KR20040107068A - Hands-free, voice-operated remote control system and method for audio sound generating devices - Google Patents

Abstract

PURPOSE: A non-contact remote control system and method based on a voice of a sound generator are provided to increase voice recognition performance even in a noisy environment by raising a signal-to-noise ratio irrespective of a voice-generating position of a speaker, to thereby improve reliability of remote control. CONSTITUTION: A non-contact remote control system includes a sound generator(110) for generating and amplifying an electric signal for generating a sound, a speaker(111) for receiving the amplified electric signal and changing the vibration of the signal to output a sound signal, and a microphone(130) for receiving the sound signal of the speaker and a voice command for remotely controlling the sound generator. The system further includes an adaptive control filter(140) for compensating transfer characteristic of the speaker or microphone and transfer characteristic according to various transfer paths of the sound signal to remove a noise signal input to the microphone, a voice recognizing unit(150) for recognizing a voice from which the noise signal has been removed by the adaptive control filter as a command, and a sound generator controller(160) for receiving the voice recognition result from the speech recognizing unit to control the sound generator.

Description

Non-contact remote control system and method by voice of sound generating equipment {HANDS-FREE, VOICE-OPERATED REMOTE CONTROL SYSTEM AND METHOD FOR AUDIO SOUND GENERATING DEVICES}

The present invention relates to a non-contact remote control system and method by the sound of the sound generating device, and more particularly, using an electrical signal of the sound generating device for driving the speaker as a reference signal, independent element analysis (ICA) algorithm or minimum By applying the mean square (LMS) algorithm to remove the noise input to the microphone, it is possible to perform remote control by voice of the sound generating device in a non-contact way by increasing the recognition performance of the voice in all other environments such as a noisy environment. It relates to a non-contact remote control system and method by the sound of the sound generator.

Most TVs, video cassette recorders (VCRs), satellite receivers, audio, and digital versatile discs (DVDs) provide remote control devices that allow users to remotely control without having to get close to the equipment they want to control. . The remote control device is generally small enough to be held in hand, and when the user presses a specific button on the keypad attached to the remote control, a command signal for driving the function is transmitted and controlled in a wired / wireless form from the remote control. The target device receives this signal and interprets it to operate the channel selection, power on / off, and volume adjustment commands.

This type of remote control has some disadvantages, one of which is to find the remote control first. Because of their small size, they may be hidden by other objects, embedded in gaps, invisible, and sometimes lost. In addition, since different remote controls are used depending on the devices, when there are several devices at the same time, it is troublesome to find a desired one among the remote controls.

In the case of a user who has an uncomfortable operation or a vision problem, such a hassle may act as a great inconvenience or a disability, and a general user may also be a cause of discomfort. In addition, when the user is doing other work by using the hand, in order to use the remote control, the work must be stopped, thereby reducing efficiency or inconvenience.

In order to overcome this disadvantage of the remote control requiring physical contact, remote control schemes have recently been developed to operate by a voice command of a user, US Patent No. 5,199,080; 5,247,580; 5,247,580; 5,267,323; 5,267,323; 5,226,090 and the like are examples. Unfortunately, these methods are still not seen as true contactless remote controls in that they require minimal physical contact, and these methods are called "talk" before giving a voice command to tell the device that a user's voice command is being issued. You need to press the "talk switch".

In addition, U.S. Patent No. 5,852,804 uses two pieces of microphones equidistantly spaced from a speaker, and uses the information such as the delay time difference between two signals coming into the two microphones after A / D conversion of the signals from each microphone. Through the signal processing process to extract only the signal from the sound source located at the same distance from the microphone is put into the voice recognizer. Instead of using the "torque switch", this method utilizes special location information between two microphones and the speaker to reduce the ambient noise, thereby increasing the discrimination between the voice and the ambient noise from the speaker.

However, this method has a disadvantage that the degree of improved signal-to-noise ratio (S / N) is not so large as about 3 dB and can be applied only when the speaker's voice position is fixed as in a vehicle.

Therefore, for a system driven only by the speaker's voice in a completely non-contact manner, it is essential to reduce unnecessary noise by reducing the ambient noise input to the microphone to a level that can be clearly distinguished from the input voice.

The present invention is to solve the problems and suggestions made as described above is to remove the noise component input to the microphone by using the electrical signal of the acoustic generator for driving the speaker as a reference signal By improving the signal-to-noise ratio without greatly limiting the speaker's vocalization position, it is possible to improve the voice recognition performance in all other environments such as a noisy environment, thereby improving the reliability of the remote control. It is to provide a non-contact remote control system and method by the voice of the sound generating device to enable the remote control by the voice of the sound generating device in a truly non-contact manner.

Figure 1 is a block diagram showing a non-contact remote control system by the voice of the sound generating apparatus according to the present invention.

Figure 2 is a block diagram showing a home theater environment as an example for explaining the environment to which the present invention can be applied.

3 is a configuration diagram illustrating a microphone input relationship between acoustic radiation of a sound apparatus and radiated sound;

Figure 4 is a block diagram showing a system structure of adaptive noise removal to explain the present invention.

Figure 5 is a block diagram showing a voice driving method by the noise removal in the present invention.

6 is a waveform diagram showing an example of a microphone output signal according to the conventional method.

7 is a waveform diagram illustrating a state in which the microphone output signal of FIG. 6 is removed by the present invention.

Explanation of symbols on the main parts of the drawings

110: sound generator 120: speaker

130: microphone 140: adaptive control filter

150: voice recognizer 160: sound generator controller

The apparatus of the present invention for achieving the above object is in the remote control system by the voice of the sound generating device, including a sound generating device, a speaker, a microphone, a voice recognizer; By using the electrical signal generated by the sound generating device as a reference signal, the transmission characteristics of the speaker and the microphone and the transmission characteristics by various transmission paths of the acoustic signal are compensated for, and thereby the noise signal input to the microphone is compensated for. It is a basic feature of the technical configuration to further comprise an adaptive control filter to be removed and delivered to the voice recognizer.

In addition, the method of the present invention, in the remote control method by the voice of the sound generating device, adapted to remove the sound signal component by the reference signal from the noise input to the microphone using the electrical signal of the sound generating device as a reference signal The technical feature of the present invention is to apply a learning algorithm so that a user's voice command is input to the voice recognizer in a noise-free form so that it can be controlled by voice even during operation of the sound generator.

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

1 is a block diagram showing a non-contact raw economic system by the voice of the sound generating device according to the present invention, Figure 2 is a block diagram showing a home theater environment as an example for explaining the environment to which the present invention can be applied 3 is a configuration diagram showing a microphone input relationship between acoustic radiation of a sound device and radiated sound, FIG. 4 is a system diagram showing a system structure of adaptive noise cancellation, and FIG. 6 is a waveform diagram illustrating an example of a microphone output signal according to a conventional method, and FIG. 7 is a waveform diagram illustrating a state in which the microphone output signal of FIG. 6 is removed by the present invention.

As shown in FIG. 1, the non-contact remote control system by the voice of the sound generator according to the present invention includes a sound generator 110 for emitting / amplifying an electrical signal for sound generation, and from the sound generator 110. A speaker 120 which receives the amplified electric signal and converts the vibration signal into a vibration and outputs a sound signal, a microphone 130 which receives a sound signal of the speaker 120 and a voice command for remote control of the sound generating device, and the sound By using the electrical signal of the generator 110 as a reference signal (reference signal) to compensate for the transmission characteristics of the speaker 120 or the microphone 130 and the various transmission paths of the sound signal input to the microphone 130 An adaptive control filter 140 for removing the noise signal, a voice recognizer 150 for recognizing a voice from which the noise signal is removed by the adaptive control filter 140 as a corresponding command, and the voice recognizer The sound generating device controller 160 receives the voice recognition result from the 150 and performs a necessary operation to control the sound generating device 110.

The sound generator 110 may be, for example, TV, video, DVD, audio, and stereo.

Although a plurality of speakers 120 are generally used, the present invention will be described by applying a system for outputting various sound signals as shown in the example of FIG. 2.

2 shows an environment for a DVD (Digital Versatile Disc) home theater as an environment to which the present invention can be applied. In the case of such a DVD home theater, six (center, front-left, front-right) rear-left, rear-right, subwoofer) speakers are used.

During the operation of the sound generator 110, various sound signals are output through the respective speaker 120, and each speaker 120 as well as the voice spoken by the user even if the user speaks a command to remotely control the device by voice. Sound signals from the mix are input to the microphone 130.

In addition, the electrical signal generated from the sound generating device 110 by the respective speaker 120 is converted into a sound signal and input to the microphone 130 through various paths, respectively, the path of the sound signal transmission of FIG. In more detail through an example as follows.

The electrical signal emitted from the sound generating device 110 is transformed by the transfer function determined by the physical characteristics of the speaker 120 to be converted into an acoustic signal, and the sound is generated by the transfer function. The electrical signal of the device 110 is modified in magnitude and phase characteristics of the frequency component.

In addition, although the acoustic signal radiated from each speaker 120 is transmitted to the microphone 130 in a direct path, there is also an indirect path that is reflected by the peripheral structure such as a wall and input to the microphone 130.

Therefore, when the compensation by the transfer function of the speaker 120 and the microphone 130 and the time delay difference, the magnitude difference of the attenuation, and the phase difference caused by various transmission paths are estimated and compensated, the input to the microphone 130 is performed. By removing the noise, the user's clear voice can be delivered to the voice recognizer 150.

In addition, the speech recognizer 150 has a characteristic that the recognition rate drops rapidly when the signal-to-noise ratio (SNR) decreases in its recognition performance. In order to control it, it is necessary to remove the noise signals input to the microphone 130 through the speaker 120 and deliver a clean user's voice to the voice recognizer 150.

The adaptive control filter 140 according to the present invention solves all the necessities as described above to enable the non-contact remote control by the voice of the sound generator 110, Independent Component Analysis (ICA) ) Or Least Mean Square (LMS) algorithm can be applied to Adaptive Noise Canceling (ANC).

Here, the adaptive noise canceling (ANC) is to remove only the portion of the input mixed signal corresponding to the signal generated from the noise source when the signal generated from the signal source is mixed with the signal generated from the noise source and the signal of the noise source can be measured. Means technology.

4 shows a system structure of adaptive noise cancellation. The signal input to the microphone is a noise signal to the speech signal S (t) Will be added. Noise signal Is a noise source Is the signal added after passing through the convolutional channel W _l . The adaptive noise canceling system obtains the coefficient W _l of the adaptive control filter by an adaptive learning algorithm to approximate the system output u (t) negative signal S (t) of Equation 1 as much as possible.

(Equation 1)

From the output of the adaptive noise cancellation system of Equation 1, Equation 2 based on the Independent Element Analysis (ICA) algorithm can be derived.

(Equation 2)

Where η is the learning rate and p (u) is the probability distribution function of u .

FIG. 5 illustrates an example of driving a voice by noise elimination through the adaptive noise removing system of the present invention. As shown in FIG. 5, the adaptive control filter 140 is installed corresponding to the number of line inputs. The reference signal is received from each line input and subtracted from the output signal of the microphone 130. This compensates for the transmission characteristics of the speaker 120 or the microphone 130 and the transmission characteristics by various transmission paths of the acoustic signal, and the clean voice signal from which noise is removed is input to the voice recognizer 150 and recognized as a corresponding command. The result is transmitted to the sound generator controller 160 to perform the necessary function according to the command.

Here, the line inputs represent reference signals obtained from the sound generator 110, and if the sound generator 110 has no line output, the driving voltage of the speaker 120 may be used as the reference signal.

Meanwhile, FIGS. 6 and 7 are waveforms in which microphone outputs are recorded by sequentially speaking voice commands in a DVD home theater listening environment in order to explain an example of practical application of noise reduction to the present invention. FIG. 6 is a microphone according to the conventional method. 7 is a waveform illustrating an example of an output signal, and FIG. 7 is a waveform illustrating a state in which the microphone output signal of FIG. 6 is removed by the present invention.

Here, the waveform of FIG. 6 indicates that it is not easy to distinguish the voice command of the user because the voice command of the user is mixed with the sound signal from the speaker. The waveform of FIG. 7 is an independent element analysis (ICA) algorithm for noise reduction. As a result, the sound signal from the speaker is removed, thereby clearly distinguishing the user's voice, thereby enabling non-contact remote control of the sound generator by the user's voice even during operation of the sound generator.

As described above, according to the non-contact remote control system and method of the sound generating apparatus according to the present invention, the electrical signal of the sound generating apparatus for driving the speaker is used as a reference signal, but an independent element analysis (ICA) algorithm or By applying LMS algorithm and removing noise component input into microphone, user's clean voice command is inputted into voice recognizer, which effectively separates / removes noise input into microphone as well as speaker. There is an excellent effect that can greatly improve the signal-to-noise ratio without significant restrictions on the vocalization position of the speaker, thereby improving the speech recognition performance in all other environments such as a noisy environment. It is possible to remote control by sound of sound generator in true non-contact way. It can be soon as the addition is jinige useful to improve the reliability of the remote control by the voice.

Claims (4)

In the remote control system by the sound of the sound generating device including a sound generator 110, a speaker 120, a microphone 130, a voice recognizer 150; Using the electrical signal generated by the sound generating device 110 as a reference signal (reference signal) to compensate for the transmission characteristics of the speaker 120 or the microphone 130 and the transmission characteristics of the various transmission paths of the acoustic signal, and Contactless remote by the voice of the sound generating device, characterized in that it further comprises an adaptive control filter 140 to remove the noise signal input to the microphone 130 to be transmitted to the voice recognizer 150 through Control system. The method of claim 1, The adaptive control filter 140 is a contactless remote control system by the voice of the sound generator, characterized in that consisting of a digital chip to which the independent element analysis (ICA) algorithm or the least mean square (LMS) algorithm is applied. In the remote control method by the sound of the sound generator, The adaptive learning algorithm is applied to remove the acoustic signal component by the reference signal from the noise input to the microphone by using the electrical signal of the sound generator as a reference signal, thereby inputting the user's voice command to the voice recognizer in a noiseless form. Non-contact remote control method by the voice of the sound generating device, characterized in that configured to be controlled by the voice even during operation of the sound generating device. The method of claim 3, wherein The adaptive learning algorithm is a contactless remote control method using a voice of a sound generating device, characterized in that using the independent element analysis (ICA) algorithm or the least mean square (LMS) algorithm.