a low-rate linear predictive coding method of a portable terminal will be described.

In the case of low bit rate coding (CODING) and transmitting the voice signal to the mobile terminal (MS) (S10), AUTOCORRELATION operation (COMPUTATION) of the input voice signal or speech signal (S20) , LP coefficient (COEFFICIENT) is calculated (S30).

Using the coefficient value (an), the gain value (G), the pitch period value (P), and the voiced / unvoiced identification signal obtained in the operation, the parameter is coded and transmitted to the counterpart designated by the low bit rate method (Low Bit Rate). S40).

When the mobile terminal MS receives a voice signal from the other party and decodes the signal (S50), the mobile terminal MS decodes the received low-rate signal as a demodulated voice-level signal or a speech signal (S60). .

In the prior art having the above-described configuration, the autocorrelation operation and the Elp coefficient operation are performed in the process of coding or encoding the voice signal, so that the process is complicated and does not reflect the human auditory effect, so in terms of MSE (MEAN SQUARED ERROR) There is a problem that a large error occurs.

The present invention encodes a voice signal of a mobile terminal and transmits it at a low data rate, thereby processing PLP coefficients, thereby reflecting the human auditory effect and increasing the sound quality by reducing errors in the MSE concept. It is an object of the present invention to provide an improved low-rate linear predictive coding apparatus and method of a portable terminal which reduces bitrate to improve compression efficiency.

The present invention devised in order to achieve the above object, the PLP counter for outputting the coefficient value and the gain value by processing the PELP coefficient of the speech-level speech signal input to the encryption unit of the mobile terminal reflecting the auditory effect Wow; A voiceless voice discrimination unit configured to determine whether the voice level speech signal input to the PLP counter is a voiced sound signal or an unvoiced voice signal and output an identification signal; An unvoiced blocker configured to input an output signal of the voiceless voice discriminator and to block only the voiced sound signal to pass only the voiced sound signal; A pitch detector for detecting and outputting a pitch period by analyzing voiced sound signals applied from the unvoiced cutoff unit; Characterized in that the configuration comprises a coding unit for low-order low-rate parameter coding using the values applied from the PLP coefficient unit, the pitch detection unit and the voiceless voice discriminator, respectively.

In addition, the present invention devised to achieve the above object, the process of determining whether to transmit the voice level speech signal input to the encryption unit of the portable terminal at a low transmission rate; An FFT process of performing fast Fourier transform processing and filter bank processing when it is determined that the low data rate is transmitted in the above process; An auditory process of outputting the signal of the process by loudness processing and power matching processing suitable for human hearing; An inverse FFT process of performing a reverse fast Fourier transform process on the signal of the process and performing phase correction processing suitable for hearing; Characterized in that the signal of the above process is characterized in that the process consisting of a transmission process for transmitting at a low transmission rate by processing the frequency characteristics and parameter coding.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described with reference to the accompanying drawings.

4 is a functional block diagram of an improved low-rate linear predictive coding device of a mobile terminal according to the present invention, and FIG. 5 is a detailed functional configuration of a PLP counter of a mobile terminal according to the present invention. 6 is a flowchart illustrating an improved low bit rate linear prediction coding method of a mobile terminal according to the present invention.

Referring to FIG. 4, according to the present invention, an improved low-rate linear predictive coding apparatus of a mobile terminal will be described. A voice level speech signal S input to an encryption unit 500 of a mobile terminal MS will be described. [n]) processes the PEL (PLP: PERCEPTUAL LINEAR PREDICTION) coefficients to which the auditory effect is reflected and outputs the coefficient value (an) and gain value (G). An FFT unit 511 for processing a FOUR FOURIER TRANSFORM (FFT) to output a speech-grade speech signal input to the encryption unit 500; A filter bank (CRITICAL BAND INTEGRATION AND RESAMPLING) unit 512 for inputting the signal of the FFT unit 511 to filter, sample, and output the filter according to a predetermined frequency band; An equalizer process and a loudness process for inputting the signal of the filter bank unit 512 to be suitable for human hearing; A matching part (514) for inputting the signal of the loudness unit (513) and adjusting the power level to suit the hearing of a person; An inverse FFT unit 515 for inputting a signal of the matching unit 514 and outputting an inverse fast Fourier transform (FFT) process; SOLVING SET OF LINEAR EQUATIONS unit 516 for inputting the signal of the inverse FFT unit 515, and phase-compensating and outputting the image according to human hearing; A voice level speech signal, which is input to the encryption unit 500, is composed of a frequency characteristic (CEPSTRAL RECURSION) unit 517 which inputs the signal of the phase processing unit 516 and compensates and outputs the frequency characteristic suitable for human hearing. The PLP coefficient unit 510 which detects parameters of the coefficient value and the gain value reflected by the low order of the human auditory effect and outputs them to the coding unit 550,

Determining whether the voice level speech signal input to the PLP counter 510 is a voiced sound signal or an unvoiced sound signal and outputting the corresponding identification signal. The voice level speech signal input to the encryption unit 500 is analyzed and voiced sound. In the case of the VOICE SOUND signal, the voiced voice identification signal is output to the coding unit 550 as a parameter. In the case of the unvoiced sound signal, the voiced voice identification unit 520 is output to the coding unit 550 as a parameter. Wow,

An unvoiced blocker 530 which inputs the output signal of the voiced voice discrimination unit 520 and blocks the unvoiced sound signal and passes only the voiced sound signal;

The pitch detection unit detects and outputs a pitch period by analyzing voiced sound signals applied from the unvoiced cutoff unit 530. The pitch detector detects the pitch period value P of the voiced sound signal and outputs the parameter value to the coding unit. With 540,

By using the values applied from the PLP counter 510, the pitch detector 540, and the voiceless voice discriminator 520, respectively, the low order low bit rate parameter coding is performed. Input a coefficient value (an) and a gain value (G) applied as a parameter from (510), and inputs the pitch period value (P) applied from the pitch detection unit 540 as a parameter, the planetary voice identification unit ( A coding unit 550 for inputting a voiceless voice identification signal applied from 520 as a parameter, processing a parameter coding, and outputting at a low bit rate is included.

Hereinafter, the present invention having the above-described configuration will be described in detail with reference to the accompanying drawings.

The present invention provides a PLP using a lower order than a linear predictive coding (LPC) method in transmitting a speech level speech signal to a counterpart at a high compression rate and a low transmission rate in a mobile terminal. By using the PERCEPTUAL LINEAR PREDICTION coefficient unit 510, the compression rate is increased and the transmission rate is lowered at a lower transmission rate than when using the conventional LAR (LINEAR PREDICTION COEFFICIENT).

The large difference between the LPC applied in the prior art and the PLP applied in the present invention are as follows.

The LPC basically calculates an ak value such that MSE (MEAN SQUARED ERROR) is the minimum value according to the following equation.

(Formula 1)

The order of the coefficients obtained by the above formula is about 10th order (8th to 12th order) when the 8 KHz sampling rate has a sampling rate, and the LP coefficients obtained as described above use the linear prediction. Applied to various codings such as LPC, CELP, MELP, RELP, and the like, which is made by WB KLEIJIN and KK PALIWAL, which are described in more detail in SPEECH CODING AND SYNTHESIS, AMSTERDAM, THE NETHERLANDS, ELSEVIER, 1955.

The PLP method applied to the present invention, which was introduced in a 1990 paper by HERMANSKY, uses the HUMAN AUDITORY property like the existing MFCC (MEL-FREQUENCY CEPSTRAL COEFFICIENT), but DCT (DISCREET-TIME COSINE TRANSFORM) is performed. On the other hand, LEVINSON-DURBIN RECURSION is different, as does LINEAR PREDICTION, and B. GOLD AND MORGAN, SPEECH AND AUDIO PROCESSING, NEWTO, JOHN WILEY & SONS, INC, 2000, and H. HECEANSKY, PERCEPTUAL LINEAR PREDICTIVE (PLP) ANALYSIS OF SPEECH, J. ACOUST. SOC. AM. VOL. 87, PP. 1738-1752, 1990, described in more detail.

That is, in the present invention, it is important to apply a PLP (PERCEPTUAL LINEAR PREDICTION) coefficient instead of applying a LINEAR PREDICTION COEFFICIENT (LPC) coefficient in LPC.

로 표현된다. A SYNTHESSIS FILTER is constructed by applying general LPC coefficients, and when the LPC coefficient of the filter is given as ai, the ERROR PREDICTION FILTER of the synthesized filter is

It is expressed as

PITCH information and voiced / unvoiced information extracted using the above equation are applied, and the LPC coefficient, the pitch information, and the voiced voice identification signal are transmitted.

However, in the present invention, the spectrum (SPECTRUM) is obtained by applying the PLP coefficients.

Since the human auditory effect is reflected in the PLP coefficients, the MSE value of the spectrum calculated by applying the PLP coefficients is larger than the value using the LPC coefficients, but the acoustic effect is considered. The error becomes smaller.

When the SHORT-DELAY PREDICTION is performed with the PLP coefficient, when the sampling rate is set to 8 KHz, the 7th coefficient is applied to reduce the bit rate, and thus the parameter information by the voiceless voice identification signal and the pitch period signal should be transmitted. However, no lower order can be applied.

Therefore, since the seventh order is applied than the conventional tenth order, the transmission rate of about 30% is lowered.

The synthesizer, which is composed of a decoder, generates and outputs a speech speech signal and performs a synthesizer filter in the same or similar manner as in the case of using the LPC coefficient, and is expressed by the following equation.

(Formula 2)

Referring to the configuration of the present invention in more detail, when the speech level speech signal S [n] is input to the encryption unit 500, the PLP input unit 510 simultaneously inputs the same signal to the voiceless voice identification unit 520 ).

The PLP counter 510 processes the input speech signal S [n] in the FFT unit 511 by FAST FOURIER TRANSFORM, and performs the filter bank unit 512 in one step. For example, by the configuration of about 26 filter banks 512, a noise component is removed in units of frequencies by respective filters, and applied to the loudness unit 513 so that an equalizer is applied to a signal suitable for a person. EQUALIZER) processing and LOUDNESS processing.

The speech-level speech signal processed as described above is applied to the matching unit 514 to be matched with power suitable for human listening, and is applied to the inverse FFT unit 515 to be inversed. Fast Fourier Transform (FFT) processing, and applied to the phase processing unit 516 to perform phase correction processing so that a human is audibly audible phase (PHASE), and applied to the frequency characteristic unit 517 The frequency characteristic compensation process is output so that this acoustic characteristic is reflected.

As described above, the PLP counting unit 510 outputs the input speech level signal to the coding unit 550 as low-order coefficient values (an) and gain values (G) reflecting human auditory characteristics. do.

The voiceless voice discriminator 520 outputs a speech signal to the coding unit 550 and determines whether the voice speech signal input is a voiced sound signal or a unvoiced sound signal. Since it is applied to the unvoiced blocking unit 530, the unvoiced sound signal is blocked from passing, and only the voiced sound signal is passed.

The pitch detecting unit 540 which inputs only the voiced sound signal by the unvoiced cutoff unit 530 detects the pitched voice signal by analyzing the input voiced sound signal, and encodes the detected pitch period P by the coding unit ( 550) as a parameter value.

The coding unit 550 is applied from the PLP coefficient unit 510, and inputs a coefficient value (an) and a gain value (G) parameter reflecting the auditory characteristics, and receives the voiced signal from the voiceless voice discriminator 520. Input the unvoiced identification signal as a parameter, and input the pitch period (P) from the pitch detection unit 540 as a parameter.

Since the coding unit 550 processes the speech-grade speech signal using only the parameter values respectively input as described above, the coding unit 550 compresses and outputs the low-rate data with low order of coefficient.

Therefore, the present invention having the above configuration has the advantage of lowering the order by using the PLP coefficient, thereby increasing the compression rate and transmitting the voice level signal at a lower low transmission rate.

Hereinafter, the present invention will be described with reference to the attached FIG. 6, which describes an improved low-rate linear predictive coding method of a mobile terminal.

A process of determining whether to transmit a voice level speech signal input to the encryption unit 500 of the mobile communication mobile terminal (MS) at a low bit rate (S100);

If it is determined in step S100 that the low transmission rate is transmitted, the fast Fourier transform processing and the filter bank processing, the voice-quality speech signal input to the encryption unit 500 of the mobile terminal (MS) Fast Fury A transform (FFT) process (S110); An FFT process comprising a process of removing a noise component to be suitable for an auditory characteristic of a person by performing a filterbank process on the voice level speech signal of the process (S110),

Loudness processing and power matching to output the signal of the FM process, the process of passing the voice level speech signal of the FM process with sound components of a size suitable for human hearing (S130); An auditory process comprising a step (S140) of matching an output power so that a person is audibly audible to the signal of the process (S130),

Performing a reverse fast Fourier transform on the signal of the auditory process and a phase correction process suitable for hearing, and performing a reverse fast Fourier transform (IFFT) on the speech level speech signal of the auditory process; An inverse FFT process comprising a step S160 of performing a phase compensation by a DURBIN process of SOLVING SET OF LINEAR EQUATION to make the speech level speech signal of step S150 suitable for human hearing;

Capturing the signal of the inverse FFT process to a low frequency rate by processing the frequency characteristics and parameter coding to suit the hearing, the capsular riker to the speech level speech signal of the reverse fast Fourier transform process suitable for human hearing Compensating for the frequency characteristic by a CEPSTRAL RECURSION process (S170); The transmission process consists of a step (S180) of transmitting the voice level speech signal of the step (S170) by coding (CODING) with each corresponding parameter at a low bit rate (LOW BIT RATE).

Hereinafter, the present invention having the above configuration will be described in detail with reference to the accompanying drawings.

The encryption unit 500 of the mobile terminal for mobile communication determines whether to transmit the input voice level signal to the other party at a low bit rate (SLOW BIT RATE) (S100), and transmits the voice level signal at a low bit rate in the determination (S100). In this case, the fast Fourier transform process (S110) and the filter bank process to remove the noise signal (S120).

Loudness processing of the voice level signal processed as described above, and outputs a component suitable for the human auditory characteristics (S130), and a power matching process for adjusting the output level to be appropriate to hear the signal by the auditory characteristics (S140) Inverse fast Fourier transform (FFT) processing (S150).

The signal processed as described above is subjected to phase correction processing so as to be suitable for hearing by human hearing (S160), and frequency characteristic compensation processing so as to be suitable for auditory characteristics (S170).

As described above, the PLP coefficient processing unit processes the input speech level speech signal to reflect the human auditory characteristics, thereby encoding the coefficient value (an) and the gain value (G) whose parameters are lower than the coding unit 550. Since the coding unit 550 inputs the pitch period (P) parameter and the voiceless voice identification signal to each other, the coding unit 550 compresses at a high compression rate according to a low order and simultaneously codes at a low data rate to transmit to a designated counterpart. (S180).

Therefore, since the present invention uses the PLP coefficient compared to the conventional technique using the LPC coefficient, the present invention has the advantage of improving the high compression rate and further improving and transmitting the voice level signal at a lower low transmission rate.

According to the present invention having the above-described configuration, since the PLP coefficient is used to transmit a voice level signal, there is an industrial use effect of reflecting an auditory characteristic and increasing the compression ratio by lowering the order of the coefficient.

In addition, since the voice signal is transmitted at an improved low transmission rate, there is a convenient effect of using the channel efficiently.

Claims (10)

A PLP counter for outputting a coefficient value and a gain value by processing PLP coefficients in which a voice level speech signal input to an encryption unit of the mobile terminal reflects an auditory effect; A voiceless voice discrimination unit for determining whether the voice level speech signal input to the PLP counter is a voiced sound signal or an unvoiced voice signal and outputting an identification signal; An unvoiced blocker configured to input an output signal of the voiced voice discriminator and to block only the voiced sound signal to pass only the voiced sound signal; A pitch detector for detecting and outputting a pitch period by analyzing voiced sound signals applied from the unvoiced cutoff unit; An improved low-rate linear predictive coding apparatus of a mobile terminal, characterized by comprising a coding unit for coding a low-order low-rate parameter using values applied from the PLP coefficient unit, the pitch detector unit, and the voiceless voice discriminator unit, respectively. . According to claim 1, wherein the PLP counting unit, A FFT unit which outputs the voice level speech signal input to the encryption unit of the mobile terminal by performing fast Fourier transform processing; A filter bank unit for inputting the signal of the FFT unit, filtering, sampling, and outputting the filter according to a predetermined frequency band; A loudness unit for inputting a signal of the filter bank unit to output an equalizer process and a loudness process to be suitable for human hearing; A matching unit for inputting a signal of the loudness unit and adjusting a power level according to hearing of a person; An inverse FFT unit for inputting a signal of the matching unit and outputting an inverse fast Fourier transform process; A phase processing unit which inputs a signal of the inverse FFT unit and phase-compensates and outputs a phase suitable for human hearing; And a frequency characteristic unit configured to input a signal of the phase processing unit and compensate and output a frequency characteristic to be suitable for hearing of a person. According to claim 1, wherein the PLP counting unit, Improvement of the mobile terminal, characterized in that it consists of analyzing the speech level speech signal input to the encryption unit of the mobile terminal and detecting the coefficient value and gain value parameters reflecting the human auditory effect and outputting them to the coding unit. Low-rate linear predictive coding device. According to claim 1, The voiceless voice identification unit analyzes the voice level speech signal input to the encryption unit of the mobile terminal, and outputs the voiced voice identification signal as a parameter to the coding unit if the voiced voice signal, and converts the voiced voice identification signal to the coding unit if the voiced signal is a parameter. Output, And the pitch detecting unit is configured to detect a pitch period value of the voiced sound signal input from the unvoiced blocking unit, and output the coding unit as a parameter value to the coding unit. The method of claim 1, wherein the coding unit, Input a coefficient value and a gain value applied as a parameter from the PLP coefficient unit, input a pitch period value applied from the pitch detection unit as a parameter, and input a voiced voice identification signal applied from the voiceless voice identification unit as a parameter Improved linearity predictive coding apparatus of a portable terminal, characterized in that the parameter coding process and output at a low transmission rate. A process of determining whether to transmit a speech level speech signal input to an encryption unit of the portable terminal at a low transmission rate, and In the process, if it is determined that the low-rate transmission, the fast Fourier transform processing, the filter bank processing FFT process, An auditory process of outputting the signal of the process by loudness processing and power matching process suitable for human hearing; An inverse FFT process of performing a reverse fast Fourier transform process on the signal of the above process and performing phase correction processing suitable for hearing; An improved low-rate linear predictive coding method of a portable terminal, characterized by comprising: a transmission process for processing the signal of the process in a frequency characteristic suitable for hearing and transmitting at a low data rate by parameter coding. The method of claim 6, wherein the FFT process, Performing a fast Fourier transform process on the voice level speech signal input to the encryption unit of the mobile terminal; An improved low rate linear predictive coding method of a portable terminal, characterized in that it comprises a process of removing the noise component by filtering the speech level speech signal of the process. The method of claim 6, wherein the hearing process, A process of loudness passing the voice level speech signal of the fft process with a sound component having a size suitable for human hearing; And a step of matching the output power so that a person is audibly audible to the signal of the above process. The method of claim 6, wherein the reverse FFT process, Performing a reverse fast Fourier transform process on the speech level speech signal of the auditory process; An improved low-rate linear predictive coding method of a portable terminal, characterized by comprising the step of performing phase compensation of the speech level speech signal of the above process by a dubin process suitable for human hearing. The method of claim 6, wherein the transmission process, Compensating for the frequency characteristics by capsular recursion processing for the speech-grade speech signal of the inverse fast Fourier transform process, for human hearing; An improved low-rate linear predictive coding method of a portable terminal, characterized in that it comprises the step of coding the speech speech signal of the process with each corresponding parameter to transmit at a low transmission rate.

Images