JP2003216607A - Electronic translating apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To make a conversation smoother relating to an electronic translating apparatus by giving each word its degree of importance, and sequentially performing the processes of voice recognition, translation and display for input voice at each degree of importance and in order of decreasing degrees of importance. <P>SOLUTION: The electronic translating apparatus includes a voice input part; a voice recognition table in which word and pronunciation data and importance level information are stored in association; a grammar table for the language of the input voice; a voice data storage part for storing the input voice; a retrieving part for retrieving pronunciation data that can coincide with the stored voice data and for selecting word candidates which correspond to the pronunciation data retrieved; a recognizing part for determining words by referring to the grammar table as to the selected word candidates; a display part for displaying the words determined; and a control part which performs the process of selecting word candidates and determining and displaying words at each degree of importance and in the order of decreasing degrees of importance so as to display sentences in steps based on the language of the input voice. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an electronic translation device having a voice recognition function.

[0002]

2. Description of the Related Art Conventionally, electronic devices having a voice recognition function and a translation function have been developed. For example, a portable voice recognition electronic device that converts voice input from a microphone into digital data, records the data once, recognizes the digital data for each word, and then searches a translation dictionary to extract all target translation words. The dictionary is described in JP-A-4-10166.

Further, in Japanese Unexamined Patent Publication No. 7-105220, the translated sentence of the conversation sentence selected by the user is displayed to the conversation partner, and the answer selection candidates for the translation sentence are displayed to indicate the conversation partner. There is described a conversation translation device for displaying a reply selected by the user in an expression that a user can understand and realizing a conversation in a writing style.

Further, in Japanese Unexamined Patent Publication No. 9-251462, when a part of words of a conversational example sentence stored in advance is replaced and the translated sentence is displayed, the replacement candidate words (elements) are classified into categories (elements). There is described a machine translation device that classifies each category and stores it in advance and displays words (elements) belonging to a category selected by a user.

Further, in Japanese Unexamined Patent Publication No. 10-320397, features of voice input from a microphone are extracted and recognized, and a translated sentence corresponding to the result of voice recognition is searched using a table stored in advance. Then, a portable voice translation terminal that displays the recognition result and the bilingual sentence or outputs the bilingual sentence by synthesized voice is described.

[0006]

However, Japanese Laid-Open Patent Publication No. 4-1 is used.
Since the one described in Japanese Patent No. 0166 has a voice recognition function, it is possible to obtain a foreign word translation of a target word earlier than an electronic dictionary of a key input system for inputting characters, but it is only necessary to extract a translated word. Therefore, it is not always possible to extract an appropriate translated word during a general two-way conversation, and if the input voice is long, it may take too long to obtain the translated word, and quick response cannot be performed.

In the device disclosed in Japanese Patent Laid-Open No. 7-105220, the device can be shown to the other party in conversation.
There is a problem that it is necessary to select a reply, smooth conversation cannot be performed, and reply options are limited to those prepared in advance.

[0008] Further, in the method disclosed in Japanese Patent Laid-Open No. 9-251462, although it is possible to increase the variation of conversational example sentences and display the desired translated sentence quickly,
Since it does not have a voice processing function, it cannot be used as a device for smoothly performing a two-way conversation between speakers who speak different languages.

Further, according to the technique disclosed in Japanese Patent Laid-Open No. 10-320397, it is possible to display or output a synthesized voice of a bilingual sentence or an example sentence corresponding to the recognition result of the inputted voice, but understand the conversation of the other party. To do this, you must look at all the translated results, and some search is required to get the translated text, and it has enough features to facilitate two-way conversations. I can't say I'm doing it.

Therefore, the present invention has been made in consideration of the above circumstances, and has a voice recognition function and classifies the registered words according to the importance, so that the two-way conversation can be smoothly performed. It is intended to provide an electronic translation dictionary that assists the user.

[0011]

According to the present invention, a voice input section, a voice recognition table in which at least words, pronunciation data and importance information are stored in association with each other, and a grammar table for a language of input voices. And a voice data storage unit that digitizes and stores the input voice and the pronunciation data that can match the voice data stored in the voice data storage unit, and searches for word candidates corresponding to the retrieved pronunciation data. A search unit to be selected, a recognition unit for determining a word by referring to the grammar table for the selected word candidate, a display unit for displaying the determined word, a selection process for the word candidate, a word determination process, and Word display processing,
A control unit that causes the groups of the importance information to be sequentially performed in descending order of importance information, and that causes the sentence in the language of the input voice to be displayed on the display unit step by step according to the importance. The present invention provides an electronic translation device characterized by being provided.

According to this, since the sentences spoken by the other party in the dialogue scene can be gradually displayed in order from the important words, the conversation can be made smooth. In addition, the content spoken by the other party can be analogized or understood before the full text of the sentence spoken by the other party is displayed.

Further, the system further comprises a translation dictionary and a translation unit for translating the word determined by the recognition unit into a predetermined other language, and the control unit, for each group of the importance information,
Further, the present invention provides an electronic translation device, characterized in that the translation unit performs a translation process on a word determined by the recognition unit in descending order of importance information.

Further, the voice recognition table is stored in a rewritable memory, and usage frequency information used for searching or recognizing words in the voice recognition table is stored in the memory for each word, and the control unit May change the importance information based on the usage frequency information.

In the electronic translation apparatus of the present invention, the voice input section is a microphone, and the display section is an LCD or EL.
A small display device such as a display element can be used.
Further, a key for inputting not only voice but also characters and symbol data, and a key for selecting a dictionary and activating voice recognition may be provided.

For the voice recognition table, the grammar table and the translation dictionary, a read-only memory, for example, a semiconductor element such as a ROM or a recording medium such as a CD-ROM can be used, but a rewritable memory such as a RAM or a hard disk. May be stored in. The voice data storage is R
It can be stored in a rewritable memory such as an AM or a hard disk. Further, each function of the search unit, the recognition unit and the control unit is controlled by a microcomputer including a CPU, a ROM, a RAM, an I / O controller, a timer,
It can be realized by operating based on a program stored in the OM or the like.

The pronunciation data stored in the voice recognition table is sample data in which analog voices when each word is generated are digitized and stored in advance, and are digitized voice waveform data. Recognizable word candidates are selected by determining the similarity between the pronunciation data and the data obtained by digitizing the actually input analog voice.

The importance level information is information indicating whether the priority level of processing such as recognition and translation is high or low, and is given in advance for each word stored in the voice recognition table. For example, using an integer as importance information, 1
May have the highest importance, and the importance may decrease in order as the number increases.

[0019]

BEST MODE FOR CARRYING OUT THE INVENTION <Structure of Electronic Translation Device> FIG. 1 is a schematic structural view of an embodiment of an electronic translation device of the present invention.
Here, it is intended to translate between Japanese and English and has a voice recognition function for recognizing English voice and converting it into character data. However, the language to be translated is 2
The present invention is not limited to one language, but may be any existing language such as German, French, or Chinese.

In FIG. 1, reference numeral 11a is a cabinet, 11b is a display portion, 11s is a microphone,
As with the conventional electronic dictionary, a large number of keys for setting and inputting various functions and characters are provided. For example, the code 11
d is a key to select an English-Japanese dictionary that converts English into Japanese,
Reference numeral 11e is a key for selecting a Japanese-English dictionary for converting Japanese into English, and 11f is a key for executing voice recognition.

FIG. 2 shows a schematic block diagram of an embodiment of the electronic translation apparatus of the present invention. As shown in FIG. 2, this device mainly includes a display unit 12 a and a key input unit 12.
c, a voice input unit 12d, and a control unit 12b. The control unit 12b includes a display control unit 12e that controls the display unit 12a, an input control unit 12g that controls the two input units 12c and 12d, and the entire apparatus. CP to control
U12f, ROM12 for storing programs and data
h and RAM 12i.

For the display section 12a, an EL element or the like can be used in addition to the LCD. As the input key unit 12c, a keyboard, a numeric keypad, or the like can be used in addition to the keys shown in FIG. The voice input unit 12d corresponds to the microphone 11s in FIG.

The ROM 12h is a program for causing the CPU to execute each function of the apparatus and information necessary for executing functions such as translation and voice recognition (a voice recognition table 15b, a dictionary 15a as shown in FIG. 5). , Correspondence word list 15c, grammar table 15d, answer correspondence table 15
e) is stored so that it can be read. The dictionary 15a contains
The information stored in a so-called ordinary dictionary such as meanings of Japanese and English words, variations, phonetic symbols, example sentences, explanations, etc. is stored.

FIG. 3 shows an embodiment of the contents of the voice recognition table. Here, the word data 13a, the pronunciation data 13b, and the degree of importance 13c, which are classified for each part of speech, are stored as a set for each word. The importance 13c and the part-of-speech information are used when selecting a candidate of a word recognized by voice recognition.

FIG. 4 shows an embodiment of the grammar table 15d. This grammar table stores so-called grammatical rules consisting of Japanese and English word orders, idioms, variants.

Next, an outline of a method of using the electronic translation dictionary of the present invention and a process of translating from English into Japanese in a conversation scene will be described. Consider a case where a user whose first language is Japanese (Japanese) has the electronic translation device of the present invention and has a conversation with a foreigner who speaks English.

First, the user presses the voice recognition key 11f to bring the microphone 11s closer to a foreigner and ask the foreigner to speak an English sentence. For example, if a foreigner says "what time is it now", this voice is input to the microphone. The input voice is recognized as each English word by voice recognition, and the word is displayed in the order of predetermined importance set in advance.

For example, as shown in FIG. 6 (a), the most important words "what", "time", and "now" are first displayed on the display section 12a, and at the same time, the translation process is performed. Is also executed and a Japanese translation is displayed below each English word. Next, in the order of importance, the remaining words are similarly subjected to voice recognition, translation processing and the like. Then, when the entire sentence is recognized, the completed English sentence including the input English sentence including "is" and "it" and the Japanese translation "What time is it now?" Are displayed in Fig. 6 (b). It is displayed as shown in.

In other words, according to the present invention, the results are not displayed at once after the speech recognition and translation processing of the entire input speech are performed, but the words are sequentially processed in the order of high preset importance. The feature is that some words are displayed even during translation. Further, after the translation of all the words is completed, the entire input voice is displayed as a sentence using both the input language and the translated language.

<First Embodiment> Next, the general operation of the electronic translation apparatus of the present invention will be described. FIG. 7 shows a flowchart of the operation of an embodiment of the electronic translation apparatus of the present invention. First,
When the user presses the voice recognition key 11f shown in FIG. 1 (step 21a), the CPU of FIG. 2 detects that there is this key input, via the input control unit 12g.

Next, when the other party of the conversation or the user makes a voice toward the microphone 12d, the voice input to the microphone is converted into an analog electric signal and subjected to analog-digital conversion (A / D conversion). , Are stored in the storage device such as the RAM 12i as digital signal data (step 21b). At this time, it has not been recognized as a character or a word.

Next, while comparing this digital signal data with the pronunciation data 13b stored in advance in the voice recognition table, words that are recognition candidates are searched and extracted (step 21c). Here, simply, the voice recognition table 15
Instead of sequentially searching b, the search is performed in order from the most important word assigned to each word in the dictionary, and the comparison is attempted. That is, first, a word of importance 1 having the highest importance assigned in advance is compared and collated, and pronunciation data that can be matched with digital signal data is extracted, and words stored in correspondence with this pronunciation data are recognized as recognition results. It is temporarily stored in the RAM as a candidate for (step 21d).

At this time, the number of candidates may be one or two.
More than one may be extracted. For example, "she", "s
If all three words “ee” and “sea” are set to the importance level 1, all three words may be extracted as candidates for the voice input “sea”. However, "she" has a priority of 1 and "se"
If "e" and "sea" are set to importance level 2,
Only "she" is extracted as a recognition candidate. However, in the process of adjusting the sentence by referring to the grammar table in the subsequent step 21i, the difference in the part of speech and meaning of each word is recognized, and the grammar and the meaning that are more suitable than the position in the pronounced sentence of the word are found. To be selected. At this time, if the previously displayed word of importance level 1 is not suitable, it is immediately selected from the candidates and changed to importance level 2.

Further, in step 21e, the CPU
Uses the grammar table 15d to determine words that match the grammar from the extracted word candidates while considering the word order and the like. As a result, the input voice data is divided into words in consideration of importance, and words that match the grammar are extracted.

Next, in step 21f, the determined words are rearranged according to the grammar and temporarily stored in the RAM. For example, the entered "what time is it now"
If "what", "time", and "now" are set as the words of importance level 1, these three words are extracted first, and the grammar table such as word order is referred to.
The data is stored in the RAM in the order of "what", "time", and "now".

Next, the words thus arranged are displayed in that order on the display unit 12a (step 21g). For example, in the above example, the three English words "what" and "tim
e ”and“ now ”are displayed. Here, in FIG. 6 (a), only the upper English word is displayed, but a user who is a Japanese person can see what the other word is by seeing this English word even if he / she cannot hear it correctly. You can quickly predict or understand what you said.

Although not all the uttered words are displayed, the word of the highest importance set in advance is displayed. Therefore, even if the other words are not known in the middle of the recognition process. , In some cases, you can quickly analogize or understand the entire utterance content of the other party.

Next, in step 21k, the English-Japanese dictionary is referred to, the Japanese translation of the determined English word is searched, and the Japanese translation of that word is displayed below the English word displayed in step 21g. Here, so-called voice-recognized importance 1
For each English word, it means that each word has been translated into Japanese. For example, "what" for "what",
If the Japanese translations "time" for "time" and "now" for "now" are searched, the Japanese translations are displayed below each English word in the upper row, as shown in Fig. 6 (a). It

In the state of FIG. 6 (a), only the speech recognition, translation and display of the word of importance 1 have been completed, but the recognition processing and translation processing of all input speech have not been completed yet. However, even during such processing, it is easy for the Japanese user to easily understand what the foreigner said by just looking at the English words and Japanese translations that are considered to be the most important. In some cases it can be understood quickly. In the display of FIG. 6 (a), the sentence is incomplete and its meaning is unknown, but during the actual conversation, "what
Just looking at the meaningless text "time now" and "what, time, now" means what the other person wants to say. In this example conversation, "I want to know the current time." In some cases, the user can empirically infer.

Therefore, the user can promptly take the next response (for example, reply to a question) only by seeing a part of the word and the translated word in the uttered voice although it is incomplete, and the conversation is smooth. Can be
That is, a smooth conversation can be assisted.

On the other hand, after recognizing all the input voices and further translating all the recognized words and creating an entire translated sentence that can be understood according to the grammatical rules, sentences corresponding to all the input voices are not formed. In the case of display, this display is made after a considerable amount of time has elapsed than the display shown in FIG. 6A, and the result display is lacking in speediness. If the displayed sentence is an accurate translated sentence, it is possible to take corrective action immediately after seeing it, but the displayed translation result sentence may be an incomplete sentence that is difficult to understand, Has to guess what the other person said by looking at the displayed incomplete sentence, and cannot take the next action promptly.

In reality, various factors such as the input state of voice, the accuracy of voice recognition, the ability of machine translation, the user's comprehension to infer the content of conversation by looking at some words displayed during the process, etc. Therefore, it is not always possible to make the conversation smooth at the stage of seeing the display as shown in FIG. 16 (a), but the translated texts displayed after all the search, recognition and translation processes are not understood. Therefore, as in the present invention, it is very effective in an actual conversation scene that a smooth conversation can be quickly assisted by displaying during processing.

As described above, after displaying some of the words in the process of processing in step 21k, next, in step 21h, it is determined whether or not the sentences for all the inputted voices are completed. If the recognition and translation processing has been completed for all voices, the processing ends. If the sentence is not completed, for example, as shown in FIG. 6A, only the words of importance 1 have been processed, step 21i.
Go to.

In step 21i, the sentence is adjusted by referring to the grammar table. Specifically, for example, when only the word “what, time, now” of importance 1 is displayed, the basic five sentence patterns (S + V, S + V + C, S + V +
(O, S + V + O + C, S + V + O + O) are adjusted by supplementing the missing words. Next, in step 21j, of the word candidates selected in step 21c, the word that has not been processed yet and has the next highest importance is selected.

For example, after the display processing of the word of importance 1 is performed, the word of importance 2 is selected. Then, the processing of steps 21d to 21k described above is executed for the word of importance 2. That is, words of importance 2 are selected, rearranged, stored in RAM, displayed, and translated into Japanese.

When the above process is completed for the word of importance 2 and the sentence is not completed yet (step 21h), steps 21i, 21j and 21d to 21k are executed for the next word of high importance. The process of is executed. That is, for all word candidates selected in step 21c, search, recognition, translation,
Processing such as display is performed, and the processing is repeated for all word candidates until the sentence is completed.

When all word candidates have been processed,
As shown in FIG. 16A, the English translation may be displayed and the Japanese translation may be displayed below each English word. In this case, even if you look at the Japanese translations in order, it is a sentence whose meaning content is unknown as a Japanese sentence, but even if the meaning is unknown as a sentence, just by looking at the translation of each word, the other party said what Sometimes you can understand. It takes a relatively short time to display the Japanese translation for each word in this way, but it takes a considerable amount of time to display a translated sentence that passes the full meaning. Sequentially displaying the translations is effective in that the next response (for example, reply) can be promptly made.

In the above-mentioned example of "what time is it now", if "is" and "it" are assigned the importance level 2, the Japanese translation is selected for "is" and "it". . At this time, the Japanese translations of "is" for "is" and "it" for "it" are selected and displayed below the English word.

However, when selecting a word by referring to the grammar table in step 21e, it is determined that the voice input from the word order is a question sentence, considering a word of importance 1 that has already been translated. However, instead of applying the translated words of “is” and “it”, a Japanese translation in which the meaning is understood may be generated for the entire sentence. That is, when the processing of "is" and "it" is completed, or when the processing of all input words is completed, it takes a little time to process, but Japanese that is meaningful as a sentence is generated and displayed. You may do it. For example, in FIG.
As shown in (b), the Japanese translation "What time is it now?" Is displayed.

The English words and their Japanese translations corresponding to the voices input as described above are displayed. First, since the voice-recognized word candidates are classified and translated according to preset importance levels, they are first important. Only the words with high frequency are displayed quickly, and it is possible to infer the content of the conversation even in the middle of translation, and it is possible to quickly understand the content of the other party and take the next response (reply) during the conversation. It will be possible.

Further, since the voice recognition and translation processing is performed in order of importance set for each word, the recognition target in the dictionary stored in the RAM is narrowed down, and high speed processing is possible. Moreover, since the preset highly important words are preferentially translated and displayed, the user can infer and understand the conversation content at an earlier stage than translating and displaying the input speech in the word order. be able to. Here, the criterion for assigning the degree of importance set for a word is not particularly determined uniformly, but may be changed appropriately. The importance may be set on the basis of information such as the frequency of use in everyday conversation for each word, the part of speech, the position in the sentence, and the accent (inflection of pronunciation).

For example, in a dialogue, there are many patterns of "question" and "answer to it".
In order to speed up the analogy and understanding of conversational sentences, it is conceivable to set high importance for question words, pronouns, and common nouns, and set low importance for articles and adverbs. In addition, it is preferable to set a high degree of importance to words that specify time and place, and words that are relatively easy to use and are frequently used in junior high school.

Further, in order to rewrite the degree of importance set in advance, a flash memory or an EPRO is used.
It must be stored in a rewritable and non-volatile memory such as M. The importance level may be rewritten by the user's own judgment using key input, etc., but it is automatically changed in consideration of the frequency of use of words from the viewpoint of preventing erroneous operations. It is preferable to be able to. In addition to the frequency of use, a so-called learning function may be provided in consideration of grammar, words before and after, intonation of pronunciation, position in a sentence, etc., so that the importance can be dynamically changed.

<Second Embodiment> In the above embodiments, the voice of the conversation partner is input and the translated sentence is displayed. However, when the conversation partner asks a question, the voice recognition is performed. It is also possible to judge that the sentence generated as a result is in the question form and display the candidates for the answer to the question.
Hereinafter, an example of displaying the answer candidates will be described. The electronic translation device of the present invention is provided with an "answer" key 11k as shown in FIG. When the user presses this "answer" key 11k, the answer display function is activated.

FIG. 8, FIG. 9 and FIG. 10 are explanatory views of an embodiment of the display contents of the present invention. For example, "what tim
It is assumed that when the voice of the conversation partner "e is it now" is input, the recognition result and the translated text as shown in FIG. 8A are displayed. In this display state, the user "answers"
When the key 11k is pressed, the answer sentence as shown in FIG. 8B is displayed. If the user sees this display and determines that it is correct, he or she may speak the displayed English sentence by himself or show this display to the conversation partner. Figure 8 (b)
If the sentence displayed as is correct, the user can reply to the conversation partner easily and quickly by simply pressing the "answer" key 11k once.

The same applies to FIGS. 9 and 10, but in the case of FIG. 10, a plurality of answers are displayed for the question of FIG. 10 (a). When a plurality of candidates are considered as answers, all of them are displayed so that they can be selected by the user's judgment. By the way, the sentences that are candidates for answers are stored in the answer correspondence table 15e in the ROM.
Is preferably stored in advance. However, RA
An answer sentence originally defined and input by the user may be stored in a rewritable memory such as M.

FIG. 11 shows an explanatory diagram of an embodiment of the answer correspondence table 15e. For example, the answer correspondence table is
The keywords that are likely to appear in the question sentence and the keyword that is the center of the answer sentence or the answer corresponding to this question are stored in association with each other. However, even if the answer that cannot be uniquely determined, such as the time, is stored as "10:00 AM", instead of displaying the answer as it is, the current time of the built-in clock should be reflected. Is preferred. In Figure 11, "how + get t
For the question containing the word "o", two answers are prepared in advance.
Two answer sentences can be displayed as shown in 0 (b).

FIG. 12 shows an operation flowchart of an embodiment of the processing for displaying this answer. This process is shown in FIG.
This is started when the "answer" key 11k is pressed in the state where the recognition result as shown in (b) is displayed. First, the displayed English sentence of the recognition result is divided into words (step 33a). Next, a key word or a phrase is picked up from the divided words and English sentences (step 33b). At this time, the key word / idiom can be picked up by using the importance data described above.

Next, it is checked whether or not the picked-up words and the like and the words in the "question" of the answer correspondence table 15e match, and the "answer" corresponding to the matched "question" in the table 15e. Is retrieved and extracted (step 33c). Then, referring to the grammar table, rearrange the words in this "answer" so that they are grammatically correct,
Alternatively, an answer sentence is created by supplementing the necessary words so that the sentence becomes grammatically meaningful (step 33d).
Finally, the prepared answer sentence is displayed (step 33).
e). The user may look at this display and, if correct, say a reply or show the display to the other party. As described above, if the "answer" sentence can be displayed, it is possible to respond more appropriately and quickly in a conversation scene.

<Third Embodiment> Next, an embodiment for changing the displayed English sentence to another one will be described.
For example, you may want to display an English sentence of another expression that has a similar meaning to the displayed English sentence, or you may want to search for the sentence of another answer. In such a case, use the following function. Here, the following function keys are provided.

First, as shown in FIG. 11, a "sub" key 11i is provided. The "sub" key 11i is a key for switching the dictionary to be used when the dictionary is provided in plural. Each time the "sub" key 11i is pressed, the dictionary used as the main and the dictionary used as the sub are switched. Further, a "switch" key 11l is provided to switch between the main display when the main dictionary is used and the sub display when the sub dictionary is used. "Return" key 11j
Is used to erase the contents of the sub-display and return to the main dictionary and the main display. The "search" key 11g is used when a process such as word search using the main dictionary is newly performed.

For example, when the "sub" key 11j is pressed while the display as shown in FIG.
While the display content of (a) is stored in the RAM, the display is switched to that shown in FIG. That is, the sub screen for the search is displayed so that the search process using the English-Japanese dictionary which is the sub dictionary can be performed.

By inputting an English word / idiom such as "time" in the state where the sub-screen of FIG. 13 is displayed, the Japanese translation can be searched for. Further, during the search by this sub screen, the "switch" key 11l can be pressed to return to the main screen, and the "switch" key 11l can be pressed again to switch to the sub screen. Also, before pressing the "answer" key while displaying the "question" as shown in FIG.
By pressing the "sub" key, the "answer" may be searched and displayed on the sub screen.

Further, the "switch" key 11l only switches the display screen while holding the states of the main screen and the sub screen respectively, but if the "return" key 11j is pressed while the sub screen is displayed, All the information on the sub screen may be erased, leaving only the main screen.

When the "sub" key 11i is pressed with the sub screen displayed, the information on the main screen is erased, and the information on the sub screen is temporarily stored as the information on the main screen for retrieval. May be continued. At this time, if the "sub" key 11i is further pressed, a new sub screen is displayed and a new search can be performed. further,
When the "search" key 11g is pressed while either the main screen or the sub screen is displayed, the information on both screens may be erased and a new search may be performed on the main screen.

<Fourth Embodiment> Next, an embodiment of the function of generating a sentence having the opposite meaning in the example sentence search will be described. Here, the "jump" key 11p shown in FIG.
And the "opposite" key 11o. "Jump" key 1
Reference numeral 1p is a key for selecting a word in the example sentence displayed on the screen, and each time the key is pressed, a cursor is displayed on the sequentially selectable word. The “opposite” key 11o is a key for searching for and displaying an opposite word of the currently selected word.

FIG. 14 is an explanatory diagram of an embodiment of generation of a sentence using an antonym. When the "jump" key 11p is pressed while the example sentence as shown in FIG. 14 (a) is displayed,
A word registered as a dictionary entry word is selected. FIG. 14A shows a state in which the word “good” 51c is selected from the example sentences in the screen 51a, and the “anti” 51d in the lower right corner indicates that the selected word is an opposite word. It shows that it is possible to jump.

When the "opposite" key 11o is pressed in this state, the selected word "good" is changed to a word having the opposite meaning as shown in FIG. 14 (b). Here, "good" 5
1c changes to "poor" 51e, and "good at" changes to "weak". In order to search and display the antonyms, the antonym dictionary 15a, the corresponding word list 15c, and the grammar table 1 shown in FIG.
5d is used.

FIG. 15 shows an operation flowchart of the opposite word selection process of the present invention. As a premise, it is assumed that an English example sentence and its Japanese translation as shown in FIG. 14A are displayed. First, when the CPU detects that the jump key 11p has been pressed (step 52a), it selects one of the words forming the example sentence and is capable of jumping,
The cursor is displayed (step 52b). The user looks at this display (FIG. 14 (a)), confirms that the word for which the opposite word is to be searched is displayed in the cursor, and if it is not the word to be searched, the user may press the jump key 11p several times. , Move the cursor to the desired word.

Next, the user presses the opposite key 11o and presses C
When the PU detects this push, the antonym dictionary 15a
Then, the corresponding word list 15c is searched to select a corresponding word having the opposite meaning (step 52d). For example, if "good" and "poor" are stored in the corresponding word list 15c as opposite words, "poo" is stored.
The word "r" is selected. Also, if there are other words associated with "good", all of them are selected at this point.

Next, referring to the grammar table 15d, an antonym suitable for the example sentence is selected (step 52e). "Go
When there are a plurality of words corresponding to "od", a word that is compatible with the currently displayed example sentence as an antonym is selected. Here, for example, whether or not it can be matched as an opposite word can be determined by the meaning of the original sentence shown in the grammar table or the like, and the meanings of the words before and after the word to be jemp. Here, the Japanese translation is selected along with the English word having the opposite meaning. Next, an example sentence is created by replacing the word selected by the jump key 11p with the opposite word selected in step 52e (step 52f), and the example sentence is displayed (step 52g).

As described above, by performing a simple key operation with two keys, it is possible to easily search and display an example sentence using a word having an opposite meaning with respect to a certain word in the example sentence currently displayed. Can be made. Further, even when a conversation is conducted using the device of the present invention, by using this function, a sentence having the opposite meaning can be easily and quickly transmitted to the other party.

FIG. 16 shows an explanatory view of another embodiment for generating an example sentence of the opposite word of the present invention. FIG. 16A shows the example sentence 61a currently displayed, and the jump key 11p
Indicates that "often" 61b is selected. FIG. 16B shows a display after pressing the opposite key 11o in the state of FIG. In FIG. 16B, an example sentence "I never go ..." using "never" 61j as an antonym of "often" and an example sentence "I have never been to. .. ”is displayed (61h).

<Fifth Embodiment> As another embodiment of the present invention, a sentence in which the selected word in the displayed example sentence is replaced with a word having a synonymous meaning may be created. Good. In this case, the jump key 11p and the "synonymous" key 11n shown in FIG. 1 are used. Similar to the search for the opposite word, the jump key 11p is used to select a desired word, and then the “synonym” key 11n is pressed to display an example sentence converted into a synonym. FIG. 17 shows an explanatory diagram of an embodiment for generating an example sentence of a synonym of the present invention. FIG. 17
(A) is the same example sentence as shown in FIG. 16 (a).

FIG. 17 (b) shows the state of FIG. 17 (a).
This is the display state 61e after pressing the "synonym" key 11n. Here, as a synonym for "often" 61b, "som
Two of "etimes" 61f and "rarely" 61g are displayed. In addition, Japanese translations corresponding to each synonym are also displayed. Also in this case, by using the synonym dictionary 15a, the corresponding word list 15c, and the grammar table 15d, the synonym dictionary 15a shown in FIG.
The same process as the above may be performed. According to this, a sentence using a synonym can be displayed by a simple operation similarly to the opposite word.

[0076]

As described above, according to the present invention, importance is assigned to a word in advance, and voice recognition, translation, and display processing are sequentially performed in accordance with the importance, so that it is possible to quickly perform a dialogue in an actual dialogue scene. It is possible to provide an electronic translation device capable of analogizing or understanding the content spoken by the other party and having a smooth conversation.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram of an embodiment of an electronic translation device of the present invention.

FIG. 2 is a configuration block diagram of an embodiment of an electronic translation apparatus of the present invention.

FIG. 3 is an explanatory diagram of an embodiment of a voice recognition table of the present invention.

FIG. 4 is an explanatory diagram of an embodiment of a grammar table of the present invention.

FIG. 5 is an explanatory diagram of various information used in the present invention.

FIG. 6 is an explanatory diagram of contents displayed on the display unit of the present invention.

FIG. 7 is an operation flowchart of an embodiment of the electronic translation apparatus of the present invention.

FIG. 8 is an explanatory diagram of contents displayed on the display unit of the present invention.

FIG. 9 is an explanatory diagram of contents displayed on the display unit of the present invention.

FIG. 10 is an explanatory diagram of contents displayed on the display unit of the present invention.

FIG. 11 is an explanatory diagram of an embodiment of an answer correspondence table of the present invention.

FIG. 12 is an operation flowchart of an embodiment of the answer display processing of the present invention.

FIG. 13 is an explanatory diagram of a display example of a sub screen according to the present invention.

FIG. 14 is an explanatory diagram of an example of generation of a sentence using an antonym of the present invention.

FIG. 15 is an operation flowchart of an opposite word selection process of the present invention.

FIG. 16 is an explanatory diagram of another embodiment for generating an example sentence of the opposite word of the present invention.

FIG. 17 is an explanatory diagram of an example of generating an example sentence of a synonym of the present invention.

[Explanation of symbols]

11a cabinet 11b display section 11d English-Japanese key 11e Japanese-English key 11f voice recognition key 11g Search key 11i subkey 11j Return key 11k answer key 11l switch key 11n Synonym key 11o Opposite key 11p jump key 11s microphone 12a display device (LCD) 12b microcomputer 12c Input device (key) 12d input device (microphone) 12h ROM 12i RAM 13a Part of speech data 13b pronunciation data 13c Importance data 15a Various dictionaries 15b voice recognition table 15c Corresponding word list 15d grammar table 15e Answer correspondence table 16a Display example in the middle of voice recognition and translation 16b Display example when voice recognition and translation are completed 31a, 31c, 31e Display example of question text 31b, 31d, 31f Display example of answer sentence 51a Example of cursor display of jump candidate 51b Display example after antonym search 51c, 51e cursor 51d antonym icon 61a Cursor display example of jump candidate 61b, 61f, 61g, 61i, 61j cursors 61c antonym icon 61d Synonyms icon 61e Display example after synonym search 61h Display example after antonym search

─────────────────────────────────────────────────── ─── Continued Front Page (51) Int.Cl. ⁷ Identification Code FI Theme Coat (Reference) G10L 3/00 537C

Claims (3)

[Claims] 1. A voice input section, a voice recognition table in which at least words, pronunciation data, and importance information are stored in association with each other, a grammar table for the language of the input voice, and the input voice is digitalized. A voice data storage unit that has been stored in a converted form, a search unit that searches the pronunciation data that can match the voice data stored in the voice data storage unit, and selects a word candidate corresponding to the searched pronunciation data;
A recognition unit that determines a word by referring to the grammar table for the selected word candidate, a display unit that displays the determined word, a selection process of the word candidate, a word determination process, and a word display process, A control unit that causes the groups of the importance information to be sequentially performed in descending order of importance information, and that causes the sentence in the language of the input voice to be displayed on the display unit step by step according to the importance. An electronic translation device characterized by being provided. 2. A translation dictionary and a translation unit that translates the word determined by the recognition unit into a predetermined other language,
The control unit causes the translation unit to perform a translation process on a word determined by the recognition unit for each group of the importance information and in descending order of importance information. Electronic translation device. 3. The voice recognition table is stored in a rewritable memory, and usage frequency information used for searching or recognizing a word in the voice recognition table is stored in the memory for each word, and the control unit. The electronic translation device according to claim 1 or 2, wherein the importance level information is changed based on the usage frequency information.