KR20040016692A - The translator for portability and real time and it's method - Google Patents

Abstract

PURPOSE: A portable simultaneous interpretation device and a method thereof are provided to reproduce a voice as the second national language as a voice by recognizing the voice being inputted by the first national language. CONSTITUTION: A multilingual translation is executed using a natural language voice recognition method. A multilingual interpretation is executed using a natural language voice recognition method and a multilingual translation method simultaneously. A mountable/dismountable memory card method is used for the above procession. A database is constructed based on a multilingual dictionary and the multilingual dictionary is mounted in a memory card. A user selects a reproduction voice being outputted through a speaker.

Description

Portable translator and its method {The translator for portability and real time and it's method}

Prior to the present invention, patented technologies are broadly divided into 1) an interpreter and 2) an interpreter, which are briefly described as follows.

1) Patent proposals have been made for cases where the interpreter is near and far. If the interpreter is located at a short distance, the interpreter uses the wired / wireless communication modulation method to interpret each person directly to the person who speaks each language. After the telephone connection, the interpreter interprets each person directly to the people who speak each language and re-delivers it by voice telephone method.

2) A brief description of a case where a patent proposal is made in such a way that an interpreter can be performed without an interpreter is as follows.

2-1. Once the interpretation area has been subdivided and the information of all the contents that can be spoken by the person in the subdivided area is made into a voice file in the first and second languages, and the person speaks to this device, A patent has been proposed regarding a method of outputting a searched wave file to a speaker by searching a sentence in a database.

2-2. At this point in time, networking is used to mount foreign language translators on servers instead of remote interpreters. That is, a patent has been proposed in connection with a method of connecting to a server equipped with a foreign language translator using a wired / wireless communication method, and then reproducing and outputting a sentence translated into a foreign language required for each foreign language using a voice synthesis unit. have.

Therefore, the present invention not only secured the disadvantages of the method proposed in the 2-1 method, that is, it is impossible to output the sentences other than the recorded voice file and the disadvantage of only outputting the recorded voice through the speaker. It is a big advantage that the -2 method can be implemented to interpret faster than other patent technologies in real time as possible by mounting the role of the server in the embedded system to eliminate the problems in the patented method, that is, the call charges and server fees.

The present invention proposes to solve the above problems, and the object of the present invention is to select a language to be interpreted among the languages of the countries of the world when a person using the first language speaks, the speech recognition unit of the native language speech recognition unit To print the recognized native language text. The output text is then transformed once by the first translation unit into the format of the foreign language translator I so that the translation can be made without error before being input into the foreign language translator I, and the result is displayed on the LCD monitor. ). If the text printed on the LCD monitor has the same meaning as the user intended, the user permits the translation. If the meaning is different, the user does not permit the translation or blocks only the wrong part of the text and converts it to the desired text. Then, when the text intended by the user is input to the foreign language translator I, the foreign language translator I outputs the second language text, and the output second language text is reproduced as the second language voice by the speech synthesis unit B. At this time, the voice reproduced and output is configured to select one of the given celebrity names and output the celebrity voice when the user wants to output the desired person, for example, the celebrity voice. The reproduced voice is then output through the speaker and output to the LCD monitor.

On the other hand, when a second language speaker who listens to the output second language voice speaks mic in his native language, the second language speech recognition unit II recognizes the word and then reproduces the first language voice through the above-described process. Output through speaker.

The LCD monitor is provided with on and off switches for turning on and off the operation, and various language translation and interpretation devices are configured to be detached and used as needed by separating them for each language and storing them in a flash memory card. do.

-The voice recognition unit A recognizes the voice of the person who speaks the first language and outputs the recognized text.

-The first translator converts the text output from the voice recognition unit A into a foreign language translator I format and inputs the foreign language translator I to translate the text without error.

-The foreign language translator I translates the text output from the first translation unit into a second language and outputs the translated text sentence.

-The speech synthesizer B synthesizes the text output from the foreign language translator I as a second language voice and outputs the voice selected by the user to the speaker.

The speech recognizer B recognizes a voice of a person who speaks a second language and outputs the recognized text.

-The second translator converts the text output from the speech recognizer B into the format of the foreign language translator II and inputs the foreign language translator II.

-The foreign language translator II translates the text output from the second translation unit into the first language and outputs the translated text sentence.

-The speech synthesizer A synthesizes the text output from the foreign language translator II as a first language voice, and then outputs the voice selected by the user to the speaker.

2 is a system block diagram of a simultaneous interpretation device.

First, after configuring the embedded system as shown in FIG. 2, only the bootloader is fusing to the flash memory. When the system is started, the system is booted by flash memory and the RTOS kernel and device driver loaded on the memory card, and simultaneous interpretation applications (voice recognition engines A, B, 1st and 2nd translation unit, and foreign languages). Configure translators I and II to access.

The following briefly describes the operation of the simultaneous interpretation application.

When the first Korean voice is input through Mic, the voice recognition unit A performs voice recognition by a voice recognition algorithm (DTW, HMM, ANN, etc.), and as a result, Text 1 is output. Text 1 output as a result of speech recognition is converted into the sentence format of the foreign language translator I, which is inputted to the first translation unit, so that translation can be performed without error before input to the second language translator. It outputs to and configures so that the meaning of what the user says is the same. Subsequently, the modified Text 2 is inputted into the foreign language translator I, is translated into the second language, and at the same time, the output Text 3 is again output on the LCD monitor. Text 3 translated into the second language is inputted to the speech synthesizer B to be reproduced as the second language voice, and the reproduced and output voice is one of the celebrity names given when the user wants to output the voice as a celebrity voice. The person is selected and configured to output the celebrity voice. The reproduced voice is then output through the speaker and output to the LCD monitor.

When a person who speaks a second language hears a word spoken by a speaker and inputs a response into Mic, the voice recognition unit B recognizes the voice and outputs a recognition result, that is, Text 4. The text 4 output as a result of speech recognition is inputted to the second translation unit and transformed into the sentence format of the foreign language translator II in order to ensure that the translation is performed without error before being input to the first language translator, that is, the foreign language translator II. At the same time, the modified output Text 5 is output on the LCD monitor to confirm that the person speaking the second language has the same meaning as what he said. Subsequently, the modified Text 5 is inputted into the foreign language translator II, translated into the first language, and simultaneously outputted to the LCD monitor. Text 6 translated into the first language is input to the voice synthesizer A to be reproduced as the first language voice, and the reproduced and output voice is one of the celebrity names given when the user wants to output the voice as a celebrity voice. The person is selected and configured to output the celebrity voice. The reproduced voice is then output through the speaker and output to the LCD monitor.

3 is a flowchart illustrating the presently described contents.

Therefore, the present invention has the disadvantage of the method disclosed in the patented method 2-1 of the published patent technology related to simultaneous interpretation up to the present point, that is, it is not possible to output sentences other than the recorded voice file and only the recorded voice is output through the speaker In addition to securing the shortcomings, it is not only secured, but also solves the problem of the method disclosed in the 2-2 method, that is, the role of the server is embedded in the embedded system in order to eliminate call charges and server charges, so that it executes faster than other patent technologies in real time as possible. The big advantage is that it can be implemented and portable.

In addition, by changing only the external memory card, it is possible to not only multi-language simultaneous interpretation but also translation.

Claims (5)

Embedded System for Multilingual Translation Using Natural Language Speech Recognition Embedded system for simultaneous interpretation of multiple languages using natural language speech recognition and multilingual translation Embedded system using a removable memory card method to satisfy the claims 1 and 2 Multi-language dictionary is databased and mounted on memory card User selects playback voice output through Speaker