JPH0155507B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an electronic translator that performs language translation between different languages, and particularly to an electronic translator that can output an input language or a translated result as a voice.

2. Description of the Related Art Conventionally, electronic translators have been provided that can output externally input sentences or translated results as audio.
For example, example sentences and information for outputting them aloud are simply stored in memory for a number of example sentences in advance, a desired example sentence is retrieved by key operation, and audio notification is performed using the corresponding audio information. There are things, etc.

This method not only cannot store a large number of example sentences in a small memory capacity, but also has the disadvantage that example sentences for which audio information is not stored in the memory cannot be output as audio.

The present invention has been made based on the above-mentioned drawbacks of conventional electronic translators, and particularly aims to provide an electronic translator that can easily notify voice information even for example sentences and words for which voice information is not stored in memory. This is what I did.

The features of the electronic translator according to the present invention are listed below.

(1) It is possible to store in memory audio information corresponding to various characters of sentences and words that are stored in advance in memory, and to be able to pronounce the spelling of sentences and words that cannot be completely output aloud.

According to this method, voice notification of sentences and words can be made even with a small amount of voice information. Furthermore, when transcribing the translation result, it can be easily done while listening to the audio output.

(2) Also, based on the audio information corresponding to the various characters, when a character is input by operating a character key, the audio corresponding to the character is output.

If voice notification is required in an electronic translator, key operations can be confirmed not only for the translated language, that is, a foreign language, but also when inputting keys in the native language. Furthermore, especially in the case of Japanese, the words are pronounced as they are, making it easy to confirm. It can also be used for blind people.

(3) It is possible to output words and sentences entered using the keys character by character or repeatedly with some degree of connection.

Generally, when inputting kana characters, voiced or semi-voiced characters are input using a combination of a clear sound key and a voiced sound key or a semi-voiced sound key, but when making a voice notification, they can be pronounced normally.

The same applies to German umlauts, etc.

Also, the Japanese words te, ni, wo, ha, no, ga,
When pronouncing particles such as ``to'', ``he'', ``yori'', and ``kara'', it is normal and natural to output the words following the previous word. According to the present invention, when a sentence that has been stored in a memory in advance is to be outputted as a voice, it is possible to continuously output particles and words with particles attached thereto. This also applies to sentences input using the keys.

Among the particles mentioned above, in particular, "ha" is pronounced as "wa" and "he" is pronounced as "e", but even in such cases, according to the present invention, audio output is performed with more natural pronunciation. Furthermore, for example sentences pre-stored in memory or sentences input using the keypad, combinations of characters with many connections are converted into compressed codes and then output as audio. Since audio information such as "Deska" is stored in memory in advance in correspondence with a compressed code, it is possible to output well-connected words as audio with a small memory capacity.

EMBODIMENT OF THE INVENTION Hereinafter, the electronic translation machine of this invention will be explained in detail based on an Example.

FIG. 1 is a plan view showing the appearance of an embodiment of the present invention.

In the figure, 1 is a key section, and the key section 1 is composed of keys for inputting alphabets, katakana, symbols, etc., and function keys for executing various functions such as translation. Further, 2 is a display section,
The display section 2 displays a character display section 3, a language display section 4 that displays symbols indicating the currently designated native language (input language) and foreign language (output language), and various symbols indicating setting status, etc. It consists of various symbol display sections 5. Further, 17 is a mode changeover and power switch.

To call up a desired example sentence and output its translation, proceed as follows.

Set the switch 17 to sentence mode and operate the desired category key. The category keys are keys arranged in an L-shape on the key part 1 ("A" to "O" and "Space" keys), and if the category is "Meals", the "B" key is pressed. By pressing this key several times, example sentences belonging to the "meal" category are displayed one after another. When the desired example sentence is displayed, by pressing the "translation" key, the translated sentence is displayed and the corresponding audio is output. If you want to do this about words, switch 1.
7 is set to word mode. Furthermore, if the switch 17 is set to dictionary mode, words or sentences can be input and translated using the character keys. In this case, the category key operates as a character key. Also,
Each time a character key is input, the character is pronounced, and the spelling of the word or sentence input by operating the voice key is continuously uttered sound by sound.

FIG. 2 is a block diagram showing the system configuration of the same embodiment.

Reference numeral 2 denotes the display section shown in FIG. 1, and character display, lighting and non-lighting of each symbol, etc. on the display section 2 are controlled by an output signal from the display drive control IC 6. 7 is a key matrix circuit constituting the key unit 1 shown in FIG. 1;
and 11 are ROM modules, respectively, which accommodate linguistic and audio information such as sentences and words in the electronic translator.

These language ROMs 9, 10 and 11 are connected to a one-chip microcomputer 8 by an address bus 12 and a data bus 13.
By selecting a predetermined ROM using the chip select signals CE ₁ , CE ₂ , CE ₃ , CE ₅ , CE ₆ , and CE ₇ , various information can be read from the selected ROM. Note that _CE4 is a chip select signal of the display drive control IC6. 9 has four inside it
It is a module with ROM, and 10 is a module that stores only language information without storing voice information.
A module 11 having a ROM is a module having a built-in ROM for storing voice information or language information. 14 is a latch circuit for outputting an audio signal or a key strobe signal;
Latch the data of 3. Reference numeral 15 denotes a drive circuit for D/A converting the output of the latch circuit 14 and driving the speaker 16. In the case of key reading, the latch output of 14 is outputted to the key matrix circuit 7 as a key strobe signal. The latch outputs are also used to output the CE ₅ , CE ₆ , and CE ₇ chip select signals. The signal PU is input to a gate for switching ON/OFF of the power supply of the drive circuit 15 and for generating a chip select signal. Since the contents of the latch circuit 14 change during audio output, the chip select signals CE ₅ , CE ₆ and CE ₇ obtained in this way are used as chip select signals for the language ROM. For example, module 9 has a Japanese language ROM, an English language ROM,
It has built-in English voice ROM, Alphabet and kana voice ROM. module 10
For example, the module 11 has a built-in German language ROM and a German voice ROM. FIG. 3 is a diagram showing the ROM format within the module.

Language ROM is broadly divided into text data and word data compression tables.

Words are not stored in their original spellings, but are compressed and stored in the language ROM.
For example, in the case of Japanese ROM, the spelling ``Giya'' (character string) is converted to one compression code (1-byte code) CC <sub>1</sub> , and the spelling ``Giya'' is converted to the compression code CC ₂ and stored. has been done. Similarly, some frequently occurring spellings of English words are selected, and each of the selected spellings is converted into a corresponding compressed code and stored. and,
The correspondence between each of the selected spellings and the compression code is stored in the compression table area. More specifically, when matching input characters with words stored in the language ROM, the input words are converted into a code string including a compressed code based on the compression table, and then the words are stored in the language ROM. The structure is such that the words (stored in the form of a code string including compressed codes) are sequentially compared. The compression table is also used when restoring words stored in the language ROM to their original spelling in order to display them.

The compression table differs depending on each language, and is determined to give the highest compression rate for each language.

In addition, words and sentences are classified into categories and arranged in order for each category. On the other hand audio
As shown in FIG. 3, the ROM is roughly divided into a character table, a word table, a sentence table, and an area for storing audio data. The character table is a table for searching audio data in order to perform the pronunciation corresponding to each character and compression code such as alphabets and 50 sounds, and the word table is a table for searching audio data in order to perform the pronunciation corresponding to each word. The table and the sentence table are tables for pronunciation corresponding to each example sentence.
These tables store the presence or absence of audio information for each character, each word, or each example sentence, and if so, address information of the corresponding audio data is stored. In addition, the identification code is
Codes are stored in advance to distinguish whether the ROM is a language ROM or a voice ROM and to identify the type of language.

FIG. 4 is a flowchart showing the operation of an electronic translator according to an embodiment of the present invention.

The operation will be explained below.

When the power is turned on, first press the mode switch 16.
The state of the key is read, that is, whether it is in sentence mode or word mode, and then the state of the key is read. If a desired category key is currently pressed, example sentences or words belonging to that category are read from the memory (n ₅ →n ₇ →n ₈ ). Also, the above-mentioned audio ROM is searched to see if the example sentence or word read at this time has audio information. That is, first, it is determined whether or not there is a native language audio ROM in the installed module. If it exists, the aforementioned sentence table or word table is searched to determine whether or not voice information is stored. The read example sentence or word is displayed. If you then press the generation key, the contents will be output audibly. If the voice information is stored in the ROM, complete voice output is performed (n ₁₀ →n ₁₁ ). That is, audio information is read out and audio output is performed based on the sentence table or word table described above. If voice information is not stored but voice information for characters is stored, the read sentence or word is pronounced one by one (n ₁₂ →n ₁₃ ).
Also, if no audio information is stored, a specific sound such as “pipipi” will be output (n ₁₂ →
_n14 ). Next, when the same category key is pressed, the next example sentence or word in the same category is read out from the memory (n ₇ →n ₁₅ →n ₈ ). In this way, if you call up the desired sentence or word and press the translation key, the translated sentence or translated word will be read out from the ROM specified as a foreign language and displayed (n ₃ → n ₁₆ →
n ₁₇ ), and the contents are also output as audio. Audio notification is performed in the same manner as described above. Furthermore, even if audio information is stored, if the audio information for each character is stored, the spelling will be output aloud each time the spelling key is pressed (n ₁₈ → n ₁₉ →
_n20 ). When a character is input using a character key, the character is input on the buffer and its contents are displayed, and if audio information corresponding to the character is stored, the character is pronounced at the time the key is pressed. Furthermore, the input character string is converted into a compressed code based on the aforementioned compression table. Therefore, if you press the voice key after entering the desired word or sentence in this way, the voice will be output based on the compressed code, such as "Giya", "Giyu", "Giyo", etc. Text is output aloud in the same way as it would normally be read. For example, when inputting the word "denshi", the sound produced when pressing a letter key is not pronounced when the voiced sound key is pressed;
'. For example, when inputting the word "Gi-young", it will be pronounced "ki-yanku", but if the word is converted into a compressed code and then announced by voice, "gi-young" will be pronounced as "ki-giyanku-gu". FIG. 5 is a flowchart showing the control of single tone generation described above. The explanation will be given below.
The content to be output as audio is stored in the display buffer, and P and g are its pointers, and B(P)
The expression means the content of the Pth character of the buffer. The overall process involves compressing and encoding the content input to the buffer based on a compression table, and then making a voice notification, and determining that only one character is sandwiched between spaces.
Treat it as a particle and pronounce it consecutively with the previous word, and if that letter is “ha” then “wa”,
If it is “he”, it is pronounced as “e”.
First, determine the character code from the first digit of the batshua,
The digit up to the space code or end code is compressed as one segment and converted into a compression code based on the compression table (n ₁ →n ₂ →n ₃ →n ₄ →n ₅ →).
For example, when the content shown in FIG. 6 is input as the content of a buffer, the word is separated by "anata", but there is no compressed code for this, and the sound is pronounced one by one (n ₆ ). Thereafter, it is determined whether the next character is a single character or not, and if it is a single character "ha" as in the example (that is, if the next second digit is a space), that character is pronounced.
In this case, “ha” is pronounced as “wa” (n ₉
→n ₁₀ →n ₁₁ ). Also, if it is "he", it is pronounced as "e" (n ₁₂ →n ₁₃ ), and other chords are pronounced as they are (n ₁₄ ). It is determined whether that word or the next word is a single character (n ₁₅ →
n ₇ → n ₈ ). In the example of FIG. 6, it is not a single character, but is compressed in the same way as described above (n ₁₆ →n ₁₇ →n ₁₈ →n ₁₉ ). As shown in Figure 6, "te" and the voiced sound code are combined into the code "de", and "ma", "su", and "ka" are converted into one compressed code called "maska" and output as audio. . In this case, the signal is output after waiting for several meters to several tens of milliseconds (n ₁₈ →n ₂₀ →n ₂₁ ). Therefore, anatawa and dekimasuka are separated to a certain extent and can be heard naturally. In the example explained above, if a character exists between two spaces, it is pronounced following the utterance of the previous word, and is also pronounced as a particle, but in order to reduce the number of exceptions, for example, noun It is possible to treat the next character as a particle. In other words, whether or not a word can be a noun is stored in a memory for each word, and the processing of a character following the target word can be determined depending on whether or not the target word is a noun. Also, as particles “yori” and “kara”
There are some characters that are not a single character, such as, but in order to process these without exception, you need to use a space and a space.
This can be done by further determining whether these particles exist between the sentences. Another problem is how to pronounce the long sound symbol. For example, ``kuko'' or ``row'' may be expressed as ``kuko'' or ``giyo'' and entered using the keypad. As mentioned above, a combination of characters including these long sound symbols may be used as a compressed code, and audio information may be stored in each code, but in order to reduce the memory capacity, it is possible to configure the code as described below. Can be done. In other words, the long sound symbol code is converted to the vowel code of Aiueo so that ``Ku-ko-'' is pronounced as ``Ku-koo'' and ``Gi-yo'' is pronounced as ``Gi-yo.'' If applied to what was explained in Figure 5, the content of buffer B is identified character by character, and if it is a long sound symbol, it is converted into the code of the vowel that should be added to the previous character, and then the above-mentioned Compression is performed based on the compression table as shown in FIG. 7. The vowel code to be added to the previous character can be easily selected by setting a code table for kana characters in order of kana characters.

As described above, according to the electronic translator of the present invention,
The storage capacity of audio information can be reduced, and with this audio information, verbal information such as documents and words can be effectively output as audio, and confirmation by audio can be performed very easily.

In particular, it is possible to output documents, words, etc. in an easy-to-understand manner with a certain degree of connection, and even when complete audio output is not possible, the spelling can be outputted, making it easy to read the contents of the audio output. can be grasped.

[Brief explanation of drawings]

Fig. 1 is a plan view showing the external appearance of an embodiment of the electronic translator of the present invention, Fig. 2 is a block diagram showing the system configuration of the embodiment, and Fig. 3 is a ROM format in the module in the block diagram. Figure 4 showing
The figure is a flowchart showing the operation of the same embodiment.
The figure is a flowchart showing control of single tone generation in the same embodiment, and FIGS. 6 and 7 are diagrams for explaining the same. In the figure, 1: key section, 2: display section, 6: display drive control IC, 7: key matrix circuit, 9-1
1: ROM module, 14: Latch circuit, 1
5: Drive circuit, 16: Speaker, 17: Mode switching/power switch.

Claims (1)

[Claims] 1. A first memory that stores a plurality of linguistic information such as documents and words that are retrieved when performing translation, and a second memory that stores audio information corresponding to the contents of the first memory. An electronic translator capable of outputting voice along with translation by reading out the contents of the first memory and second memory in correspondence, the contents of the first memory being read as the voice information. a second memory having a first voice information storage section for vocally expressing linguistic information such as a document or a word, and a second voice information storage section for pronouncing spellings constituting the linguistic information; Discrimination means for discriminating whether or not linguistic information such as documents and words to be outputted as voice is stored in the first voice information storage section of the second memory; means for reading the voice information corresponding to the language information when it is determined that the language information is stored in the first voice information storage section of the second memory; is determined,
means for reading audio information corresponding to spellings constituting linguistic information such as documents and words from a second audio information storage section of a second memory; An electronic translator characterized by comprising: audio output control means for outputting the output audio information singly or in a combination thereof.