KR100238470B1 - Multiple Speaker Selection Apparatus and Method for Video Conference System - Google Patents

Abstract

1. TECHNICAL FIELD OF THE INVENTION

A circuit and method for selecting multiple speakers for audio mixing in a videoconferencing system.

2. The technical problem to be solved by the invention

An apparatus and method for accurately selecting a plurality of speakers in real time is provided.

3. Summary of Solution to Invention

It is characterized by counting the number of times that the sign of each voice signal remains the same and the number of times the magnitude exceeds the reference value for a certain period of time, and selecting the speaker by comparing the sign / size count value at the point of speaker selection.

4. Important uses of the invention

It is used to accurately select the actual speaker's voice by using the characteristics of the voice signal.

Description

Multiple Speaker Selection Apparatus and Method for Video Conference System

1 is an exemplary configuration diagram of a video conference system according to the present invention.

2 is a configuration diagram of a video conference control device in FIG.

3 is a detailed configuration diagram of an audio mixing unit in FIG. 1.

4 is a flowchart illustrating a comparison process for selecting a plurality of speakers according to the present invention.

5 is a flowchart illustrating a process of selecting a speaker at the time of speaker selection based on the comparison result of FIG.

6A is a graph showing characteristics of a voice signal.

FIG. 6B is a waveform diagram of an audio signal for explaining the case where only one speaker is selected according to the comparison of magnitudes.

6C is a sound signal waveform diagram of a first terminal for explaining a case where a speaker is selected according to a comparison of a sign and a magnitude.

6d is a sound signal waveform diagram of a second terminal for explaining the case where a speaker is selected according to a comparison of a sign and a magnitude.

* Explanation of symbols for main parts of the drawings

2a: ISDN BRI interface 2c: H.221 framing

2b: ISDN PRI interface part 2d: data separation extracting part

2e: Audio Mixer 2f: Video Switching Unit

2g: data processor 2h: controller

2i: multiplexer 2j: H. 221 reframing unit

10: decoder 12: code check unit

14: comparison unit 16: count unit

18: Speaker selector 20: Encoder

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to circuits and methods for selecting a plurality of speakers for audio mixing in a videoconferencing system, and more particularly to circuits and methods for selecting a speaker's voice using characteristics of the voice signal.

Common forms of video conferencing include point-to-point video conferencing and multi-point video conferencing in a group.

Referring to the video conferencing having the H.221 frame structure in the H series of the international standard ITU-T, for example, FIG. 1 shows an example of the configuration of the video conferencing system according to the present invention. The video conferencing control device 100 controls the plurality of terminals T1 to Tn to participate in the video conference at the same time through the ISDN.

FIG. 2 is a detailed configuration diagram of the video conferencing control device 100 of FIG. 1, wherein a data line connecting each part is MVIP (Multi-Vendor Integrated Protocol) and the detailed operation of each part is as follows.

The line interface units 2a and 2b perform a signal processing procedure according to ITU-T Q.931 to connect the videoconferencing control device to the ISDN network, and may be configured only with either BRI 2a or PRI 2b. The H. 221 framing unit 2c is a module that searches for and establishes the frame structure specified in H. 221 and synchronizes a frame alignment signal (FAS) and a bit rate allocation signal (BAS). The demultiplexer 2d extracts the image data composed of the synchronized sound, image, and data into respective independent data and delivers them on the bus. The audio mixing unit 2e includes an energy measurement unit and a mixing unit, and extracts terminals having a large acoustic energy from the acoustic data input from the demultiplexer 2d and mixes them. The video switching unit 2f extracts and supplies an image of a desired terminal from the image of each terminal input from the demultiplexing unit 2d. The data processor 2g processes the data input from the demultiplexer 2d. The control unit 2h controls the input / output according to the control code inputted from the H. 221 framing unit 2c, and the CCITT Recommendation H. 231 / H. 242 / H. 243 is a module that collectively controls the overall operation of the videoconferencing controller. The data combiner 2i is a multiplexing module for recombining the data processed from the modules 2e, 2f, and 2g into one data structure. The H. 221 reframing unit 2j converts the data input from the data combination unit 2i into data having an H. 221 frame structure. The converted data is then converted into the line interface unit 2a or 2b. Is transmitted through the ISDN network.

The videoconferencing control device 100 having such a configuration differs in its control method depending on whether the videoconferencing control is set to the audio mode or the design mode. In the former case, the speaker is searched for according to the size of the audio signal and the voice is mixed. In the latter case, since the speaker is designated by the chairman terminal, a separate speaker search operation for audio processing is not required.

On the other hand, there are two conventional methods of selecting one speaker to deliver the current speaker's voice signal to the conference participants.

First, the loudest audio signal is first determined as the speaker by comparing the magnitudes of the voice signals of the participants participating in the conference, and if the value is larger than the predetermined value, the true speaker is determined. FIG. 6B shows the waveform of the audio signal for explaining such a case.

When the method of simply comparing the sizes of voice signals is applied to the selection of a plurality of speakers, the speaker must be selected in the order of the size of the voice at each input. Therefore, even if noise occurs and is recognized as voice, there is no way to prevent it. As a result, when selecting the main speaker and the remaining incubators at the meeting, there is a risk that the coin will change every time it is entered.

Secondly, the speaker was determined by obtaining auto-correlation coefficients of the signals without comparing the magnitudes of the voice signals for a certain sampling period and comparing the magnitudes of the coefficients.

In the method of obtaining autocorrelation coefficients and comparing the magnitudes of the voice signals for a certain period of time, the calculation amount increases at the time of calculating the autocorrelation coefficients. There is a problem, and if the period is shortened to solve it, the same error as the first method described above occurs.

In conclusion, the method of comparing the magnitude of the voice signal or comparing the autocorrelation number for a certain period of time is not suitable for the video conferencing system requiring the selection of three or more speakers.

Accordingly, an object of the present invention is to provide an apparatus and method for accurately selecting a plurality of speakers in real time.

In order to achieve the above object, the present invention counts the number of times that the sign of each voice signal remains the same and the number of times the magnitude exceeds the reference value for a predetermined period, and compares the sign / size count value at the time of speaker selection. Characterized by selecting a speaker.

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

First, in adding reference numerals to the components of each drawing, it should be noted that the same reference numerals have the same reference numerals as much as possible even if displayed on different drawings. In addition, in the following description, many specific details such as components of specific circuits are shown, which are provided to help a more general understanding of the present invention, and the present invention may be practiced without these specific details. It will be obvious to those skilled in the art. In the following description of the present invention, if it is determined that a detailed description of a related known function or configuration may unnecessarily obscure the subject matter of the present invention, the detailed description thereof will be omitted.

The audio signal processing portion of the videoconferencing system according to the present embodiment includes an A / μ-1aw PCM data decoder, a speaker selection portion, and an A / μ-1aw PCM data encoder as shown in FIG. And the input signal is the PCM signal of the G.711 Recommendation of ITU-T.

In the case of G.711, the period of the input signal is 125 s in consideration of 80 samples per frame, and the operation of the present embodiment will be described based on the speaker selection every 10 seconds.

As shown in FIG. 6A, the audio signal has a graphical characteristic in the form of oscillating the + value and the − value region. In this embodiment, this characteristic of the voice signal is used.

Referring to FIG. 4, in step 4a, the variable is initialized, and in step 4b, the voice signal of the participant is input from the terminal participating in the conference. In step 4c, whether the sign of the input voice signal is a positive value or a negative value is checked, and the result of the check is stored as a binary value 0 or 1 in steps 4d and 4e. Thereafter, when the binary value is larger than the reference value in step 4f, the size count is increased in step 4g. In step 4h, the current count is compared with the previously stored code, and if it is not the same value, the code count is increased in step 4i. Repeat this action for each participant.

The repeated process is continued until the speaker selection point is reached. When the speaker selection point is reached, as shown in FIG. 5, the variable is initialized in step 5a and proceeds to step 5b to determine the sign / size count of each terminal. In step 5c, first, three speakers are selected in large order by comparing the size of sign counts. If there are two or more terminals having the same size of the code count as shown in step 5c, the process proceeds to step 5e and selects the terminal having the largest size count as the speaker. At this time, if the size count is less than the reference value, the voice signal of the previously selected speaker is maintained as the speaker. In step 5e, if the terminal having the same size of the code count is not 2 or more, the process proceeds to step 5f to determine the terminal having the maximum code count as the speaker.

When the above method is described further with reference to the waveform of the audio signal,

6B is an audio signal waveform diagram for explaining the case where only one speaker is selected according to the size comparison. This represents a conventional method. According to this method, instantaneous loud voice or ambient noise is recognized as a speaker. In other words, if you actually speak, the waveform will show a continuous shape, but if you do not speak and the loud noise is made momentarily due to noise or other environmental reasons, the waveform will change dramatically. It is mistaken.

FIG. 6C is a waveform diagram of a voice signal of a first terminal for explaining a case of selecting a speaker according to a comparison of a sign and a size, and FIG. 6D is a second diagram for explaining a case of selecting a speaker according to a comparison of a sign and a size. Sound signal waveform diagram of the terminal. According to the two waveform diagrams, assuming that T1 is a speaker selection point, the code counts corresponding to the waveforms of the 6c and 6d diagrams at the time T1 are all four. However, the size count is 3 for FIG. 6C and 4 for FIG. 6D. Therefore, the terminal showing the waveform of FIG. 6d is selected as the speaker.

On the other hand, in the present embodiment, in the conference system, not only one speaker selection but also a plurality of speaker selection of about three people can be processed in software in real time.

In conclusion, according to the present embodiment, two speaker selection situations may occur.

First, when a conference system delivers a speaker's voice to the other conference participants, it is necessary to select the person who is actually speaking during the conference. In this case, you can see if the current participant is speaking by checking the sign count. If there are several people speaking, the size count can be checked to select the person with the larger level.

Second, if many participants need to mix the voice signal of the current speaker and send it to the listeners, they can check the number of signs and select as many as necessary.

As described above, the present invention has advantages in terms of boundary and processing time because it can reduce a large amount of computation and memory required in the method using the autocorrelation coefficient by using the characteristics of the audio signal. It can be applied without modification to the selection of three or one speakers required for audio mixing and voice activated mode. In addition, in the conference system to which G.711 (PCM), G.722 (ADPCM), and G.728 (LD-CELP) of ITU-T are applied, the speaker selection module can be implemented simply by changing the reference value. .

Meanwhile, in the detailed description of the present invention, specific embodiments have been described, but various modifications may be made without departing from the scope of the present invention. Therefore, the scope of the present invention should not be limited to the described embodiments, but should be defined not only by the scope of the following claims, but also by the equivalents of the claims.

Claims (3)

A plurality of terminals are controlled to simultaneously participate in the video conference, and each terminal participating in the video conference has a schedule for mixing audio of the terminals participating in the video conference in the conference control apparatus having a code and a size counter for the audio signal. A method of selecting a speaker by a period, the method comprising: checking the arrival of a speaker selection point, checking a sign count for each terminal participating in a video conference until the speaker selection point, and selecting a terminal having a maximum value; And selecting a terminal having a maximum value by checking a size count if the code counts for two or more terminals are the same as a result of the check of the code count. The method of claim 1, wherein the performing of the code and the size count for each terminal comprises: inputting a voice signal from a terminal participating in a video conference, checking a code of the input voice signal, Storing a previous code, converting the input voice signal into a pulse code modulated signal, and comparing the magnitude of the pulse code modulated signal with a predetermined reference value to increase the size count by a unit value if the reference value is larger than the reference value. And checking whether the current code is the same as a previously stored code and increasing the code count by a unit value if the same is the same. A video conferencing system having a single terminal and a conference control apparatus for performing video processing and audio mixing such that the plurality of terminals participate in a video conference simultaneously, wherein the conference control apparatus is any one that participates in a video conference. A voice input unit for inputting a voice signal from the terminal, a signal converter for converting the input voice signal into a pulse code modulated signal, a code count, a size counter, and terminals to participate in a video conference are registered and a predetermined binary code is input. Check the memory for storing and the code of the voice signal input from the registered terminal to give a binary code according to the positive or negative, and check whether the binary code currently stored in the memory is the same as the previously stored binary code If it is the same, the sign counter is increased by a unit value, and the magnitude of the pulse code modulation signal is compared with a predetermined reference value. The counter control unit increases the size counter by a unit value if the reference value is greater than the reference value, and checks the sign count for each terminal participating in the video conference until a certain period, and selects the terminal as the speaker if there is one terminal having the maximum value. And a speaker selector configured to select a terminal having a maximum value as a speaker by checking the size counts of the corresponding terminals when the code counts for the two or more terminals are the same.