KR20030075316A - Speech recognition method and speech recognition system for juke box system - Google Patents

Abstract

PURPOSE:A method for recognizing a voice for a karaoke machine and a system therefor are provided to extract a command word corresponding to the karaoke machine by removing a noise not corresponding to a voice of a man and recognizing the voice. CONSTITUTION:A noise removing unit(100) extracts only an area corresponding to a voice bandwidth of a man from a command inputted in a voice, and converts the voice command of the extracted area into digital signals for removing a noise included in the voice bandwidth of the man. A voice recognizing unit(200) recognizes wavelength of the voice command for comparing the same with syllable wave form of a command word preliminarily stored to search for the consistent command. A serial communication unit(400) converts the retrieved command word into a code a karaoke machine is capable of interpreting, and transmits the converted code to the karaoke machine by using a port communication with the karaoke machine.

Description

Speech recognition method for karaoke machine and its system {SPEECH RECOGNITION METHOD AND SPEECH RECOGNITION SYSTEM FOR JUKE BOX SYSTEM}

The present invention relates to a voice recognition method and system for a karaoke machine, and more particularly, to a method and system for recognizing a voice by actively removing noise.

Karaoke has been developing rapidly since the introduction of a new concept, and as with the PC room, which is similar in character, karaoke is also creating expectations for new ways. A system designed to meet the needs of these customers is a voice recognition karaoke selection system. The voice recognition selection system refers to a system capable of selecting a desired song through voice recognition, compared to a system that was selected by pressing a conventional number.

Speech recognition technology is divided into various types according to the classification criteria. First of all, according to the speaker to be recognized, it is classified into speaker-independent and speaker-dependent recognition technology. First, the speaker dependent system is a system for recognizing a voice of a specific speaker, and a voice dialing system that is currently installed and used in a mobile phone is a representative example. In speaker-dependent systems, a pattern matching technique is generally used to store and register a user's voice before using the system, and to compare the pattern of the input voice with the stored voice when performing actual recognition.

On the other hand, the speaker independent system is for recognizing the voice of an unspecified majority speaker, and there is no need for a user to register the voice prior to the operation of the system as in the speaker dependent system. The speaker independent system collects the voices of multiple speakers and trains a statistical model, and performs recognition using the trained model. Therefore, the characteristic characteristics of each speaker disappear, and the characteristics common to each speaker are highlighted. If you use the same amount of training data for the same vocabulary, the speaker-dependent system generally outperforms the speaker-independent system. However, in the speaker dependent system, if a person other than the speaker whose voice is registered uses the system, the recognition rate is greatly reduced.

Next, it is divided into isolated language recognition system and continuous language recognition system according to the form of pronunciation. In the isolated word recognition system, it is assumed that each word is pronounced again and again, and there is a silent section of sufficient length between each word. Thus, the focus of recognition is on how different each word is from other words and the influence of adjacent words is ignored. In contrast, the continuous word recognition system refers to a system that performs recognition in units of sentences. Each sentence is pronounced as usual, with no special silence between words.

As mentioned above, speech recognition technology is divided into various types according to classification criteria. Each classed technology has its own advantages and disadvantages, so we can't say that one is superior to the other.

In the case of karaoke machines, it is common to employ a speaker independent system because many users use them.

These karaoke machines are typically made up of various operating systems and hardware for different manufacturers. Therefore, it is necessary to interface with the karaoke machine in order to deliver the voice recognition result to the karaoke machine using a method such as a speaker independent system. In order to interface with a karaoke machine like this, the program must be redesigned for the operating system of the corresponding machine. If you repeat these tasks for different operating systems by manufacturer and by product, there are significant problems in applying them to general-purpose products.

In addition, in the case of a karaoke machine there is a problem that the speech recognition rate is lowered due to a lot of noise present in the vicinity.

In order to solve the above problems, the technical problem is to remove noise that does not correspond to human voice and perform voice recognition.

Another object of the present invention is to provide a speech recognition system that can be operated independently of a karaoke machine.

1 is a block diagram showing a voice recognition system for a karaoke machine according to an embodiment of the present invention.

2 is a flowchart illustrating a voice recognition method for a karaoke machine according to an embodiment of the present invention.

3 is a flowchart illustrating an active noise canceling method according to an embodiment of the present invention.

The present invention solves this problem by transmitting a voice recognition result to a karaoke machine using an active noise canceling method and a code in which the remote controller communicates with karaoke.

According to a first aspect of the invention there is provided a speech recognition system for a karaoke machine comprising a noise canceling unit, a speech recognition unit and a serial communication unit. The noise canceling unit extracts only a voice command by removing noise from a voice input together with the noise. The voice recognition unit recognizes a wavelength of the voice command and searches for a matching command by comparing the syllable waveform of the pre-stored command. The serial communication unit converts the retrieved command into a code that can be interpreted by the karaoke machine, and transmits the converted code to the karaoke machine using a port through which the remote control of the karaoke machine communicates with the karaoke machine.

In this case, the noise canceling unit preferably includes a band pass filter, an analog-to-digital converter, and a digital signal processing unit. The band pass filter filters out noise outside the human voice bandwidth, and the analog-to-digital converter converts the filtered voice signal into a digital signal. The digital signal processor removes noise in a human voice bandwidth from the digital signal.

Also, the serial communication unit converts the searched command into a code used when the remote control commands the karaoke machine, and transmits the converted code to the karaoke machine using an RS232C serial communication port or an IR (infrared) port. Do.

The speech recognition system may further include a database that stores syllable waveforms of commands in advance.

According to a second aspect of the invention there is provided a speech recognition method for a karaoke machine. According to this method, first, a user removes noise from a voice input command, and recognizes a voice command from which the noise is removed to search for a corresponding command. Next, the searched command is converted into a code that can be interpreted by the karaoke machine, and the remote control of the karaoke machine transmits it to the karaoke machine through a port that communicates with the karaoke machine.

In this case, when removing the noise, it is preferable to extract only the area corresponding to the human voice bandwidth from the voice input command, convert the extracted voice command into a digital signal, and then remove the noise included in the human voice bandwidth. Do.

In addition, when recognizing the voice, it is preferable to recognize the wavelength of the voice command and compare the syllable waveform with the pre-stored syllable waveform to extract a matching command.

Also, when transmitting to a karaoke machine, it is recommended to convert the retrieved command into the code used by the remote control to command the karaoke machine, and transmit the converted code to the karaoke machine using the RS232C serial communication port or IR (infrared) port. Do.

Next, a voice recognition method and system according to an embodiment of the present invention will be described in detail with reference to the accompanying drawings.

First, a voice recognition system according to an embodiment of the present invention will be described with reference to FIG. 1.

1 is a block diagram showing a voice recognition system for a karaoke machine according to an embodiment of the present invention.

As shown in FIG. 1, the voice recognition system of karaoke according to an embodiment of the present invention includes a noise removing unit 100, a voice recognition unit 200, a song name database 300, and a serial communication unit 400. .

The noise canceller 100 includes a band pass filter 110 for filtering portions other than the human voice bandwidth, an analog-digital converter 120 for converting an analog voice signal into a digital signal, and a digital for removing noise from the digital voice signal. It includes a signal processor 130, and actively removes the noise for the voice input from an external voice input device (not shown) and delivers only the remaining pure command to the voice recognition unit 200.

The active noise removing method removes the frequency band corresponding to the noise from the input signal and receives only the frequency bandwidth corresponding to the human voice as an input to remove the remaining noise.

The song name database 300 stores a song name and a singer name corresponding to the karaoke database.

The speech recognizer 200 recognizes a wavelength of the speech signal from which the noise is removed by the noise remover 100 and performs an acoustic analysis thereof. As the acoustic analysis method, a matching method or a state sequence method may be used. An embodiment of the present invention will be described using a matching method, but is not limited thereto.

Using the matching method, the speech recognition unit 200 sequentially compares the speech signal with syllable waveforms stored in the song name database 300 by one syllable to find a matching syllable waveform. If a matching syllable waveform is found, the voice recognition unit 200 extracts a song name and a mantissa name corresponding to the matching syllable waveform.

The serial communication unit 400 converts the speech recognition result into a code that can be interpreted by the karaoke machine and transmits the speech recognition result to the machine after the speech recognition is completed in the speech recognition unit 200. When sending commands to a karaoke machine, the port that the karaoke machine and the remote control communicate with is used.

In other words, in the karaoke machine, the remote controller transmits the code of the same protocol to the karaoke machine. When transmitting, the remote control communicates using the RS232C serial communication port or IR (infrared) port. In an embodiment of the present invention, the voice recognition result of the voice recognition unit 200 is converted into a code used in the remote control, and the converted code is transmitted to the karaoke machine using such a communication port.

In this way, the hardware and software of the karaoke machine can be operated independently.

Hereinafter, a voice recognition method for a karaoke machine according to an embodiment of the present invention will be described with reference to FIG. 2.

2 is a flowchart illustrating a voice recognition method for a karaoke machine according to an embodiment of the present invention.

When the user first inputs a song name or singer name of a song to be selected through a voice input device such as a microphone as the voice, the input voice is input to the voice recognition system 10 together with ambient noise (S201). The noise removing unit 100 of the speech recognition system 10 extracts a portion corresponding to the pure voice from the voice input together with the noise and transmits the portion corresponding to the pure voice to the speech recognition unit 200 (S202). . This active noise cancellation method will be described in detail below with reference to FIG. 3.

The speech recognizer 200 recognizes the wavelength of the speech signal from which the noise is removed from the noise remover 100 and compares the speech signal with syllable waveforms stored in the song name database 300 one by one syllable (S203). If a matching syllable waveform is found, the voice recognition unit 200 extracts a song name and a mantissa name corresponding to the matching syllable waveform and transmits the same to the serial communication unit 400. If the matching syllable waveform is not found, the voice recognition unit waits for a voice input again. S204).

The serial communication unit 400 converts the result recognized by the voice recognition unit 200 into a code that can be interpreted by the karaoke machine (S205), and transmits the code to the karaoke machine using the RS232C serial communication port or IR port. (S206).

At this time, the screen displayed on the karaoke machine is changed according to the command input by the user. For example, when the user inputs a song name, when there are two or more corresponding songs, the song name and the singer name are displayed together on the karaoke machine and wait for the user's selection input again. Similarly, when a singer name is entered, a list of song names of the corresponding singer is displayed on the karaoke machine and awaits the user's selection input again. In addition, when a user enters a song name and a singer name together, the corresponding song is selected by the karaoke machine.

Next, an active noise canceling method according to an exemplary embodiment of the present invention will be described with reference to FIG. 3.

3 is a flowchart illustrating an active noise canceling method according to an embodiment of the present invention.

First, the noise removing unit 100 removes the noise deviating from the bandwidth of the human voice frequency through the band pass filter 110 in the voice input together with the noise (S301). In general, the bandwidth of a human voice frequency is between 300 kHz and 3400 kHz. The voice input filtered by the band pass filter 110 is converted into a digital signal by the analog-digital converter 120 (S302).

The converted digital signal is removed from the digital signal processing unit 130 and noise corresponding to the bandwidth of the voice frequency is transmitted to the voice recognition unit 200 (S303). The method of removing noise may use existing methods.

In the exemplary embodiment of the present invention, only a case of inputting a song name or a singer name by voice input has been described. However, the present invention is not limited thereto, and a command for adjusting a function of a karaoke machine may also be input by voice. For example, a voice up or down command, a vocal insertion command, a tempo control command, or the like may be input by voice, and the operation in the case where such a command is input may be performed in the technical field to which the present invention belongs. In the following description, it will be easily understood by those skilled in the art.

As described above, according to the present invention, a command corresponding to a voice of a person may be removed from a command input by voice, and voice recognition may be performed to extract a command corresponding to a karaoke machine. In addition, by using the port and code that the remote control and the karaoke machine communicates, it can be operated independently of any kind of karaoke machine.

Claims (9)

Noise canceling unit that removes only the voice command by removing noise from the voice input with the noise, A voice recognition unit recognizing a wavelength of the voice command and searching for a matched command by comparing with a syllable waveform of a pre-stored command, and A serial communication unit for converting the searched command into a code that can be interpreted by a karaoke machine, and transmitting the converted code to the karaoke machine using a port through which a remote control of the karaoke machine communicates with the karaoke machine. Speech recognition system for karaoke machine comprising a. The method of claim 1, The noise removing unit A band pass filter that filters out noise outside the human voice bandwidth, An analog-digital converter for converting the filtered voice signal into a digital signal, and Digital signal processing unit for removing noise in the voice bandwidth of the person from the digital signal Speech recognition system for karaoke machine comprising a. The method of claim 1, The serial communication unit converts the retrieved command into a code used when the remote control commands the karaoke machine, and transmits the converted code to the karaoke machine using an RS232C serial communication port or an IR (infrared) port. Speech recognition system for karaoke machines. The method of claim 1, And a database for storing syllable waveforms of the commands in advance. A first step of removing noise from a voice input command by the user, A second step of recognizing the voice command from which the noise is removed and searching for a corresponding command; A third step of converting the searched command into a code that can be interpreted by a karaoke machine, and A fourth step of transmitting the converted code to the karaoke machine using a port through which the remote control of the karaoke machine communicates with the karaoke machine; Speech recognition method for a karaoke machine comprising a. The method of claim 5, The first step is Extracting only an area corresponding to a voice bandwidth of a person from the voice input command; Converting the voice command of the extracted region into a digital signal, and Removing noise included in the voice bandwidth of the person from the converted digital signal Speech recognition method for a karaoke machine comprising a. The method of claim 5, The second step is to recognize the wavelength of the voice command to compare the syllable waveforms of the pre-stored syllable waveforms and extract the matching command voice recognition method for a karaoke machine. The method of claim 5, The third step is a voice recognition method for a karaoke machine to convert the searched command to a code used when the remote control commands to the karaoke machine. The method of claim 5, In the fourth step, the converted code is transmitted to the karaoke machine using an RS232C serial communication port or an IR (infrared) port.