JPH0934895A - Translation system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To simply configure the translation system among multi-languages at a low cost by translating a prescribed language inputted in each translation device into a common language and sending the translation result into the common language to other translation device. SOLUTION: Upon receipt of a voice in a language A, a voice recognition section 201 recognizes the voice and provides an output of the voice recognition result to a translation section 202. Then the translation section 202 translates the voice recognition result for the voice in the language A into a language C used in common for all voice translation devices 200, 210, and provides the result of translation to a MODEM 205. Furthermore, the MODEM 205 sends the translation result into the common language C to other voice translation device 210 other than the voice translation device 200 via a radio channel or a wired line. Moreover, the MODEM 205 receives the translation result into the common language C and gives it to the translation section 204. Then the translation section 204 translates the translation result into the common language from the MODEM 205 into the language A and provides an output of the translation result to a voice synthesis section 203.

Description

Detailed Description of the Invention

[0001]

TECHNICAL FIELD The present invention relates to a translation system. In particular, the present invention relates to a translation system in which a predetermined language is translated into a certain common language (common language) so that translation between many languages can be easily performed.

[0002]

2. Description of the Related Art FIG. 7 shows an example of the configuration of a conventional translation system (speech translation system). In this translation system, the input voice is translated into another language, and the translation result is output as a synthetic voice.

That is, this translation system is composed of a speech translator 100 for translating a language A into a language B and, conversely, a speech translator 110 for translating a language B into a language A. The voice translator 100 or 110 is composed of a voice recognition unit 201, a translation unit 102, and a voice synthesis unit 213, or a voice recognition unit 211, a translation unit 112, and a voice synthesis unit 203, respectively. In the following, a translator that translates a language X into a language Y will be referred to as an X / Y translator.

Speech (speech of language A) uttered by a user who speaks language A (speech of language A) is input to speech recognition unit 201 and is recognized there. The voice recognition result by the voice recognition unit 201 is supplied to the translation unit 102, where it is translated into the language B. The translation result into the language B is supplied to the speech synthesis unit 213, where a synthesized sound corresponding to the translation result into the language B is generated and output. The user of the language B understands the utterance content of the user of the language A by listening to the synthesized sound of the language B output from the speech synthesis unit 213 as described above.

On the other hand, the voice uttered by the user of language B (the voice of language B) is input to the voice recognition unit 211,
There it is voice recognized. The voice recognition result by the voice recognition unit 211 is supplied to the translation unit 112, where it is translated into the language A. The translation result into the language A is supplied to the speech synthesis unit 203, where a synthesized voice corresponding to the translation result into the language A is generated and output. The user of the language A understands the utterance content of the user of the language B by listening to the synthesized sound of the language A output from the speech synthesis unit 203 as described above.

The translation system shown in FIG. 7 integrates the speech translators 100 and 110 so that it can translate bidirectionally from languages A to B and from B to A. However, they can be configured as separate devices. Further, the translation result can be displayed on a CRT, LCD or the like in addition to being output as a synthetic sound.

[0007]

When a user of language A and a user of language B have a conversation, for example, the A / B speech translator 100 used by the user of language A is used.
Focusing on, the translation unit 10 that translates from language A to B
2 is enough. However, in addition, if a user of another language joins and conducts a conversation,
There is a problem that the speech translation unit 100 becomes large in size because translation units are required for the number of combinations with all other languages.

Further, in this case, when translating, rules and dictionaries for translating each language are required, but it is not easy to create this for many languages, and it is enormous. Since it takes man-hours, there is a problem that the cost of the system increases.

The present invention has been made in view of such a situation, and makes it possible to easily perform translation between multiple languages.

[0010]

A translation system according to the present invention is a translation system comprising a plurality of translators, and the translator translates an input predetermined language into a certain common language. Means and a transmitting means for transmitting the translation result to the common language by the first translation means to another translator, and the translation result to the common language transmitted from the other translator to a predetermined language It is characterized in that it has a second translation means for performing the above, and an output means for outputting a translation result into a predetermined language by the second translation means.

In this translation system, when the predetermined language is a common language, the translator is configured by excluding the first and second translation means, and the transmission means receives the input predetermined language. It is possible to cause another output device to transmit the translation result as it is, and to the output means to output the translation result to the common language transmitted from the other translation device as it is. The common language is
It can be natural language. Also, the common language can be English. The translator can further include a voice recognition unit that recognizes the input voice by voice, and in this case, the first translation unit can translate the voice recognition result by the voice recognition unit.

When there are two common languages, the first
Translation means translates a predetermined language into one common language, and the second translation means translates one common language into a predetermined language. The translators are connected to each other to form a first network. The first translation means translates the predetermined language into the other common language, and the second translation means translates the other common language into the predetermined language. First and second when configuring the network of
In the network, the first translation means translates one common language into the other common language, and the second translation means uses the translator that translates the other common language into one common language as a gateway. Can be connected.

The translation system having the above-mentioned configuration is provided with a plurality of translators. In each translator, the first translation means translates the input predetermined language into a certain common language, and the transmission means. Is adapted to transmit the result of translation into the common language by the first translation means to another translator. The second translation means translates the translation result to the common language transmitted from another translator into a predetermined language, and the output means
The result of translation into a predetermined language by the second translation means is output.

[0014]

DESCRIPTION OF THE PREFERRED EMBODIMENTS FIG. 1 shows the configuration of an embodiment of a speech translation system to which the present invention is applied. In the figure, the same reference numerals are given to portions corresponding to the case in FIG.

This speech translation system includes a speech translator 20.
It consists of two voice translators, 0 and 210,
Communication is possible between the two parties.

The voice translator 200 includes a voice recognition unit 201,
The translation unit 202, the voice synthesis unit 203, the translation unit 204, and the modem 205 are included. Voice recognition unit 201
When the voice of the language A is input thereto, the voice recognizes the voice and outputs the voice recognition result to the translation unit 202. In the voice recognition unit 201 (the same applies to the voice recognition unit described later), for example, HMM (Hidd
en Markov Model) speech recognition is performed according to the method. However, the voice recognition algorithm used for the voice recognition processing is not limited to the HMM method.

The translation unit 202 outputs the speech recognition result of the language A speech (language A speech recognition result) output from the speech recognition unit 201 to all speech translators (in this embodiment,
A language commonly used by the voice translators 200 and 210) (herein referred to as a common language) C,
The translation result is output to the modem 205. In addition, in this embodiment, the common language is a natural language. Further, the translation algorithm in the translation unit 202 (the same applies to the translation unit described later) is not particularly limited. However, it is preferable to adopt a translation algorithm corresponding to the characteristics (features) of the language A and the common language C.

The modem 205 transmits the translation result to the common language C supplied from the translation unit 202 according to a predetermined protocol via a wireless line or a wired line to a voice translator (other than the voice translator 200). In the embodiment of FIG. 1, it is adapted to send to a speech translator 210). The modem 205 is also configured to receive the translation result into the common language C transmitted from the voice translator 210 (modem 215), which is another voice translator, and supply it to the translation unit 204.

The translation unit 204 is adapted to translate the translation result from the modem 205 into the common language C into the language A, and output the translation result to the voice synthesis unit 203.
The speech synthesis unit 203 generates a synthesized sound corresponding to the translation result into the language A supplied from the translation unit 204, and the speaker 3
08 (FIG. 2).

The speech translator 210 (similarly to a speech translator to be described later) is also constructed similarly to the speech translator 200. That is, the voice translator 210 includes the voice recognition unit 20.
1, translation unit 202, speech synthesis unit 203, translation unit 204,
Alternatively, it is composed of a voice recognition unit 211, a translation unit 212, a voice synthesis unit 213, a translation unit 214, or a modem 215 that are configured similarly to the modem 205. However, the voice recognition unit 211 performs voice recognition of the language B voice, and the translation unit 212 translates the voice recognition result of the language B into the common language C. Further, the translation unit 214 is adapted to translate the common language C into the language B, and the speech synthesis unit 213 is adapted to output a synthesized voice corresponding to the translation result into the language B.

According to the translation system configured as described above, the user of the language A or B can have a conversation by using the voice translator 200 or 210, respectively.
That is, when the user of the language A inputs a voice into the voice recognition unit 201, the voice is recognized. Then, the speech recognition result of the language A obtained as a result is supplied to the translation unit 202, where it is translated into the common language C. The translation result is supplied to the modem 205, and from there, transmitted to the speech translator 210 via a communication line (not shown).

In the speech translator 210, the speech translator 200
The translation result into the common language C transmitted from (modem 205) is received by the modem 215, and the translation unit 21
4 is supplied. The translation unit 214 translates the translation result into the common language C into the language B, and the translation result is output to the speech synthesis unit 213. The speech synthesis unit 213 generates and outputs a synthetic sound corresponding to the result of translation into the language B. It is also possible to output the translation result by displaying it on a monitor such as a CRT or an LCD, instead of outputting the synthesized sound.

On the other hand, the voice uttered by the user of language B is input to the voice recognition unit 211 and is recognized there. The speech recognition result of this language B is supplied to the translation unit 212,
It is translated into the common language C and output to the modem 215.
Upon receiving the translation result into the common language C, the modem 215 transmits it to the voice translator 200.

In the voice translator 200, the voice translator 210
The result of translation into the common language C transmitted from (modem 215) is received by the modem 205, and the translation unit 20
4 is supplied. The translation unit 204 translates the translation result into the common language C into the language A, and the translation result is output to the speech synthesis unit 203. The voice synthesis unit 203 generates and outputs a synthesized voice corresponding to the translation result into the language A.

By the way, FIG. 1 shows a block diagram in the case where the speech translator 200 (the same applies to the speech translator 210) is divided into functional blocks, but its hardware is actually For example, as shown in FIG. 2, a modem 205, a microphone 301, an A / D converter 302, a DS
P (Digital Signal Proccessor) 303, CPU (Cen
tral Processor Unit) 304, IC card 306, D
The A / A converter 307 and the speaker 308 are included.

The voice recognition unit 201 is provided with the microphone 30.
1, A / D converter 302, DSP 303, IC card 3
06, and the translation units 202 and 204 have the CPU 3
04 and IC card 306. Further, the voice synthesis unit 203 includes a DSP 303, an IC card 306, a D /
It corresponds to the A converter 307 and the speaker 308.

In the microphone 301, the voice of the language A input therein is converted into a voice signal as an electric signal, and A
It is output to the / D converter 302. In the A / D converter 302, the audio signal from the microphone 301 is A / D converted (quantized and converted into a discrete time), and the resulting digital audio signal is output to the DSP 303. DSP
Upon receiving the voice signal from the A / D converter 302, the 303 executes a program (voice recognition program) for recognizing the voice of the language A, and thereby recognizes the voice input to the microphone 301. The recognition result is supplied to the CPU 304 via the bus. When the CPU 304 receives the speech recognition result of the language A, it executes a program (translation program) for translating it into the common language C,
As a result, the speech recognition result of the language A is translated into the common language C. The translation result is output to the modem 205 and transmitted.

On the other hand, the result of translation into the common language C received by the modem 205 is output to the CPU 304 and the CPU 3
Upon receiving the translation result for the common language C from the modem 205, 04 executes a program (translation program) for translating it into the language A, and thereby translates the translation result for the common language C into the language A. . The translation result is supplied to the DSP 303 via the bus. DSP303
Upon receiving the translation result into the language A, executes a program (speech synthesis program) for generating a synthesized voice corresponding to the translation result, and D / S the resultant synthesized voice data.
Output to the A converter 307. In the D / A converter 307,
The synthesized voice data is D / A converted and supplied to the speaker 308. As a result, the language A is output from the speaker 308.
A synthesized voice corresponding to the result of translation to is output.

A translation program executed by the CPU 304 and data necessary for translation (eg, rules for translation and a dictionary) are stored in the IC card 306. Card 30
Read necessary data (including program) from 6,
The processing as described above is performed. Similarly, the voice recognition program executed by the DSP 303, the voice synthesis program, and the data necessary for them are also stored in the IC.
The data is stored in the card 306, and the DSP 303 reads necessary data from the IC card 306 via the bus and performs the above-described processing. The IC card 306 is attachable to and detachable from the voice translator 200.

Therefore, the speech translator 200 can be made into a speech translator which translates a language other than the language A into the common language C by changing the IC card mounted therein. That is, a voice translator that translates various languages into the common language C
The hardware can be made common, and the configuration can be achieved only by changing the IC card mounted therein (by changing the software). Therefore, the flexibility of the speech translator and the expandability of the system can be improved, and the manufacturing cost thereof can be reduced.

Next, FIG. 3 shows a configuration example of a speech translation system in which three or more speech translators are connected to form a network. Speech translator 200 or 21
0 is adapted to perform bidirectional translation between the language A or B and the common language C, respectively, as described in FIG. The speech translators 401 and 402 are configured similarly to the speech translator 200 (speech translator 210) described with reference to FIG. However, the voice translator 401 uses the common language C
Between the languages D and C (common language C to common language C), and the speech translator 402 bidirectionally translates between the language D and the common language C. .

In this case, the language of the translation result transmitted from the voice translator 200 (similarly to other voice translators) to the other voice translator is the common language C, and the other voice translations are included therein. Since the language of the translation result transmitted from the device is also the common language C, the voice translator 200 can be directly connected to the network without changing the configuration as in the case shown in FIG. That is, in this case, the speech translator 200 can translate the languages C and D into the language A in addition to the language B without changing the configuration.

However, as described above, the speech translator 200
When connecting to another voice translator, the communication protocol is determined in advance and the modem 205
It is necessary to make the input and output of the above into a form that can be networked.

In FIG. 3, a network is constructed by connecting four speech translators,
The number of voice translators to be connected is not particularly limited.

Hereinafter, a network configured to exchange the common language C as shown in FIG. 3 will be referred to as a network of the common language C as appropriate.

By the way, the speech translator 401 shown in FIG.
As described above, the translation between the common languages C is performed, but the translation of the common language C into the common language C has the same result whether it is performed or not. Therefore, the voice translator 401 shown in FIG.
It is not necessary to provide blocks corresponding to 0 translation units 202 and 204. Therefore, as shown in FIG. 4, a voice translator 401 for translating between common languages C, a voice recognition unit 501 that recognizes a voice of a common language C, and a voice synthesis unit 401 that generates a synthesized sound of a common language C. , And a common language C for transmitting and receiving a modem 505. That is,
The voice translator 401 can be configured easily and at low cost as compared with other voice translators.

As a result, as shown in FIG. 3, when a plurality of speech translators are connected to form a network,
Of the languages used by users, it is desirable that the most common language is the common language. In this case, since the speech translator used by the user of such a language can be manufactured at low cost as described above, the cost of the entire system can be reduced.

The above effects can be obtained by selecting the common language from natural languages. Also, some conventional translation methods use so-called intermediate languages, but it is difficult to determine what kind of intermediate expression a certain semantic expression is, and this method is used for various languages. Is practically impossible. In this embodiment, since the natural language is adopted as the common language, the above problems can be avoided. Note that, for example, English or the like can be adopted as the common language. This is because it can be expected that there are dictionaries and rules for translating into English for many languages, which makes it possible to reduce the man-hours for creating the dictionaries.

By the way, there may be a language F that is easy to translate into a certain language but difficult to translate into the common language C. That is, when, for example, English is adopted as the common language C, a dictionary or rule for translating into English has not been prepared yet, but a dictionary or rule for translating into Chinese, for example, is prepared. F may be present. In this case, Chinese is, so to speak, a second common language (sub-common language) E, and a network of this second common language E is configured. As shown in FIG. By connecting a voice translator that performs bidirectional translation between C and the second common language E as a gateway, voice translation can be performed without preparing a dictionary or rules for translating the language F into the common language C. It is possible to configure the system.

That is, the network of the common language C is constructed as described with reference to FIG. 3, and further, the network of the second common language E is constructed in the same manner. Note that FIG.
In the above, speech translators 200 and 210 for bidirectionally translating each of the languages A, B or C and the common language C,
Alternatively, 401 constitutes a network of the common language C, and a voice translator 404 which performs bidirectional translation between the language F and the second common language E constitutes a network (subnetwork) of the second common language E. There is.

Then, these networks are connected via a voice translator 403 which performs bidirectional translation between the common language C and the second common language E. In this case, the voice translator 4
03 translates the common language C transmitted / received in the network of the common language C into the second common language E, outputs it to the network of the second common language E, and in the network of the second common language E. The second common language E that is transmitted and received is translated into the common language C and output to the network of the common language C.

Therefore, the common language C output from the speech translators 200, 210 and 401 is the speech translator 40.
Is translated into the second common language E by way of 3,
The speech translator 404 is also supplied to the speech translator 404.
The second common language E output from the speech translator 403 is
Is translated into the common language C by way of and is supplied to the voice translators 200, 210 and 401, so that the users of the languages A, B, C and F can talk with each other.

The connection between the networks as shown in FIG. 5 can be expanded. That is, the third
Common language G is further set to configure a network of the third common language G, and the network of the second common language E is connected to the second common language E and the third common language G. A voice translator for bidirectional translation can be connected as a gateway. In addition, the network of the third common language G and the network of the common language C can be connected by using a speech translator that performs bidirectional translation between the common language C and the third common language G as a gateway.

Next, FIG. 6 shows a configuration example of the speech translator 403 of FIG. The voice translator 403 functions as a gateway as described above, but is also configured to be able to perform the same processing as other voice translators. That is, the speech translator 403 includes a speech recognition unit 201, a translation unit 202, a speech synthesis unit 203, a translation unit 204, or a speech recognition unit 601, a translation unit 602, which have the same configurations as the modem 205 described in FIG. Speech synthesizer 60
3, the translation unit 604, or the modem 605, switches SW1 and SW2, and a modem 606 are provided.

The voice recognition unit 601 performs voice recognition of the voice of the second common language E, and the translation unit 602 performs the voice recognition.
The speech recognition result of the second common language E is translated into the common language C. Further, the translation unit 604 is adapted to translate the common language C into the second common language E, and the speech synthesis unit 603 outputs a synthesized sound corresponding to the translation result into the second common language E. It is designed to be done.

Further, the switch SW1 is provided in the voice recognition section 6
The switch SW2 is provided between 01 and the translation unit 602 between the speech synthesis unit 603 and the translation unit 604, respectively. The modem 605 is connected to the common language C network, and the modem 606 is connected to the second common language E network.

The speech translator 403 configured as described above
, Is used like any other speech translator, the switch SW1 or SW2 is switched to the terminal b1 or b2 side. As a result, the output of the speech recognition unit 601 is supplied to the translation unit 602 via the switch SW1, and the output of the translation unit 604 is supplied to the speech synthesis unit 603 via the switch SW2. Therefore, the voice translator 4
03 will operate as a speech translator that can be used by a user of language (second common language) E in the network of common language C.

Next, when the speech translator 403 is used as a gateway, the switch SW1 or SW2 is
It is switched to the terminal a1 or a2 side. This allows
The output of the modem 606 is supplied to the translation unit 602 via the switch SW1, and the output of the translation unit 604 is supplied to the modem 606 via the switch SW2.

As described above, the modem 605 is connected to the network of the common language C, and the modem 606 is the second
The common language C from the common language C network is connected to the modem 6 because it is connected to the common language E network.
05, translation unit 604, switch SW2, and modem 6
Through 06, it is translated into the second common language E and output to the network of the second common language E. The second common language E from the network of the second common language E is the modem 606, the switch SW1, the translation unit 60.
2 and via the modem 605, it is translated into the common language C and output to the network of the common language C. That is, the voice translator 403 operates as a gateway that connects the network of the common language C and the network of the second common language E.

The terminals a1 and b1 of the switch SW1 are connected and the terminals a2 and b2 of the switch SW2 are connected.
When connecting with, the speech translator 403 operates as a speech translator for translating between languages E on the network of the second common language E (this is the speech translator 403). Is the speech translator 40 shown in FIG.
This is because the configuration is the same as that of 1.).

The case where the present invention is applied to a voice translation system for translating a language input by voice into another language has been described above. It is also applicable to a translation system that translates.

[0052]

As described above, according to the translation system of the present invention, a plurality of translators are provided, and each translator has:
The input predetermined language is translated into a certain common language, and the result of translation into the common language is transmitted to another translator.
In addition, the translation result to the common language transmitted from another translator is translated to a predetermined language, and the translation result to the predetermined language is output. Therefore, since the common language communicates with other translators, a system for translating between multiple languages can be configured easily and at low cost.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a first embodiment of a speech translation system to which the present invention has been applied.

2 is a block diagram showing a hardware configuration of a speech translator 200 of FIG.

FIG. 3 is a block diagram showing a configuration of a second embodiment of a speech translation system to which the present invention has been applied.

4 is a block diagram showing a configuration example of a speech translator 401 of FIG.

FIG. 5 is a block diagram showing a configuration of a third embodiment of a speech translation system to which the present invention has been applied.

6 is a block diagram showing a configuration example of a speech translator 403 of FIG.

FIG. 7 is a block diagram showing a configuration of an example of a conventional speech translation system.

[Explanation of symbols]

 200 voice translator 201 voice recognition unit (voice recognition means) 202 translation unit (first translation means) 203 voice synthesis unit (output means) 204 translation unit (second translation means) 205 modem (transmission means) 210 voice translation Device 211 Speech recognition unit (speech recognition unit) 212 Translation unit (first translation unit) 213 Speech synthesis unit (output unit) 214 Translation unit (second translation unit) 215 Modem (transmission unit)

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Yasuhiko Kato 6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Tetsuya Kagami 6-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Masao Watanabe 6-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Makoto Akabane 6-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Incorporated (72) Inventor Yuko Tanaka 6-735 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation (72) Inventor Hiroshi Tsunoda 6-7-35 Kita-Shinagawa, Shinagawa-ku, Tokyo Sony Corporation Within

Claims (6)

[Claims] 1. A translation system comprising a plurality of translators, wherein the translator translates an input predetermined language into a certain common language.
Translation means, transmitting means for transmitting the translation result to the common language by the first translation means to the other translator, and translation result to the common language transmitted from the other translator. And a second translation means for translating into a predetermined language, and an output means for outputting a result of translation by the second translation means into the predetermined language. 2. When the predetermined language is the common language, the translator is configured without the first and second translation means, and the transmission means is the input predetermined language. 2. The translation according to claim 1, wherein the translation unit directly outputs the translation result to the common language, which is transmitted from the other translation unit, to the other translation unit. system. 3. The translation system according to claim 1, wherein the common language is a natural language. 4. The translation system according to claim 3, wherein the common language is English. 5. The translator further comprises voice recognition means for voice-recognizing the input voice, and the first translation means translates the voice recognition result by the voice recognition means. The translation system according to claim 1. 6. When there are two types of the common language, the first translation means translates the predetermined language into one of the common languages, and the second translation means includes one of the common languages. The translators that translate a language into the predetermined language are connected to each other to form a first network, and the first translation unit translates the predetermined language into the other common language, When the second translation means configures a second network by connecting the translators that translate the other common language into the predetermined language, the first and second networks include: The translator in which the first translation unit translates one of the common languages into the other common language, and the second translation unit translates the other common language into one of the common languages. Connected as a gateway Claim 1 characterized in that
The translation system described in.