KR20050070338A - Apparatus and its method for providing echo cancellation using delay prediction - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

The present invention relates to a speech coding / decoding apparatus having an echo cancellation function according to a delay prediction and a method thereof.

2. The technical problem to be solved by the invention

The present invention generates a pilot signal (P ₁ (n)) out of an audible signal to a voice output unit when outputting a voice signal in a voice telephone or the like, and then inputs a pilot signal P ₂ (n) input from the voice input unit. It is an object of the present invention to provide a speech coding / decoding device having an echo cancellation function according to a delay time prediction, for effectively canceling an echo signal by accurately predicting a delay time from a difference between a signal and a signal.

3. Summary of the Solution of the Invention

The present invention provides a speech coding / decoding apparatus having an echo cancellation function according to a delay time prediction. After decoding a speech code received through a transmission channel and converting the speech code into a speech signal, the pilot signal P ₁ (n) is converted into a speech signal. In addition, speech decoding means for outputting the speech output means; The delay time is predicted by comparing the pilot signal P ₁ (n) received from the speech decoding means with the pilot signal P ₂ (n) extracted from the speech signal Vi (n) received through the speech input means. Echo canceling means for removing echo from the speech signal Vi (n) received through the speech input means in accordance with the estimated delay time; And speech encoding means for receiving the speech signal V (n) from which the echo signal is removed through the echo canceling means and encoding the speech signal with a speech code.

4. Important uses of the invention

The present invention is used for the Internet voice telephone (VoIP) terminal.

Description

Speech coding / decoding device having echo cancellation function according to delay prediction and its method {APPARATUS AND ITS METHOD FOR PROVIDING ECHO CANCELLATION USING DELAY PREDICTION}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a speech coding / decoding apparatus having an echo cancellation function according to a delay time prediction, and more particularly, to a pilot signal (P ₁ ) outside an audible signal when outputting a speech signal in a voice telephone or the like. (n)) is generated by the voice output unit, and the echo cancellation according to the delay time prediction is performed to effectively remove the echo signal by accurately predicting the delay time from the difference from the pilot signal P ₂ (n) input from the voice input unit. A voice coding / decoding device having a function and a method thereof are provided.

In the following example, an Internet voice telephone (VoIP) will be described as an example. However, the present invention is not limited to the Internet voice telephone (VoIP).

A conventional voice coding / decoding device (codec) of a VoIP terminal is a voice decoder for decoding a voice code received through a transmission channel and outputting the voice code to the speaker 13. (11) and a voice encoder (12) for encoding and compressing and transmitting the voice signal of the talker received through the microphone (14). In this process, the decoded voice signal (Vd (n), n is an integer and means the time of the quantized and sampled digital voice signal) is partially inputted through the microphone 14, and thus the terminal of the other party. Echo occurs at, and the same echo occurs in the reverse direction.

In order to solve such a problem, conventionally, the microphone and the speaker are separated by a predetermined distance to block a signal that is physically reflected, and thus a method of suppressing the generation of the echo signal is used. For example, an earphone or a directional microphone is an example.

However, this method causes inconvenience for the user because it requires additional equipment (eg, earphone). In addition, in a multi-party call environment using a speakerphone, there is a problem in that echo is not completely removed by a physical echo cancellation method that separates the microphone and the speaker more than a predetermined distance. Developed to compensate for this, a speech coding / decoding device (codec) with an adaptive echo cancellation device as shown in FIG.

As shown in FIG. 2, the speech coding / decoding apparatus including the adaptive echo canceller is configured to convert the speech signal Vd (n) output from the speech decoder 21 into the adaptive echo canceller 23. After delaying for a predetermined time and multiplying the weights, they are subtracted from the voice signal Vi (m) input from the microphone 25.

However, in the conventional method as described above, it is assumed that the distance between the speaker and the microphone is in a constant section (for example, 0.5 m), and the delay time is determined based on the calculated value considering the air velocity (Vs meter / sec) of the voice signal. It is set.

However, when the VoIP terminal is a personal computer (PC) equipped with a speaker and a microphone, the location of the speaker and the microphone is different for each user or each use time. There is a problem that it is difficult to remove the echo signal.

That is, the conventional method as described above has a problem that it is impossible to effectively remove the echo signal in a communication environment in which the positions of the speaker and the microphone are changed.

The present invention has been proposed to solve the above problems, and generates a pilot signal (P ₁ (n)) out of an audible signal to a voice output unit when outputting a voice signal in a voice telephone or the like, and then from the voice input unit. Provided is a speech coding / decoding device having an echo cancellation function according to a delay time prediction for effectively canceling an echo signal by accurately predicting a delay time from a difference from an input pilot signal P ₂ (n). There is a purpose.

The apparatus of the present invention for achieving the above object, in the speech coding / decoding apparatus having an echo cancellation function according to the delay time prediction, after decoding the speech code received through the transmission channel to convert to a speech signal, the pilot Speech decoding means for adding a signal P ₁ (n) and outputting it through the speech output means; The delay time is predicted by comparing the pilot signal P ₁ (n) received from the speech decoding means with the pilot signal P ₂ (n) extracted from the speech signal Vi (n) received through the speech input means. Echo canceling means for removing echo from the speech signal Vi (n) received through the speech input means in accordance with the estimated delay time; And speech encoding means for receiving the speech signal V (n) from which the echo signal is removed through the echo canceling means and encoding the speech signal with a speech code.

On the other hand, the method of the present invention, in the speech coding / decoding method having an echo cancellation function according to the delay time prediction, after the speech decoder decodes the speech code received through the transport channel to convert the speech code into a speech signal, the pilot signal ( A voice decoding step of adding P ₁ (n)) and outputting it through the voice output means; The echo canceller compares the pilot signal P ₁ (n) received from the voice decoder with the pilot signal P ₂ (n) extracted from the voice signal Vi (n) received through the voice input means. A delay time prediction step of predicting time; An echo canceling step for removing echo from the speech signal Vi (n) received through the speech input means according to the delay time predicted by the echo canceller in the delay time estimating step; And a speech encoding step for the speech coder to receive the speech signal V (n) from which the echo signal is removed through the echo canceller and to encode the speech signal into a speech code.

The above objects, features and advantages will become more apparent from the following detailed description taken in conjunction with the accompanying drawings. Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

3 is a block diagram of an embodiment of a speech coding / decoding apparatus having echo cancellation according to a delay prediction according to the present invention.

As shown in FIG. 3, the speech coding / decoding apparatus having an echo cancellation function according to a delay time prediction according to the present invention decodes a speech code received through a transmission channel and converts the speech code into a speech signal Vd (n). Then, the voice decoder 31 for adding the pilot signal P ₁ (n) to the speaker 34 and the pilot signal P ₁ (n) received from the voice decoder 31 and Compare the pilot signal P ₂ (n) extracted from the voice signal Vi (n) received from the microphone 35 to predict the delay time, and the voice input from the microphone 35 according to the predicted delay time. From the echo canceling unit 33 for outputting the audio signal V (n) from which the echo signal is removed from the signal Vi (n), and from the microphone 35 from which the echo signal is removed through the echo canceling unit 33. Sound to be output after receiving the audio signal of V (n) And a coder (32).

At this time, the voice decoder 31 generates a pilot signal P ₁ (n) at the time of outputting the voice signal to drive the speaker 34. Therefore, Vp (n) is P ₁ (n) in the initial N sample interval and is equal to Vd (n) in the subsequent interval. In general, since N corresponds to a voice signal section of about 0.1 second, it may be selected as a value obtained by multiplying the voice sampling frequency by 0.1 and taking an integer value.

On the other hand, the pilot signal P ₁ (n) uses a signal that cannot be recognized by a person. For example, a sinusoidal signal of a band that cannot be easily detected by the human ear in the audio frequency band may be used, and a plurality of sinusoidal signals may be used to reduce the prediction error of the delay time. At this time, the magnitude of the sinusoidal signal is determined to measure the magnitude of the signal Vi (n) input through the microphone 35 so as to easily predict the delay time Td. Typically, a sinusoidal signal generated at 30 decibels is an effective pilot signal that is not perceived by the human ear. On the other hand, if the magnitude of the sinusoidal signal is not determined to facilitate the prediction of the delay time Td, the amount of calculation of the echo canceller 332 is too large, making it difficult to implement real-time echo cancellation. Therefore, the size of the sinusoidal signal (e.g. 30 days) should be determined so that the calculation time is small while accurately predicting the delay time Td.

On the other hand, the echo canceller 33 compares the pilot signal P ₁ (n) from the voice decoder 31 with the pilot signal P ₂ (n) extracted from the voice signal from the microphone 35. A delay estimator 331 for predicting a delay time and an echo canceller 332 for removing an echo signal from a voice signal input from the microphone 35 according to the delay time received from the delay time predictor 331. do. In this case, a detailed operation process of the delay predictor 331 and the echo canceller 332 will be described in detail with reference to FIGS. 4 and 5, respectively.

4 is a detailed block diagram of a delay predictor among speech coding / decoding apparatuses having echo cancellation according to a delay prediction according to the present invention.

As shown in FIG. 4, the delay time predictor 331 according to the present invention is configured to extract the pilot signal P ₂ (n) from the voice signal Vi (n) received through the microphone 35. N-sample delay 42 for delaying the pilot signal P ₁ (n) received from the speech coder 32 in units of samples under the control of the band pass 41 and the controller 46, the band pass The subtractor 43 and the subtractor for calculating a difference between the pilot signal P ₂ (n) received from the device 41 and the pilot signal P ₁ (n) received from the N sample delay unit 42. An average calculator 44 for receiving the difference signal E (n) from 43 and calculating a mean absolute error value; a delay time Td at which the average absolute error value received from the average calculator 44 becomes the minimum; ) The minimum value predictor 45 for extracting the component and passing it to the echo canceller 332 and the controller 46, Includes a controller 46 for controlling the range of sample units in which the N sample delay unit 42 operates according to the delay time Td from the minimum predictor 45.

At this time, when there is no attenuation through the speaker 34 and the microphone 35, the pilot signal P ₂ (n) extracted from the band pass 41 is composed of a simple delay period component of P ₁ (n), In the actual implementation process, attenuation or amplification and various reflection paths are used. Therefore, the average absolute error value is calculated using the average values of P ₁ (n) and P ₂ (n) of a certain number of samples (eg 100 samples). ) And calculate the delay component that minimizes this value. In this case, the delay time component is the minimum value of the average absolute error signal calculated by the average calculator 44 while the N sample delayer 42 delays the signal of P ₁ (n) by one sample unit. 45).

In addition, the controller 46 controls the N sample delay unit 42 to operate at a predetermined total sample interval during system initial driving (when program starting or PC booting), and in the actual voice transmission / reception process, the minimum predictor The N sample delay unit 42 is controlled to operate in the sample interval within a predetermined range (+.-) By reusing the delay time extracted in (45).

Meanwhile, the output signal Td of the delay time predictor 331 is input to the echo canceller 332. The echo canceller 332 may be configured as an adaptive digital filter similar to that of the related art. However, in the present invention, since the delay time is accurately predicted, even if the order (tap or tap) of the adaptive filter is implemented less (for example, 3), the cancellation performance of the echo signal can be improved.

5 is a detailed block diagram of an echo canceller among speech coding / decoding apparatuses having echo cancellation according to a delay prediction according to the present invention.

As shown in FIG. 5, the echo canceller 332 according to the present invention uses the delay time Td received from the delay predictor 331 to output the output signal Vd (n) of the speech decoder 31. ) Is passed through a third-order linear digital filter to generate the echo signal component Ve (n). That is, after passing the output signal Vd (n) of the speech decoder 31 through the time delay unit, the linear digital The echo signal component Ve (n) is generated by multiplying the filter weights W (-1), W (0) and W (1), and the signal Ve (n) is applied to the microphone input signal. Output as a difference signal

At this time, the weights W (-1), W (0), and W (1) of the linear digital filter may be set to values that generate the least error as in the conventional adaptive digital filter, and reduce the calculation amount. To this end, weights may be obtained in advance when the pilot signal is generated. In addition, a value obtained using a conventional Least Mean Square (LMS) algorithm may be stored in a register and used.

As described above, the delay time predictor 331 according to the present invention drives the speaker 34 by adding a pilot signal P ₁ (n) to the output signal of the speech decoder 31 unlike in the related art. The pilot signal P ₂ (n) input from the microphone 35 is analyzed to accurately predict the delay time. As described above, the pilot signal used in the present invention uses an inaudible frequency far from the actual speech signal (eg, a sinusoidal signal of 3.5 KHz), and the delay time predictor 331 knows the frequency of this signal. Therefore, the band pass 41 can be used to easily extract the signal of the pilot component.

That is, in the related art, it is assumed that the distance between the speaker and the microphone (Ds meter) is in a constant section (for example, 0.5m), and the following Equation 1 is considered in consideration of the air velocity (Vs meter / sec) of the voice signal Delay time (Td) was calculated.

Td = Ds * Fs / Vs

That is, since the distance Ds between the speaker and the microphone has a constant value (for example, 0.5 meters), when the voice signal sampling frequency is 8000 samples / sec, Td = 0.5 * 8000/340 = 11 samples. Is calculated. However, in the Internet voice telephone using a personal computer (PC), the distance between the speaker and the microphone varies depending on the usage environment (for example, 0.1 to 10 meters).

Therefore, a large amount of computation may be necessary to accurately predict the delay time. In other words, considering the distance of 10 meters, we need to compare all the delay components for 10 * 8000/340 = 235 samples as shown in Equation 1 above. In addition, in the case of a wideband voice signal, the sampling frequency Fs is increased (eg, 16000 Hz), so that the complexity of the delay time calculation increases linearly (479 samples).

In contrast, in the present invention, in order to easily remove the real-time echo signal, the delay time prediction for the entire sample interval is performed only when the system is initially started (when the program is started or when the PC is booted). By using one delay time, delay time prediction is performed for a sample interval within a predetermined range (+.-).

The method of the present invention as described above may be implemented as a program and stored in a computer-readable recording medium (CD-ROM, RAM, ROM, floppy disk, hard disk, magneto-optical disk, etc.). Since this process can be easily implemented by those skilled in the art will not be described in more detail.

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

As described above, the present invention generates a pilot signal P ₁ (n) out of an audible signal to a voice output unit when outputting a voice signal in a voice telephone or the like, and then inputs a pilot signal P ₂ ( Since the echo signal is removed by accurately predicting the delay time from the time difference with n)), it is possible to provide an optimized echo signal cancellation function in real time even if the distance between the audio output unit and the audio input unit is not fixed.

1 is a block diagram of an embodiment of a voice codec in a conventional Internet voice telephone (VoIP) terminal.

2 is a block diagram of another embodiment of a voice codec in a conventional Internet voice telephone (VoIP) terminal.

3 is a diagram illustrating an embodiment of a speech coding / decoding apparatus having echo cancellation according to a delay time prediction according to the present invention.

4 is a detailed block diagram of an embodiment of a delay predictor among speech coding / decoding apparatuses having echo cancellation according to a delay prediction according to the present invention.

5 is a detailed block diagram of an echo canceller among speech coding / decoding apparatuses having echo cancellation according to a delay time prediction according to the present invention.

* Explanation of symbols on the main parts of the drawing

31: speech decoder 32: speech encoder

33: echo canceller 331: delay time predictor

332: echo canceller 34: speaker

35: microphone

Claims (8)

In the speech coding / decoding apparatus having an echo cancellation function according to the delay prediction, Speech decoding means for decoding the speech code received through the transmission channel and adding a pilot signal P ₁ (n) to output through the speech output means; The delay time is predicted by comparing the pilot signal P ₁ (n) received from the speech decoding means with the pilot signal P ₂ (n) extracted from the speech signal Vi (n) received through the speech input means. Echo canceling means for removing echo signals from the speech signal Vi (n) received through the speech input means in accordance with the estimated delay time; And Speech encoding means for receiving and encoding the speech signal V (n) from which the echo signal is removed by the echo removing means. Speech coding / decoding apparatus having an echo cancellation function according to the delay prediction including a. The method of claim 1, The echo cancellation means, The delay time is predicted by comparing the pilot signal P ₁ (n) from the speech decoding means with the pilot signal P ₂ (n) extracted from the speech signal Vi (n) received through the speech input means. Delay time prediction means for performing; And The echo signal component is generated by delaying the speech signal from the speech decoding means according to the delay time predicted by the delay time predicting means, and the speech signal received through the speech input means using the generated echo signal component ( Echo cancellation means for canceling the echo signal at Vi (n)) Speech coding / decoding apparatus having an echo cancellation function according to the delay prediction including a. The method of claim 2, The delay time prediction means, Pilot signal extracting means for extracting a pilot signal (P ₂ (n)) from the speech signal received through the speech input means; Sample delay means for delaying the pilot signal P ₁ (n) received from the speech decoding means in units of samples under control of a control means; Subtraction means for calculating a difference between the pilot signal P ₂ (n) extracted by the pilot signal extraction means and the pilot signal P ₁ (n) delayed by the sample delay means; Average absolute error value calculating means for calculating a mean absolute error value by receiving the difference signal E (n) from the subtraction means; Minimum value predicting means for extracting a delay time at which the average absolute error value received from said average absolute error value calculating means becomes minimum and transferring it to said echo cancellation means; And The control means for controlling a range of sample units in which the sample delay means operates Speech coding / decoding apparatus having an echo cancellation function according to the delay prediction including a. The method of claim 3, wherein The control means, The sample delay means controls the sample delay means to operate in the entire preset sample interval during system initial driving, and operates the sample delay within the predetermined range by reusing the delay time extracted by the minimum value predicting means in the actual voice transmission / reception process. Speech coding / decoding device having echo cancellation function according to delay time prediction, characterized in that for controlling. The method according to any one of claims 1 to 4, The pilot signal, Speech coding / decoding device having echo cancellation function according to delay time prediction, characterized in that the signal outside the audio frequency band. In the speech coding / decoding method having echo cancellation function according to delay prediction, A voice decoding step in which the voice decoder decodes the voice code received through the transmission channel, adds a pilot signal P ₁ (n) and outputs it through the voice output means; The delay time predictor compares the pilot signal P ₁ (n) received from the speech decoder with the pilot signal P ₂ (n) extracted from the speech signal Vi (n) received through the voice input means. A delay time predicting step of predicting a delay time; An echo cancellation step of removing an echo signal from the voice signal Vi (n) received through the voice input means according to the delay time predicted by the echo canceller in the delay time estimating step; And A speech encoding step in which a speech encoder receives and encodes a speech signal V (n) from which the echo signal is removed in the echo cancellation step. Speech coding / decoding method having an echo cancellation function according to the delay prediction comprising a. The method of claim 6, The echo cancellation step, The echo canceller generates a echo signal component by delaying the speech signal from the speech decoder according to the delay time predicted in the delay time predicting step, and inputs it through the speech input means using the generated echo signal component. A speech coding / decoding method having an echo cancellation function according to a delay time prediction, characterized in that the echo signal is removed from the received speech signal Vi (n). The method according to claim 6 or 7, The delay time prediction step, A pilot signal extracting step of extracting, by a band pass, a pilot signal (P ₂ (n)) from the voice signal received through the voice input means; A pilot signal delay step of N sample delays delaying the pilot signal P ₁ (n) received from the speech decoder in units of samples under the control of a controller; A subtraction step of calculating a difference between the pilot signal P ₂ (n) extracted by the subband and the pilot signal P ₁ (n) delayed by the N sample delayer; An average calculator calculating an average absolute error value by receiving a difference signal E (n) from the subtractor; And A minimum value estimating step of extracting a delay component whose minimum absolute error value calculated by the average calculator becomes the minimum and passing it to the echo canceller Speech coding / decoding method having an echo cancellation function according to the delay prediction comprising a.