JPH0230063B2 - - Google Patents

Description

[Detailed Description of the Invention] The present invention relates to an electronic translator, and an object of the present invention is to provide an electronic translator that has a relationship between the speaking speed of a source language and the speaking speed of a translated language. .

In this specification, translation refers to (1) converting the language of one country into the language of another country, (2) mutual conversion between the standard language and dialect of the same country, (3) calculations (general calculations and Japanese yen). It also refers to inputting the periodicity π and outputting the numerical value "3.14...").

The source language refers to the language and numerical value before translation, and the translated language refers to the language and numerical value after translation. For example, in an electronic translator set to translate Japanese into English, "good morning" is the original language and "good morning" is the translated language.

In addition, a translated phrase refers to a set of words, phrases, sentences, numerical values, calculation formulas, etc. in the source language to be translated, and a translated language refers to a set of words, phrases, sentences, numerical values, calculation formulas, etc. in the source language to be translated. Translation language words, phrases,
A set unit that is a collection of sentences, numbers, etc.

The electronic translator already proposed by the present inventor in Japanese Patent Application Laid-open No. 103765/1983 was constructed as shown in FIG. That is, the audio input signal of the original language applied from the microphone 1 is amplified by the input amplifier 2 to a level that allows speech recognition, and then sent to the audio analysis circuit 3.
The signal is subjected to spectrum analysis, voice recognition, and converted into a character signal, which is applied to an electronic translation circuit 4 set for Japanese and English, and is translated into a character signal of the translated language.

This translated language character signal output is converted into a voice signal by a voice synthesis circuit 5, and then output by an output amplifier 6.
After being applied to and amplified, the speaker 7 outputs the voice in the translated language.

In this case, by adjusting the output speed adjustment circuit 9, the output speed of the translated language sent from the speech synthesis circuit 5, that is, the speaking speed, can be changed to be faster or slower. Therefore, by changing the speech rate of the translated language of the electronic translator, the voice produced from the speaker 7 can be adjusted to be spoken quickly or slowly. Note that the character display circuit 8 is
First, the source language is displayed in characters, and then the translated language is displayed in characters.

However, since the above-mentioned electronic translator has such a configuration, it is not possible to change the speech rate of the translated language in accordance with the input speed (speech rate) of the source language.

In other words, with conventional electronic translation machines,
As shown in B, the source language input signal “Good morning”
The delivery time of the translation language output signal "Good morning" is (t ₂ - _{t 1} ), (t ₆ - _{t 5} ), and the delivery time of the translation language output signal "good morning" is (t ₄ - t ₃ ), (t ₈ - t ₇ ). Then, (t ₆ − t ₅ )≪(t ₂ −
t ₁ ), (t ₄ − _{t 3} )≒(t ₈ − _{t 7} ). ∴( _t6 − _t5 )≪( _t8 − _t7 )≒( _t4 − _t3 )≪
( _t2 − _t1 ).

In this way, (t ₄ -t ₃ ) and (t ₈ -t ₇ ) are determined by the output speed adjustment circuit 9 and have no relation to whether the input speed of the source language is fast or slow.

Therefore, regardless of whether the source language speaker pronounces it quickly or slowly, the translated language uttered by the electronic translator will be pronounced at a constant speed determined by the output speed adjustment circuit 9. There was nothing more than to be done. Therefore, when the source language speaker speaks quickly, the translated language also speaks quickly, and when the source language speaker speaks slowly, the translated language also speaks slowly. If the speech rate is not adjusted appropriately each time, there will be a large difference between the speaking speed of the source language and the translated language, giving an unnatural feeling. It made me feel even stronger. Therefore, such electronic translators have the drawback of not being able to communicate sufficiently and express emotions sufficiently.

The present invention eliminates such drawbacks,
The present invention provides an electronic translator having speed correlation between voice input and output.

An embodiment of the present invention will be described below with reference to FIGS. 3 and 4.

FIG. 3 is a block diagram of an electronic translator using the "speech rate matching method" which is an embodiment of the present invention, and FIG. 4 is a timing chart of the input and output of each part of the same machine.

In the figure, 11 is a microphone, which is connected to an input amplifier 12. In the input amplifier 12, the microphone 11
After amplifying the audio signal of the original language applied from the source to a level that allows audio analysis, the audio signal is sent to the audio analysis circuit 13. Here, the speech analysis circuit refers to a circuit that spectrally analyzes a speech signal, recognizes the speech, and converts it into a character signal. The character signal output from the speech analysis circuit 13 is applied to an electronic translation circuit 14 set to Japanese and English, and after being translated, it is sent to the speech synthesis circuit 15 as a character signal of the translated language.

The speech synthesis circuit 15 converts the character signal of the translation language into a speech signal and applies it to the output amplifier 16. The output amplifier 16 amplifies this audio signal and then sends it to the speaker 17. In this way, the speaker 17 produces sounds in the translated language.

During this time, the speech analysis circuit 13 calculates the input speed (utterance speed) of the speech input signal of the original language, and sends the calculated output to the control circuit 20 . This results in
The control circuit 20 controls the output speed adjustment circuit 19 so that the translated language is output from the speech synthesis circuit 15 at a speaking rate corresponding to the speaking rate of the source language.
Therefore, the speech rate of the translated language changes corresponding to the speech rate of the source language.

By doing this, if the source language speaks quickly, the translated language will speak at a correspondingly fast rate, and conversely, if the source language speaks slowly, the translated language will also speak at a correspondingly slow rate. , becomes a slow talker.

That is, in this embodiment, the speech analysis circuit 13 divides the number of characters (4) of the translated phrase "good morning" in the source language by the length of time required to apply the translated phrase,
The speech information of the original language is calculated, and the calculated output is applied to the control circuit 20. Next, control circuit 2
0 controls the delivery speed adjustment circuit 19 to match the utterance speed of the translated phrase ``good morning'' of the translated phrase ``good morning'' to the utterance speed of its source language using the calculation output according to the utterance speed. Control.

In this way, the speech rate of the translated language pronounced from the speaker 17 matches the speech rate of the source language applied to the microphone 11, so that more natural translation is possible.

Next, its operation will be explained.

As shown in FIG. 4A, when the original language "Good morning" is applied from the microphone 11 between time t ₁₁ and time t ₁₂ , after being amplified by the input amplifier 12, the original language voice "Good morning" is output. The signal is voice analysis circuit 1
3, it is converted into a character signal of the original language called "Ohayo", and is applied to the electronic translation circuit 14.
At the same time, the speech analysis circuit 13 uses the time required to apply the word to be translated ( _t12 - _t11 ) and the number of characters of the word to be translated, ``o'', ``ha'', ``yo'', and ``u''. In addition to detecting and counting the number of characters in the translated word/phrase/time required to apply the translated word/phrase =
The speech rate is calculated, and the calculation output thus calculated is sent to the control circuit 20. Here, the speaking speed is the average speaking speed.

On the other hand, the electronic translation circuit 14 translates "Ohayo" into a character signal "GOOD MORNING" in the translated language, and then the speech synthesis circuit 15 converts it into a voice signal "Good morning". The speed follows the speed adjusted by the unloading speed adjustment circuit 19.

However, the output speed adjustment circuit 19 is controlled by the control circuit 20 so that the utterance speed of "good morning" (translated phrase) matches the utterance speed of the translated phrase "good morning". Therefore, after being amplified by the output amplifier 16, the translated phrase "gutsudmorning" (time) is pronounced from the speaker 17.
The speaking speed at times t ₁₃ - t ₁₄ ) is the same as that of the translated word "good morning" applied to the microphone 11 (times t ₁₁ - t ₁₂ ).

In this example, ●Good morning...Time required for application ( _t12 - _t11 ), number of characters is 4.

●Good morning…Time required for sending (t ₁₄ −
t ₁₃ ), the number of characters is 8. (However, long sounds were counted as two characters).

Moreover, each time has the following relationship.

( _t12 − _t11 )≫( _t16 − _t15 ), ( _t14 − _t13 )≫( _t18 − _t17 ).

Number of characters in the translated word / time required to apply the translated word = speech rate of the source language, number of characters in the translated word / time required to send the translated word = speech rate of the translated language, then the speech in the source language Since speed = speech rate of the translated language = constant, 4/t ₁₂ −t ₁₁ = 8/t ₁₄ −t ₁₃ = constant holds true.

The control circuit 20 also controls the output speed so that the output speed of the speech (synthesized speech) sent from the speech synthesis circuit 15 is constant until the output of the translation language speech signal corresponding to the source language speech signal is completed. The adjustment circuit 19 is held and reset after the transmission of the translated language audio signal is completed. Then, the control circuit 20
adjusts and maintains the unloading speed adjusting circuit 19.

The same operation is repeated below.

Therefore, as shown in Figure 4B, when the translated word "Good morning" is applied to the microphone 11 in a fast manner, 4/t ₁₆ - t ₁₅ = 8/t ₁₈ - t ₁₇ = constant, as shown in Then, the translated phrase "good morning" can be uttered from the speaker 17 at the same utterance speed. In this way, if you speak quickly, your speaking speed will increase, and if you speak slowly, your speaking speed will slow down.

By doing this, the utterance speeds of the translated word and the translated word match, so that fast words are translated into fast words, and slow words are translated into slow words, and the words are pronounced.

Next, another embodiment of the present invention will be described using FIGS. 5 and 6.

FIG. 5 is a block diagram of an electronic translator using the "voice time adjustment method" which is another embodiment of the present invention, and FIG. 6 is a timing chart of input and output of each part of the same machine.

In the figure, 21 is a microphone, which is connected to the input amplifier 22. In the input amplifier 22, the microphone 2
The speech signal of the original language applied from 1 is amplified to a level that allows speech analysis, and is sent to the speech analysis circuit 23. The speech analysis circuit 23 converts the applied speech signal into a character signal and sends it to the electronic translation circuit 24. Thereafter, in the electronic translation circuit 24 set to Japanese-English, the character signal is translated into a character signal of the translated language. The translated character signal is converted into a speech signal of the translated language by the speech synthesis circuit 25, and then further amplified by the output amplifier 26 and outputted as speech from the speaker 27.

In the figure, 31 is a measurement circuit that measures the utterance time of the source language, and measures the time required to apply the word to be translated. This measurement circuit 31 is connected to the control circuit 30 and further connected to the discharge speed adjustment circuit 29.

The control circuit 30 controls the delivery speed adjustment circuit 29 so that the time required to send out the translated words matches the time required to apply the translated words measured by the measurement circuit 31 .

Therefore, when the number of characters in the word to be translated is greater than the number of characters in the word to be translated, the output speed adjustment circuit 29
The speaking speed in the speech synthesis circuit 25 is slowed down, and in the opposite case, the speaking speed is made faster, so that the speaking times of the word to be translated and the word to be translated are equalized.

Next, its operation will be explained.

As shown in FIG. 6A, when the original language "Good morning" is applied from the microphone 21, this input signal "Good morning" is amplified by the input amplifier 22 and then
The voice analysis circuit 23 converts the signal into a character signal of the original language "ohayyou" and sends it to the electronic translation circuit 24. In the electronic translation circuit 24, this character signal "Ohayo" is converted into the character signal "GOOD" of the translation language.
After being translated into “MORNING”, the speech synthesis circuit 2
At step 5, the voice signal "Good Morning" is converted into a voice signal in the translated language, and after being amplified by the output amplifier 26, the speaker 27 produces the sound "Good Morning". At this time, the time (t ₂₂ − t ₂₁ ) required to apply the original language “good morning” is:
Measurement is performed by the measurement circuit 31, and this measurement data is sent to the control circuit 30. The control circuit 30 counts the number of characters of the translated word to be pronounced, and divides the counted number of characters by the measured time to obtain the speaking speed per pronounced character. Control the unloading speed adjustment circuit 29 so that the time required to send out the "good morning" matches the time (t ₂₂ - t ₂₁ ),
The speech rate of the translated language sent from the speech synthesis circuit 25 is changed.

In this example, ● Good morning... Time required to apply the translated word (t ₂₂ − t ₂₁ ), ● Good morning... Time required to send the translated word ( t ₂₄ − _{t 23} ), ● Translated word The time required to apply = the time required to send the translated phrase, so time (t ₂₂ - t ₂₁ ) = time (t ₂₄ - t ₂₃ ) = T. Therefore, the speech rate of the source language V ₀ = Number of pronounced letters of the translated word/time (t ₂₂ − t ₂₁ ) = 4
/T, speech rate of the translated language V _T = number of pronounced letters of the translated phrase/time (t ₂₄ - t ₂₃ ) = 8/
T, however, the number of long sounds is 2 pronunciations.

As is clear from this, the rate of speech of the translated word is twice the rate of speech of the word to be translated.

With this configuration, as shown in FIG. 6B, if the translated phrase "good morning" is spoken quickly and is applied from the microphone 21 in a short period of time, the translated phrase "good morning" is heard from the speaker at the same time. It is pronounced from 27 onwards.

That is, ( _t22 - _t21 )=( _t24 - _t23 )≫( _t26 - _t25 )=( _t28 - _t27 ).

However, if the number of characters in the word to be translated is greater than the number of characters in the word to be translated, the utterance speed of the word to be translated will be faster because the utterance time is the same.

By doing this, the utterance times of both languages match, making it possible to translate more naturally.

Note that the present invention is not limited to the above two embodiments, and examples of embodiments of the present invention include the following.

(1) A detection unit for detecting the speech rate of the source language, an adjustment unit for adjusting the output speed of the translated language audio signal, and a control unit configured to control the output speed according to the input speed. An electronic translation machine with

(2) An electronic translation machine comprising the detection section, the adjustment section, and a control section configured to control the unloading speed in proportion to the loading speed.

(3) A measurement unit for measuring the utterance time of the translated word, a delivery speed adjustment unit for adjusting the utterance speed of the translated word, and a device for adjusting the utterance time of the translated word to match the utterance time of the translated word. and a control section for controlling an output speed adjustment circuit.

Additionally, since you cannot see the other person on the radio or over the phone, you have to judge the other person's emotional state based solely on their voice. Because I was pronouncing the words so fast, I couldn't even tell if the other person was speaking quickly or slowly.
However, by attaching the electronic translator of the present invention to a radio, television, telephone, etc., the speaker's emotional state can be more clearly conveyed to the other party, and communication can be deepened.

Furthermore, the electronic translator of the present invention has a function of changing the translation output signal level according to the input signal level, which has already been proposed by the present applicant (Utility Application No. 113154-1982, 1972). (Application filed on August 16th)
By combining these, the speaking speed and speaking level of the translated language can be controlled according to the speaking speed and speaking level of the source language, allowing for more natural translation.

As described above, according to the present invention, the speed or time at which audio information related to the original language input into the audio information input means is brought in, and the speed or time at which audio information related to the translated language outputted to the audio information output means is outputted. Since it is possible to create a correlation between If used in a translation phone, it is possible to not only transmit translation information, but also the speaker's emotional expression and speaking style such as fast or slow speaking. When it is difficult to hear the voice on the phone (synthetic voice), you can tell the other person to speak more slowly, and if the other person responds and speaks more slowly, what the other person said will be translated, and the person will be able to speak more slowly than before. The message is conveyed to the speaker using an easy-to-understand synthesized voice, which has the excellent effect of further enhancing communication between both parties.

[Brief explanation of drawings]

Figure 1 is a block diagram of a conventional electronic translation machine, Figure 2 is a timing chart of the input and output of each part of the machine, and Figure A is the output signal when the input signal "Good morning" is pronounced slowly. Figure B shows the time relationship of the output signal when the same input signal ``good morning'' is pronounced quickly. FIG. 3 is a block diagram of an electronic translation machine that is an embodiment of the present invention, and FIG. 4 is a timing chart of input and output of each part of the machine.
FIG. 5 is a block diagram of an electronic translation machine according to another embodiment of the present invention, and FIG. 6 is a timing chart of input and output of each part of the same machine. 1, 11, 21...Mike, 2, 12, 22...
...Input amplifier, 3,13,23...Speech analysis circuit, 4,14,24...Electronic translation circuit, 5,1
5, 25...Speech synthesis circuit, 6, 16, 26...
Output amplifier, 7, 17, 27...Speaker, 8,
18, 28... Character display circuit, 9, 19, 29...
... Unloading speed adjustment circuit, 20, 30... Control circuit,
31...Measuring circuit.

Claims (1)

[Scope of Claims] 1. A voice information input means for inputting voice information regarding a source language; a voice recognition means for recognizing voice information input to the voice information input means; and a source language recognized by the voice recognition means. a translation means for translating into information about a corresponding translation language based on information about the translated language; a speech synthesis means for synthesizing speech information corresponding to the information about the translation language translated by the translation means; audio information output means for outputting audio information related to the translated language that has been translated; audio information outputting means for outputting audio information regarding the original language inputted to the audio information inputting means; and input speed or time for inputting audio information regarding the original language inputted to the audio information inputting means; and information regarding the translated language outputted to the audio information outputting means. An electronic translator comprising input/output correlation control means for establishing a correlation between the output speed or output time of audio information. 2. While outputting the translated word/phrase audio information to the audio information output means, the input/output correlation control means controls the input speed or time of the translated word/phrase audio information and the translated word/phrase audio information corresponding to the translated word/phrase audio information. The electronic translation machine according to claim 1, wherein a correlation with a delivery speed or a delivery time is maintained. 3. The input/output correlation control means correlates the input speed or time of the translated word/phrase audio information so that the speed or time of import of the translated word/phrase audio information substantially matches the output speed or time of the translated word/phrase audio information corresponding to the translated word/phrase audio information. Claim 1 characterized in that the control is performed.
The electronic translation machine according to item 1 or 2.