KR20150112716A - Method and Apparatus for Automatic Measurement of Delay Time for Echo Path Prediction of Adaptive Filter in Set Top Box Video and Voice Call Service - Google Patents

Abstract

The present invention relates to a method and an apparatus for automatically measuring delay time for predicting a path of echo of an adaptive filter in a set-top box image and a voice call service. The method for automatically measuring delay time includes the steps of: generating a test signal with a single frequency on an input path for a received signal, which is input in a voice input and output device; encoding the received signal including the test signal; change a frequency of the encoded first signal; sensing the test signal in the second signal of which the frequency is changed; and calculating the minimum average of difference between the time of generating the test signal and the time of sensing the test signal. The minimum average is used to generate a filter coefficient of the adaptive filter.

Description

TECHNICAL FIELD The present invention relates to an automatic measurement method and apparatus for an echo path estimation of an adaptive filter in a set-top box video and voice communication service. }

Embodiments of the present invention relate to a delay time measurement technique for echo path prediction, and more particularly, to a method and apparatus for automatically measuring delay time for echo path prediction of an adaptive filter in a set- will be.

Speech echo refers to a phenomenon in which a voice signal output through a speaker is echoed and returned to the microphones. This causes a far-end speaker in the call service to delay his / her voice to hear again, thereby degrading the call quality. If the echo delay time is a short echo of less than 30 ms, it is not a serious problem. However, the echo cancellation causes remarkable deterioration of call quality and must be eliminated. For this purpose, acoustic echo cancellation technology has been continuously researched and developed.

1 is a schematic diagram of an echo cancellation model using an adaptive filter of the prior art.

As shown in FIG. 1, the basic principle of acoustic echo cancellation is to first estimate the characteristics of the echo path, generate an echo prediction signal therefrom, and remove it from the output of the microphone. Here, it is important to model the characteristics of the echo path as accurately as possible in order to generate a prediction signal similar to the actual echo signal. The characteristics of the echo path are usually modeled using a filter because it varies abruptly with time and surrounding conditions.

In FIG. 1, x (t) is a voice signal received from a communication partner to a wired / wireless network for a call, and z (t) is an estimated echo prediction signal modeled at time t. y (t) is an actual echo signal that has passed through the encoder according to time t. The final output signal e (t) at time t is a residual signal indicating the difference between the actual echo signal and the predicted signal predicted through modeling.

The adaptive algorithms used to minimize the final output signal e (t) include Least Mean Square (LMS) and Normalized Least Mean Square (NLMS). The LMS algorithm is an adaptive algorithm that minimizes the mean square error between the estimated object signal and the estimated signal, and the NLMS algorithm is a modified form of this LMS algorithm. The performance of the LMS algorithm depends on the size of the input signal. To solve this problem, a normalized Least Mean Square (NLMS) algorithm has been developed that normalizes the convergence constant with the power of the input signal.

An adaptive acoustic echo canceller for minimizing acoustic echo eliminates the influence of echo components by superimposing and creating an echo prediction signal equal to the echo component of the actual direction signal y (t) using an adaptive filter. The actual echo signal y (t) is a signal that has passed through the echo path according to the delay and is delayed and echoed by the influence of the delay of the audio output apparatus and the environment of the call environment.

In the adaptive filter, it is important to accurately predict the components of the echo path, and the filter coefficient is adjusted using the residual signal e (t) from which the echo component is removed. That is, the filter coefficient is adjusted by receiving the residual signal e (t) and the actual echo signal y (t).

Thus, the echo estimate signal z (t) of the adaptive filter is a matrix of N x 1 consisting of the adaptive filter coefficient and the actual echo signal y (t) of the actually echoed far-end signal. Where N is the length of the adaptive filter, and the longer the filter length, the greater the amount of computation, but the higher the accuracy.

The echo delay time is an important factor in estimating the echo prediction signal z (t), and is the response time of the echo cancellation. Conventional cell phones and telephone devices use built-in echo response time including device delay and initial response delay as a fixed value, and are used to cancel echo in a call, and the value is dependent only on the built-in sound device.

However, in a communication environment using a universal serial bus (USB) webcam, a television (TV), a computer speakerphone, etc., in which a sound device is not constant, it is difficult to estimate a delay time for each device, have. Particularly, when the position of the audio apparatus is changed, the previously used echo response time should be adjusted to a new value, but it is often difficult to predict or measure the delay time of the direction signal, have.

In this way, it is possible to accurately and easily predict the delay of the echo signal generated in the sound input / output device or the like by the actual echo signal of the sound apparatus in the video and / or voice call service of the acoustic apparatus in which the installation position is not constant, Is required.

Korean Registered Patent No. 10-0758206 (September 2006)

The present invention has been made to solve the problem that the voice of a far-end speaker is echoed back to a microphone through a speaker in a conventional video call, and an embodiment of the present invention is applied to an acoustic device We propose a method for efficiently setting or controlling echo cancellation adaptive filters in a video and voice call service using a set top box.

That is, a communication service using a television (TV) and a USB (Universal Serial Bus) webcam connected to a set top box HDMI (High Definition Multimedia Interface), a speaker connected to a computer, and a microphone, The adaptive filter function that performs echo cancellation due to the hardware device delay and the call environment is difficult to operate properly. Therefore, in the embodiment of the present invention, the hardware voice output delay and the minimum echo path delay time in the video call service are measured in advance and the echo path delay time measurement that can increase the accuracy and efficiency of the delay time prediction modeling of the echo canceling adaptive filter Method, and apparatus.

According to an aspect of the present invention, there is provided an automatic delay measurement method for echo path prediction of an adaptive filter for echo removal in a set-top box video and voice call service, A test signal of a single frequency is generated on the path, a received signal including a test signal is encoded, frequency-converted the encoded first signal, a test signal is detected in the frequency-converted second signal, And a minimum average value of the difference between the sensing times. Where the minimum mean value is used to generate the filter coefficients of the adaptive filter.

In one embodiment, the test signal is a single frequency within the speech band frequency. That is, the test signal is a single frequency within 0.3 to 3.4 kHz. Preferably, the test signal may be 1 kHz.

In one embodiment, an automatic delay time measurement method for echo path prediction of an adaptive filter further comprises determining validity of a minimum mean value. The validity range of the minimum average value is 30 ms or more and 500 ms or less.

According to another aspect of the present invention, there is provided an apparatus for automatically measuring delay time for echo path prediction of an adaptive filter, comprising: a test signal generator for generating a single frequency in a voice frequency band as a test signal; An audio encoding unit encoding an audio signal including a test signal; A frequency transformer for transforming the first signal in the time domain encoded by the audio encoder into a second signal in the frequency domain; A signal detecting unit for detecting a test signal which is the maximum signal on the frequency in the second signal; And a delay time measuring unit for measuring a time difference between the generation time of the test signal and the detection time in the signal detection unit and calculating a minimum average value of the time difference. Here, the minimum average value is used for generating the filter coefficient of the adaptive filter for echo cancellation in the set-top box video and voice call service.

In one embodiment, the test signal is a single frequency within 0.3 to 3.4 kHz.

In one embodiment, the delay time measurer determines the validity of the minimum mean value. Here, the validity range of the minimum average value is 30 ms or more and 500 ms or less.

According to the present invention, accuracy and efficiency of delay time prediction modeling of the echo cancellation adaptive filter can be increased by previously measuring the hardware speech output delay and the minimum echo path delay time in the video call service.

That is, the sound input device and the sound output device are fixed as in a call service using a television (TV) and a USB (Universal Serial Bus) webcam connected to a set-top box HDMI (High Definition Multimedia Interface) A method and an apparatus for measuring echo path delay time which can reliably and automatically measure the delay time due to hardware device delay and influence of a communication environment even when the echo path does not have a shape .

1 is a schematic diagram of an echo cancellation model using an adaptive filter of a conventional acoustic apparatus;
FIG. 2 is an exemplary diagram for explaining a pre-acoustic apparatus delay time measurement process of an automatic delay time measuring apparatus (hereinafter, an echo path delay time measuring apparatus) for echo path prediction of an adaptive filter according to an embodiment of the present invention.
3 is a block diagram of an echo path delay time measuring apparatus of FIG.
4 is a flowchart illustrating an automatic delay time measurement method for echo path prediction of an adaptive filter according to another embodiment of the present invention.

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

Portable terminals and general fixed line telephones are fixed type such as a speaker and a microphone (MIC). In this case, a hardware delay value such as voice input / output necessary for a certain level of call quality or the like is generally used in a form stored in a memory in a manufacturing step.

However, in a call service that connects a universal serial bus (USB) webcam and a high-definition multimedia interface (HDMI) connected television (TV) device, sound devices are not fixed, The majority of cases are large. In this case, when the path delay value is large, the adaptive filter necessary for the call quality is increased in the amount of calculation, and thus it is difficult to exert its performance due to filter distortion. For this reason, the audio output device and the input device for the conventional call service are usually fixedly installed, and it becomes difficult to change it.

Therefore, in the embodiment of the present invention, a method of preliminarily measuring a minimum echo path including a delay of a sound device in a communication device is provided to estimate an echo signal by modeling an initial adaptive filter, The device and the input device can be changed and used.

In the following description, the contents related to the initial echo signal estimation are first described through echo signal modeling in echo cancellation, and the echo path is modeled using the echo path check signals shown in the embodiments of the present invention. An echo canceller that does not depend on a sound device will be described.

Referring to FIG. 1, a signal received in a general acoustic device is denoted by x (t), an impulse response of an echo path estimated at time t (hereinafter referred to as a first filter coefficient) [h , And the second filter coefficient of the actual echo path is ci, the echo replica signal z (t) and the actual echo signal y (t) are given by the following equations (1) and (2).

According to Equations (1) and (2), the final output signal e (t) to be minimized in the present embodiment is expressed by the following Equation (3).

Equation 3 is an algorithm that minimizes the final output signal e (t). The filter coefficient is obtained by the LMS (Least Mean Square) algorithm.

In Equation (4), μ is a constant that determines the convergence speed stability.

If the estimated echo signal z (t) of Equation (1) is rewritten by applying the NLMS (Normalized Least Mean) algorithm to the value obtained in Equation (4), it can be expressed as Equation (5).

In estimating the adaptive essential coefficients of the above two methods, the shock response time h _i (t) of the echo path, that is, the delay response time due to the actual echo path, is assumed to be a negligible distance between the speaker and the microphone , It is most affected by the delay time of the audio output device. In other words, when the response time becomes longer, the length of the filter must be increased, whereby not only the amount of calculation but also the filter convergence time is increased.

For example, when a mobile phone or a public switched telephone network (PSTN) telephone has a delay time of 20 to 30 ms or less, the influence of the echo cancellation response time is considerably large, such as device delay in echo signal estimation modeling not. However, since the delay of the audio output device of the HDMI-connected TV is about 80 to 300 milliseconds for the HDMI delay and the TV sound output, there is a problem that the delay values are different for each device. Therefore, the audio output apparatus of the TV connected with HDMI should cancel the difference of the delay time in the impulse response time of adaptive filter coefficient prediction and echo cancellation estimation modeling.

2 is an exemplary diagram illustrating a video and voice call service system employing a delay time measurement apparatus for echo path prediction of an adaptive filter according to an embodiment of the present invention.

Referring to FIG. 2, the video and voice call service system equipped with the echo path delay time measuring apparatus according to the present embodiment includes a near-end speaker and a tail talker, and is adaptive to at least one of a near- A filter 21, a filter coefficient generator 20, and a delay time measuring device 25.

The far-end speaker in the video and audio call service system includes a sound device connected to the TV through the HDMI of the set-top box and a delay time measuring device 25, and a near- (20) and an adaptive filter (21). Here, the near-end speaker can be implemented as a TV, a computer device, a portable terminal, or the like.

The adaptive filter 21 filters the audio signal input on the communication channel through the near-end speaker microphone 14. The filter coefficient generator 20 supplies a filter coefficient (Filter Coef.) To the adaptive filter 21. The adaptive filter 21 and the filter coefficient generation unit 20 may be mounted on a set-top box such as a digital television to which a near-end speaker's sound apparatus or a sound apparatus is connected.

The delay time measuring device 25 is an echo path delay time measuring device that measures the minimum delay (23) of the echo path leading from the far-end speaker 22 to the microphone 24. [ The delay time measuring device 25 obtains filter coefficient information (an average value of the final output signal or the like) that minimizes the echo path delay time and provides it to the filter coefficient generator 20 of the near-end speaker.

The delay time measuring device 25 measures the minimum delay time of the echo path using a single frequency signal. A single frequency signal uses a sinusoidal 1 kHz single frequency signal as a test signal. That is, since the test signal measures the delay time in a call situation, a single frequency within the speech band frequency 0.3 to 3.4 kHz can be used for the composition of the situation closest to the actual call.

The reason why the single frequency is used as the test signal in the present embodiment is that the error caused by the actual calculation delay can be minimized by minimizing the calculation delay by encoding and FFT (Fast Fourier Transform) conversion of the signal input into the MIC to be.

The minimum delay measurement time for the echo path is denoted by Te, and it can be expressed by the following equation (6).

In Equation (6), Tc is the time at which the test signal is echoed into the microphone, and Tp is the test tone generation / output time.

Since the echo path delay measurement time (Te) increases proportionally with the distance between the actual speaker and the microphone, when measuring the minimum delay measurement time, the speaker and microphone can be positioned as close as possible to minimize the measured value.

3 is a block diagram of an echo path delay time measurement apparatus coupled to the acoustic apparatus of the far end speaker of FIG.

3, the echo path delay time measuring apparatus according to the present embodiment includes a test tone generating unit 31, a signal input unit 32, an audio encoding unit 33, an FFT transforming unit 34, (35) and a delay time measuring unit (36).

The echo path delay time measuring device corresponds to the delay time measuring device 25 of the far-end speaker shown in Fig. The echo path delay time measuring apparatus can be included in the apparatus of the near end speaker according to its relative role similar to the sender and the receiver of the video and voice call service or can be mounted on both the near end speaker and the far end speaker.

More specifically, the test tone generating unit 31 generates a test tone signal having a single frequency (for example, 1 kHz) in the speech frequency band as a test signal at a predetermined period (for example, 20 ms units). The test tone generator 31 may be implemented as a local oscillator using a quartz oscillator as a test signal generator.

The signal input unit 32 refers to a predetermined function unit of the audio input device to which the echo signal of the test tone is input by the test tone generation unit 31 or a configuration unit that performs a function corresponding to the function unit. For example, the signal input unit 32 may be implemented by including a signal input terminal or an analog digital converter connected to the signal input terminal and the signal input terminal.

The audio encoding unit 33 digitally codes the test tone audio input signal. The audio encoding unit 33 can encode the echo signal including the test tone and encode it at a low bit rate.

The FFT transforming unit 34 performs frequency FFT on the digital input signal. That is, the FFT transforming unit 34 transforms a signal in the time domain into a signal in the frequency domain as a frequency domain transformer. In this embodiment, the FFT conversion unit 34 optimized as the frequency conversion unit for converting the signal in the time domain into the signal in the frequency domain is used for the sake of convenience, but the present invention is not limited thereto.

The signal detecting unit 35 detects that the maximum signal on the frequency of the signal output from the FFT transforming unit 34 is a 1 kHz test signal.

The delay time measuring unit 36 measures a test tone generation time and a time difference in which the test tone is sensed by the signal sensing unit 35, and calculates a minimum average value.

In the present embodiment, a 1 kHz signal obtained by sampling the 8 kHz signal in the test tone generating section 31 is output to the actual voice output apparatus without delay in order to make the environment as similar to the actual voice call as possible, And the test signal is outputted from the voice output device by resampling (re-sampling). By doing so, it is possible to accurately predict the delay time of the echo signal including the device delay in the actual call environment.

In the audio encoding unit 33 of the present embodiment, the audio signal encoding rate is digitally encoded at 48 kHz to measure the echo path delay time. This is because the signal-to-noise ratio is higher than that of the 8-kHz encoding, and the signal detection unit 35 can increase the test tone detection efficiency.

In addition, the phase signal passed through the audio encoding unit 33 is frequency-spectrum-converted by the FFT transforming unit 34. The signal detection unit 35 determines whether the frequency of the signal having the highest intensity is found on the frequency spectrum and whether the frequency of the signal matches the test tone 1 kHz.

If the frequency of the FFT-transformed signal coincides with the test tone, the delay minimum value for the echo path prediction including the acoustic device delay is measured by the delay time measuring unit 36.

In order to reduce the measurement error of the delay time, the average value can be obtained by discarding the minimum range 10% and the maximum range 10% in the measurement range of the N measurement results. In this case, the range of the minimum average value validity may be set to 30 ms or more and 500 ms or less. The reason for this is that the time is too short for devices using HDMI and TV in case of less than 30 ms, and it is not valid if it exceeds 500 ms. Because.

The measured echo path delay minimum value Te can be used as the minimum delay time in the echo path estimation signal, thereby improving the response time and the filter coefficient accuracy of the adaptive filter echo canceller, as well as speeding up the convergence of the filter.

The echo path remover is designed with a limited filter length (taps), assuming that the reverberation time of the sound is typically 100 to 400 ms in an office or living room that is not particularly soundproofed. The adaptive filter is difficult to converge and the basic operation of the echo canceller becomes difficult if the device delay is not used for the adaptive filter estimation. Therefore, in this embodiment, by automatically measuring the echo path delay time and controlling the filter coefficient of the adaptive filter, it is useful and effective when the audio device delay is large. And is particularly suitable for a TV using an HDMI with a large device delay, but the present invention is not limited thereto.

4 is a flowchart illustrating an automatic delay time measurement method for echo path prediction of an adaptive filter according to another embodiment of the present invention.

Referring to FIG. 4, an automatic delay time measurement method (hereinafter referred to as an echo path delay time measurement method) for echo path prediction of an adaptive filter according to an embodiment of the present invention includes a system A microphone for communication or a television (TV) is connected to the main body (S41). The system body may include a set-top box or the like. The sound apparatus includes a speaker and a microphone (or a microphone), and the system body includes a set-top box or a computer apparatus in which a television, a USB webcam, or the like is connected by HDMI.

A set-top box or a computer device includes a memory for storing a program and a processor for executing a program stored in the memory, and the processor executes a series of the present embodiment for implementing an automatic delay time measurement method for estimating an echo path of an adaptive filter by a program Or a component that performs a function corresponding to this means.

When a microphone for communication or a TV is connected, the system body generates a single frequency sample signal through the test tone generator of the sound device and records the single frequency sample signal generation time Tp (S42). Here, the sample signal corresponds to a test signal.

Next, a microphone for communication or a TV is connected (S43). A test tone echo signal, which is a single frequency sample signal, is input to an audio input device through a signal input unit connected to a microphone or a TV, and an input sample signal and an echo signal are encoded.

Next, the audio signal encoded through Fast Fourier Transform (FFT) is converted into a frequency signal (S44), and a sample signal is detected from the maximum signal of the converted frequency signal (S45).

When the sample signal is detected from the maximum signal of the frequency signal, the average value E (n) obtained by repeatedly measuring the minimum delay measurement time Te of the echo path a plurality of times (N) is calculated (S46, S47). Here, a plurality of times (N) may be ten.

In order to reduce the measurement error of the delay time, the minimum value 10% and the maximum value 10% of the measurement value are discarded from the generation of the sample signal 10 times and the detection result of the sample signal (S48).

Next, it is determined whether the average value obtained in the above step (48) is valid (S49). The validity range of the average value is set to 30 ms or less and 500 ms or more. The reason for this is that the time is too short for devices using HDMI and TV in case of 30 ms or less, and it is not valid, and since the delay of 500 ms or longer can not make a call, it is determined that the value is invalid.

As a result of the determination step S49, if the average value is valid, the echo path delay is set based on the echo path delay time measurement result obtained as the effective average value (S50).

On the other hand, if it is determined that the average value is not valid as a result of the determination step S49, the process is terminated without setting the echo path delay as the setting failure S51. Of course, after the setting failure, the method of the present embodiment can be implemented to perform again the above-described process to automatically measure again the echo path delay time within a predetermined number of times.

According to the present embodiment, in order to prevent the degradation of the communication quality due to the use of the input / output device due to the use of any audio such as a TV and a webcam, an echo path minimum delay required for echo path modeling of an adaptive filter It is possible to provide a method of measuring the time in advance. In particular, by providing a method that can be used whenever a TV and a webcam audio input / output device are changed by using a single frequency sine wave signal, the filter coefficients of the particular TV and webcam are not limited to adaptive filters that are fixed in advance , The automatic variable filter coefficient enables users to receive high-quality video calls regardless of the position of a speaker or a microphone, thereby maximizing user convenience as well as performance of a sound apparatus.

In addition, in order to secure the performance of the echo canceller, which is essential for the call quality, conventionally, only a dedicated audio input / output device built in the product itself for communication has been used and it has been difficult to change it. However, It is possible to measure the minimum value of the echo path by measuring the delay of the direct audio input / output device, thereby making it possible to efficiently use the adaptive filter for eliminating the echo signal.

According to the present embodiment, it is possible to apply the present invention to an apparatus equipped with a call service, but in particular to an IPTV set-top box in which an audio input / output device is not fixed and an echo path delay A time measuring apparatus and method can be provided.

In addition, TV audio output apparatuses connected through HDMI and S / PDIF (Sony / Philips Digital Interface) output audio, and since the device delay is particularly large, an echo path An apparatus and a method for measuring a delay time can be provided.

In addition, by using the apparatus and method for measuring echo path delay time according to the present embodiment, when it is desired to change the arrangement position of the acoustic input device and / or the acoustic output device in the call service, The position of the device and / or the sound output device can be changed.

It will be understood by those of ordinary skill in the art that various changes in form and details may be made therein without departing from the spirit and scope of the present invention as defined by the following claims . It is therefore to be understood that the above-described embodiments are illustrative in all aspects and not restrictive.

25: Delay time measuring device
31: Test tone generator
32: Signal input section
33: Audio coding unit
34: FFT transform unit
35: Signal detection unit
36: delay time measuring unit

Claims (8)

A method for automatic measurement of delay time for echo path prediction of an adaptive filter for echo cancellation in a set top box video and voice call service,
A test signal of a single frequency is generated on the input path of the received signal to the acoustic input / output device,
A reception signal including the test signal is encoded,
Frequency-converts the encoded first signal,
Detecting the test signal in the frequency-converted second signal,
Calculating a minimum average value of the difference between the generation time and the detection time of the test signal,
And using the minimum mean value to generate a filter coefficient of the adaptive filter. The method according to claim 1,
Wherein the test signal is a single frequency within a speech band frequency. The method according to claim 1,
Wherein the test signal is a single frequency within 0.3 to 3.4 kHz. The method according to claim 1,
Further comprising determining the validity of the minimum mean value. ≪ Desc / Clms Page number 20 > The method of claim 4,
Wherein the effective range of the minimum average value is 30 ms or more and 500 ms or less. A test signal generator for generating a single frequency within a voice frequency band as a test signal;
An audio encoding unit encoding an audio signal including the test signal;
A frequency transformer for transforming the first signal in the time domain encoded by the audio encoder into a second signal in the frequency domain;
A signal detecting unit for detecting the test signal which is the maximum signal on the frequency in the second signal; And
A delay time measuring unit measuring a time difference between the generation time of the test signal and the detection time of the signal sensing unit and calculating a minimum average value of the time difference;
, ≪ / RTI &
Wherein the minimum average value is used for generating a filter coefficient of an adaptive filter for echo cancellation in a set-top box video and voice call service. The method of claim 6,
Wherein the test signal is a single frequency within 0.3 to 3.4 kHz. The method of claim 6,
Wherein the delay time measuring unit determines the validity of the minimum average value and the effective range of the minimum average value is not less than 30 ms and not more than 500 ms.

Images