KR19990026316A - VSI's Subtitle Processing System and Method - Google Patents

Abstract

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a system and method for subtitle processing of VSIal using a DSP IC for speech recognition. In the related art, a voice signal of VSIal can be visually recognized by a TV receiver to enable a user to study a language or to be visually recognized by a deaf person. There is a limit to actively use the VSI system because no method for captioning with texts in the text box has been devised.

According to the present invention, it is possible to determine whether the VSI is powered on initially (Step 100), and sequentially determine whether the current VSI is in the regeneration mode. Step (step 101), and subsequently determining whether or not the caption mode key 40 is on (step 102), and if it is determined in this step that the caption mode key 40 is turned on, the digital audio signal in the analog state A step (step 103), storing a voice signal of the digitalized audio signal in a memory (step 104), and subsequently determining whether a section of the currently digitized voice signal is a silent section (step 105); If it is determined in the step that the speech recognition DSP IC 44 is in the silent section, it continuously recognizes the speech data of the continuous sound section and converts it into text data (step 106); Transmitting the character data converted by the voice recognition DSP IC 44 to the microcomputer 45 (step 107), and when the text data is applied to the microcomputer 45 in the step, the text data is again transmitted. Transmitting the text data to the TV receiver 46 (step 108), and outputting the text data to the TV receiver screen when the text data is sent to the OS unit 46 (step 109). It is characterized by judging whether the current VSI mode is the regeneration mode, and if the regeneration mode continues, the process returns to step 103 and ends (step 110).

Description

VSI's Subtitle Processing System and Method

The present invention relates to a closed caption processing system and method using a DSP IC for speech recognition, and in particular, when a user turns on the closed caption mode key provided on one side of the set of the visual receiver or on the key input unit of the remote control. The audio signal is processed into text data by the DSP IC for speech recognition, and the text data is subtitled in OSD to be output to the TV receiver screen under the control of Micom which has received the text data from the DSP IC for speech recognition. will be.

Korean character broadcasting is designed to insert KBIPS signal data into the back porch side of the horizontal synchronization section of the video signal and to use various kinds of information.

For example, the KB PS broadcast signal that is being executed in Korea has data such as the name of a broadcasting station, titles of various pros, time, time information, and so on. The use of such data allows the BSI system to perform reservation recording and the like. Doing.

A type capable of outputting an OSD screen with a simple reservation recording process in a VSI system using such a KBP signal is known as shown in FIG. 1.

Here, the microcomputer 31 for controlling the VSI system, the servo circuit 32 for driving control of the motor on the system deck side, the remote control unit 33b and the set side connected wirelessly to the remote control transmitter 33c The recording signal is taped from a timer circuit 33 to which the key portion 33c is connected, an audio circuit 35 for processing a reproduction or audio recording signal in the system, and an RF signal provided from the tuner and the IF unit 38. The video circuit 37 is provided to the side or processes the reproduced video signal, and the like, and the DMSS unit 36 is configured to control the audio signal to be provided in the normal state at the time of double speed reproduction.

In addition, the system includes a KBPS unit 34 which decodes data such as characters and graphics, such as KBPS, and displays an OSD menu screen for selecting characters, graphics, and functions of various information on the screen, and an antenna. Tuner and eye F 38 for tuning signals received from and generating intermediate frequency signals, power circuit 39 for providing system power, audio / video for selecting external audio / video input and broadcast measurement signals It consists of selection switch (ASW, VSW), etc.

In addition, the KB PS part 34 of the system showed this as shown in FIG.

In this case, the first switch 34b for connecting the antenna-side signals provided from the tuner and the IF unit 38 to the buffers 34c and 34j, and the RF signal provided from the buffer 34c side are supplied with the horizontal synchronization signal ( Inputting a synchronization separator (34d) for separating the HS and the color sync signal (CS) from each other, a data slicer (34e) for detecting the KBP data DT from the composite video signal, and the KBP data DT; Receiving and decoding data contents to provide an On Screen Display (OSD) signal or the like to the second switch 34k, and various program contents executed by the KPS processor 34a. And a programmable ROM 34g that stores data, a system RAM 34f, a character font ROM 34h, and the like.

In addition, a switch is 34L for providing a signal provided from the second switch 34k to a digital / analog (D / A) converter 34i or providing a signal to a video output terminal of the system. It consists of an AD converter 34i which converts an analog signal provided from the city 34L into a digital signal and provides it to the KB PS decoder 34a, a low pass filter 34m connected to the external caption section 4, and the like. consist of.

On the other hand, in the system having the OSD function as described above, it is designed to provide the channel number, subtitle characters, symbols, etc. required for the TV set using the KBPS data to the OSD screen, but conventionally to actively use such a function There was a limitation.

For example, in the related art, there is no method for capturing VSI's audio signal into text on the TV receiver screen so that the user can study language or visually recognize the visually impaired. There was a limit to active use.

An object of the present invention is to automatically turn on the subtitle mode key provided on one side of the Vial set or the key input unit of the remote control in the playback mode of the Vial to improve the problems of the prior art as described above. VSI subtitle processing system and method not only helps the language study of the user by being subtitled in OSD and output to the TV receiver screen, but also allows the hearing impaired to visually recognize VSI's audio output visually and conveniently. To provide.

Particularly, in order to realize the above object, the present invention provides a subtitle provided on one side of a VSI set or a key input unit of a remote control so that an input signal is applied to the MICOM of VSI when the user turns on when the user wants to view a subtitle on a TV receiver screen. A mode signal, a video signal processor and an audio signal processor for processing separate video / audio signals during playback of a videotape, an A / D converter for receiving an analog audio signal from the audio signal processor and digitizing the same; After receiving the digitalized voice data from the D converter and storing it in the memory, the voice data is first analyzed by determining whether the input voice data is a silent section, and if it is determined that it is a silent section, the voice data of the continuous sound section during the silent section. DSP IC for Speech Recognition to Convert Text Data Characterized in that it comprises a microprocessor for controlling the negative o is receiving text data from the edible parts of the DSP IC SD, SD O unit such that under the control of the microcomputer outputting a subtitle text to the TV set screen.

In addition, the present invention is the step of determining whether the V-Sial is initially powered on, and if the V-Sial is powered on in this step, the step of sequentially determining whether the current V-Sial is in the playback mode, and subsequently the subtitle mode key is turned on Determining whether the subtitle mode key is turned on, digitalizing an audio signal in an analog state, storing a voice signal of the digitalized audio signal in a memory, and continuing the current digitalization. Determining whether the speech signal section is the silent section, and if it is determined that the speech section is the silent section by the voice recognition DSP IC, continuously recognizing the speech data of the continuous sound section and converting the speech data into text data; Transmitting the character data converted by the DSP IC for recognition to the microcomputer, and in the step, When the data is applied to the microcomputer, the step of transmitting the text data to the OS again, and if the text data is sent to the OS D and subtitles the text data to output to the TV receiver screen, and when the step is over again the current V It is characterized by judging whether the mode of the sial is the playback mode, and if the playback mode is to continue the A / D processing of the audio signal, or terminates sequentially.

Hereinafter, more specific features of the present invention will be understood by describing the same in detail with the accompanying drawings.

1 is a system circuit block diagram of a Video Cassette Recorder (VCR) having a general KBPS reception function;

FIG. 2 is a reference diagram illustrating a KBPS decoder unit in FIG. 1; FIG.

3 is a block diagram showing a system configuration of the present invention;

4 is a flowchart illustrating a subtitle processing method of BCial according to the present invention.

* Explanation of symbols for main parts of the drawings

40: caption mode key 41: audio signal processor

42: audio signal processor 43: A / D converter

44: DSP IC 45 for voice recognition: micom

46: OSDIVE

3 is a block diagram showing a closed caption processing system according to the present invention.

First, when the user wants to watch a text caption through a TV receiver screen to study a language or to visually recognize voice data of a videotape, the caption mode is provided on one side of a VSI set or a key input unit of a remote controller. When the key 40 is turned on, an input signal is applied to the microcomputer 45 of VSI and thus the microcomputer 45 performs the subtitle mode.

That is, the separated video / audio signals are processed through the video signal processor 41 and the audio signal processor 42, respectively, during the playback of the video tape. In this case, the audio signals are digitalized through the A / D converter 43. .

In addition, if the digitized voice data stored in the A / D converter 43 is stored in the memory under the control of the voice recognition DSP IC 44, the voice recognition DSP IC 44 analyzes the firstly input voice data. After determining whether the input voice data is a silent section, if it is determined that it is a silent section, the voice data of the continuous sound section is recognized during the silent section and then converted into text data.

In addition, the microcomputer 45 receiving the text data from the voice recognition DSP IC 44 controls the OSD 46 to output the text caption on the TV receiver screen.

4 is a flowchart illustrating a subtitle processing method of VSIal according to the present invention. Hereinafter, the processing of the method will be described in more detail with the flowchart of FIG. 4.

That is, the microcomputer 45 on the VSI set side initially determines whether the VSI is powered on by the user (step 100), and if the VSI is powered on in this step, whether the current VSI is in the regeneration mode sequentially. It is determined (step 101).

At this time, the microcomputer 45 determines whether or not the caption mode key 40 provided on the set side of VSI is turned on by the user (step 102).

When the caption mode key 40 is turned on in this step, the analog audio signal is sequentially digitalized by the A / D converter 43 in binary (step 103).

At this time, the DSP IC 44 for speech recognition provided at the rear end of the A / D converter 43 stores the digitalized voice signal in a memory and continuously determines whether the current digitalized voice signal section is a silent section. (Step 104, step 105)

For reference, a filter (not shown) is provided at the front end of the A / D converter 43 to limit the frequency band to about 1KHz to 4KHz. That is, the filter serves to limit the frequency band of the audio signal digitized through the A / D converter 43 to the voice region of 1KHz to 4KHz.

If it is determined in the step that the speech recognition DSP IC 44 is in the silent section, it continuously analyzes the speech data in the continuous sound section and converts it into text data (step 106).

At this time, the algorithm for recognizing the voice data of the continuous sound intervals is typical, such as DTW (Dynamic Time Warping), HMM (Hidden Markov Model), ANN (Artificial Neural Network), each of which will be described as follows.

First, DTW is a method of obtaining the similarity between the pattern of the voice signal as a reference and the input voice signal using a dynamic program. Unlike DTW, which is an intuitive algorithm, HMM is a probabilistic theory-based algorithm that has been widely used from isolated word recognition to continuous speech recognition. It is assumed that speech can be modeled using the Markov process. And recognition process. First, in the learning process, from the observation date 0 of the training data, the conditional probability λ estimates the parameters of the HMM λ according to the optimization criteria, and determines the model of the word to be recognized. On the other hand, in the recognition process, the conditional probability P (0 / λ) of the input speech pattern for each HMM is obtained, and the word corresponding to the HMM having the maximum conditional probability is regarded as the recognized result. HMM is currently adopted in a large number of commercially available products, and has many advantages, especially speaker-independent and continuous speech recognition, and the calculation is smaller than the DTW mentioned above. Disadvantages include lack of discrimination between models and neglect of the association between voice signals when there is a lack of training data.

ANN is called artificial neural network, which is designed to model human information processing process and to find and process information inherent in input pattern through learning by connecting many processing elements in parallel.

Generally, the speech recognition system is carried out through the steps of speech signal → speech detection → feature extraction → similarity measurement → final recognition determination → result through the reference model set. In reality, it is difficult to implement an ideal speech detector and in most cases zero crossing It is designed to be less affected by the surrounding environment by detecting the speech section in real time using only the parameters of the time domain with a small amount of computation such as rate and energy, and adapting the threshold of the parameter that detects the voice according to the change of the surrounding environment. do. In addition, the various algorithms described above are combined to perform each step described above in an optimal state.

On the other hand, when the voice data of the continuous sound section is converted into text data by the voice recognition DSP IC 44, the voice recognition DSP IC 44 then transfers the converted text data to the microcomputer 45 (step). 107), the microcomputer 45 which has received the text data sends the text data to the OS 46 again (step 108).

At this time, the OS 46 receiving the text data subtitles the text data and outputs it to the TV receiver screen (step 109).

Finally, the microcomputer 45 determines whether VSI is currently in the playback mode (step 110). At this time, if the playback mode is continuously performed, the microcomputer 45 returns to step 103. Will end.

In the present invention, when the user turns on the caption mode key 40 provided on one side of the VSIAL set or the key input unit of the remote control, the audio signal is converted into character data by the DSP IC 44 for voice recognition. The language study of the user is performed by processing the text data under the control of the microcomputer 45 receiving the text data from the DSP IC 44 for speech recognition and outputting the caption data on the TV receiver screen. In addition to helping the hearing impaired, the hearing-impaired person can visually recognize the voice output of VSI.

Claims (2)

When the user wants to watch a text caption through the TV receiver screen, the caption mode key 40 is provided on one side of the VSI set or the key input unit of the remote control so that an input signal is applied to the microcomputer 45 of VSI. A video signal processor 41 and an audio signal processor 42 for processing separated video / audio signals during playback of a tape, and an A / D converter for receiving an analog audio signal from the audio signal processor 42 and digitizing the same. 43) and the digitalized voice data received from the A / D converter 43 are stored in the memory, and the voice data is analyzed first to determine whether the input voice data is a silent section, and then determined to be a silent section. And a voice recognition DSP IC 44 for recognizing voice data in a continuous sound section and converting the voice data into text data during a silent period. Microcomputer 45 to control the OS dee 46 by receiving the text data from the; and the OS dee 46 to output text subtitles on the TV receiver screen under the control of the microcomputer 45, etc. Subtitle processing system of V'Sial characterized in that. Initially determining whether the VSI is powered on (step 100); if the VSI is powered on in this step, sequentially determining whether the current BD is in the playback mode (step 101); Determining whether or not the key 40 is on (step 102), and if it is determined that the subtitle mode key 40 is turned on (step 102), digitalizing the audio signal in an analog state (step 103); Storing the audio signal of the audio signal in the memory (step 104), and subsequently determining whether the section of the current digitalized voice signal is a silent section (step 105); and in the step, the speech recognition DSP IC 44 If it is determined that the silent section is determined, the speech data of the continuous sound section is continuously recognized and converted into text data (step 106), and the text data converted by the DSP IC 44 for speech recognition. Transmitting the character data to the microcomputer 45 (step 107), and transmitting the character data to the OS-D 46 again when the character data is applied to the microcomputer 45 (step 108); When the text data is sent to the SD unit 46, the text data is subtitled to be output to the TV receiver screen (step 109), and after the above steps, it is determined whether the current VSI mode is the playback mode to continue playback. If it is in the mode, the process returns to step 103, or ends (step 110).