JPH0493980A - Language learning system - Google Patents

Abstract

PURPOSE:To visually discriminate between both a standard pronunciation and a learner's voiced pronunciation and to learn a language by using their similarity by spectrum analyzing both pronunciations respectively and generating voice formant signals, and displaying them on a monitor, etc., one over the other. CONSTITUTION:When learner inputs his or her voice pronunciation through a microphone 24, the voice pronunciation signal is amplified temporarily by an amplifier 25 and sent to an LPF switch 7 and a spectrum analyzer 8 detects its power spectrum to generate a voice pronunciation formant signal, which is passed through a learner voice pronunciation storage circuit 17 and mixed by a mixing circuit 21, so that its image is displayed on the monitor where the character and symbol frame of a video channel and the standard pronunciation formant signal are displayed one over the other. Here, the shape is generated with a sampling frequency signal sent to a control circuit 14. Then the learner discriminates and recognizes whether or not the standard pronunciation formant signal and voice pronunciation formant signal are visually similar, or whether or not the learner correctly pronunces according to the standard pronunciation.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a language learning system, and more specifically, the present invention relates to a language learning system, and more specifically, it is capable of producing both standard pronunciation by a teacher, etc., and learner-generated pronunciation by a learner, respectively. The present invention relates to a language learning system in which a visually recognized shape signal is displayed on a monitor so that a learner can learn a language by visually recognizing coincidence or mismatch (similarity or dissimilarity) between both signals.

[Conventional technology]

Conventionally, language learning systems have consisted of a text, a tape recorded with the content of the text as standard pronunciation by a teacher, etc.
The system is equipped with an auditory comprehension device such as a tape recorder that plays a disc, and the learner uses the auditory comprehension device to repeatedly play standard pronunciation recorded on a tape etc. while referring to the above text to grasp the auditory comprehension. After that, the learner makes an audio pronunciation imitating the learned standard pronunciation, records and plays it on the auditory comprehension device, and learns by comparing the standard pronunciation with the learner's recorded and played pronunciation. It has become a learning system.

This learning system has the advantage that learners can study in their free time and as needed.

By the way, each ethnic group has a different language, and each has its own Φ spoken language. However, these spoken languages are all means of human self-expression, and when the need arises, the content that should be uttered through logical thinking is expressed as voice based on memories recalled to the cerebral center. However, this spoken language is not just a string of characters, but also requires information such as strength, weakness, and flexibility. and,
The voice enters the speaker's ears at the same time as it is uttered, and is processed in the same way as when listening to someone else's voice, and is compared with the words still in short-term memory. Through this comparison, the words that have just been uttered are confirmed to be correct. Semi-consciously, we check whether the words are articulated or not, control the movements of our mouths, observe the attitude and actions of the person we are speaking to, notice mistakes and rephrase them, so that we can utter the correct language.

On the other hand, the recognition of the voice uttered by another person begins with the voice entering the ear, and the voice vibrates the eardrum as a sound wave, and the vibration is observed as an electric pulse of -22-ron. It is sent to the auditory nerve network, processed, and changes form until it reaches the auditory center, the temporal lobe of the brain. If the sound is a word, the brain compares it with the word in memory as input information, and if it is similar to the word in memory, it is said to have been heard. Also, if there is no similar word in memory, the input information is broken down and each part is compared with memory, and a word that combines the words of the colleagues is input. In this way, unfamiliar words cannot be recognized, but familiar words can be easily recognized. Furthermore, if there is a word predicted based on situational judgment, it is assumed that the predicted word is heard even if it is slightly different, and it is said that the essence of human speech recognition is that the predicted word is heard.

Now, if we consider how language learning is originally carried out, we can understand the state in which humans learn words by considering the state in which infants and young children memorize words for the first time. In other words, when a mother who is a teacher touches an infant's body and encourages the infant to vocalize while repeating vocal pronunciations, the infant recognizes the mother's facial expressions,
Feel what you need to say out loud through your movements and vocalize it. In this case, the mother encourages the infant to practice repeatedly until she thinks that the infant's vocal pronunciation is the same as her own, and the infant watches the mother's mouth movements and tries to produce the same pronunciation as the one heard through the ears. I will do it. On the other hand, when the pronunciation is the same, the mother expresses her emotions strongly through words, emotions, and actions, and the infant sees the mother's facial expressions and memorizes this voice strongly in his brain, and the infant can easily repeat the same voice next time. They begin to pronounce sounds, and this state is said to be the origin of how humans learn words.

Next, considering the case of language learning based on the above points, the language learning is based on the standard pronunciation of audio pronunciation,
It starts with the learner imitating the pronunciation, and continues until the learner's vocal pronunciation is recognized as similar to the standard pronunciation.
The system repeatedly learns the vocal pronunciation, and when the vocal pronunciation is recognized as similar to the standard pronunciation, the vocal pronunciation is strongly memorized in the brain as the standard pronunciation. in this case,
In language learning, it is ideal for teachers to compare and recognize pronunciations, but for boat fishing, we have chosen a learning method that uses teaching materials from the language learning system described above.

[Problem to be solved by the invention]

However, in the case of such language learning systems, the learner's auditory judgment is used to determine whether the learner's pronunciation matches (is similar to) the above-mentioned standard pronunciation, and this discrimination is usually insufficient. The reality is that the progress of learning cannot be monitored. In other words, with conventional teaching materials, it is not possible to compare and identify standard pronunciations, so it takes a long time for learners who do not have the ability to identify standard pronunciations.
It is easy to make the mistake of recognizing incorrect pronunciation as standard pronunciation. For this reason, even if students are able to pronounce similar words, their hearing ability does not improve, so their progress in learning is slow and they often stop learning midway through.

Based on the above points, the present inventors believe that by equipping a learning device with the ability to compare and identify standard pronunciations, learners will be able to memorize the correct standard pronunciation in their brains, and in the same way that infants and toddlers can memorize words, the present inventor will Focusing on the point that it may be possible to understand the language, we considered conventional language learning materials.

In the case of conventional teaching materials, they only work on the auditory sense of the learner, and since there is a concept that "sound" is "something heard with the ear," vocal pronunciation is temporary as far as it is based on the auditory sense. We learned that there are individual differences in the comparative discrimination ability of vocal pronunciations, and that it is impossible to ask learners to compare and discriminate vocal pronunciations due to individual differences in auditory ability.

Therefore, the present inventor investigated a method using visual sense from the viewpoint of whether there is any sense other than hearing among the five human senses that can identify sounds. Although there are originally individual differences in vision as well as hearing, it can be said that shape recognition is an ability that does not cause individual differences. From this, if it is possible to express the audio pronunciation in a form, it becomes possible to express the standard pronunciation and the audio pronunciation in the form of a form. In addition, if it becomes possible to formulate phonetic pronunciations, it will be possible for any learner to easily and accurately compare and identify phonetic pronunciations. It becomes possible to obtain.

The question then becomes what kind of means is preferable for visually comparing and identifying vocal pronunciations. First, vocal pronunciations have pitches. In other words, when human vocalizations change the pitch, the fundamental frequency changes, but when heard as a voice, it sounds like the same sound with a certain range of range. This is because when producing a sound, even if the pitch changes, the vocal organs from the glottis to the tip of the lips hardly change. The area from the glottis to the lips is called the vocal tract.The vocal tract has a shape, and as long as the shape of the vocal tract is the same, the voice information heard by the human ear is independent of the sound produced by the sound source. This means that everyone hears the same audio information. The exchange of voice information by the human ear begins with the same shape of each person's vocal tract, and the shape of the vocal tract matches the frequency characteristics of the vocal tract, resulting in vocal pronunciation that is as close to standard pronunciation as possible. The shape of the vocal tract can be said to be the same, and the frequency characteristics of the vocal tract can be said to be the same.

Therefore, by detecting the frequency characteristics of the vocal tract, it becomes possible to compare voice information and perform voice recognition. Speech recognition compares and identifies voice information,
This method performs comparative recognition of unique speech information based on standard pronunciation, so it is currently possible to use complete speech! ! The fire is not completed.

However, in the case of learning, the purpose of the learner repeating the vocal pronunciation by imitating the standard pronunciation is for the vocal pronunciation to be similar to the standard pronunciation. Therefore, by detecting the shape of the vocal tract of the taYs pronunciation, that is, the frequency characteristics of the vocal tract, it becomes possible to compare and identify the standard pronunciation and the vocal pronunciation as similar pronunciations. We learned that complete speech recognition is possible only through learning.

Then, the vocal pronunciation is shaped by detecting the frequency characteristics (formants) mentioned above, and the standard pronunciation waveform is displayed as a still image on the TV monitor screen as a standard pattern, and the vocal pronunciation is also processed in the same way. The utterance sound waveform is displayed as a utterance pattern superimposed on the standard pattern, and the learner can visually compare and recognize the similarity between the standard pattern on the monitor and the utterance pattern. here,
The standard pattern can be displayed at any time,
Learners compare and identify the pronunciation pattern that is displayed each time they pronounce the pronunciation with the standard pattern, and repeatedly learn the pronunciation until the pronunciation pattern resembles the standard pattern. We have determined that it is possible to perform learning.

The present invention has been devised in response to the above-mentioned problems, and its purpose is to visually identify the standard pronunciation and the vocalized pronunciation produced by the learner by shaping them. However, it is an object of the present invention to provide a language learning system that enables language learning based on the similarities.

[Means to solve the problem]

The language learning system of the present invention as a means for solving the above problems spectrally analyzes standard pronunciation to detect a power spectrum, extracts characteristic parameters related to phonology from the power spectrum, and A standard pronunciation visual recognition display means that obtains a pronunciation shaping signal and can visually recognize and display the standard pronunciation, and detects a power spectrum by spectral analysis of learner's uttered pronunciation, and detects phonological characteristics from the power spectrum. A learner's vocal pronunciation visual recognition display means is capable of extracting feature parameters related to the learner's vocal pronunciation to obtain a learner's vocal pronunciation shaping signal and displaying the signal through visual recognition of the learner's vocal pronunciation; The signal comparison means visually compares the standard pronunciation shaping signal displayed for visual recognition of pronunciation and the learner vocalization shaping signal displayed for visual recognition of the learner's vocalization, and the signal comparison means The system is configured so that language learning can be performed by understanding the degree of similarity between both signals.

Further, in the language learning system of the present invention, in the above configuration, if necessary, the standard pronunciation and the learner's uttered pronunciation are subjected to spectrum analysis using a common spectrum analyzer to detect the power spectrum of the pronunciation. It may also be configured to extract feature parameters related to gender.Alternatively, the standard pronunciation shaping signal and the learner's utterance pronunciation shaping signal can be displayed as images on the monitor, and the signals can be generated by superimposing both signals. A configuration that allows comparison to be made visually may also be used. In addition, the signal comparison means for comparing the standard pronunciation shaping signal and the learner's vocalization shaping signal can detect signal coincidence, mismatch, signal sum, or dissimilarity using an LED.
Visual display may be provided using a light emitting device such as a lamp. Furthermore, a standard pronunciation record is created by recording an audio pronunciation of a word or sentence and a word length detection signal that specifies the length of the word or sentence that constitutes the audio pronunciation. is configured as a standard pronunciation, and a standard pronunciation recording body is formed as a video disk, and characters and symbol frames such as suberu and phonetic symbols of the standard pronunciation are recorded along with images on the video channel of the disk, and In one channel, audio such as explanatory audio and standard pronunciation audio corresponding to the video of the video channel is recorded, and in another channel, standard pronunciation and word length detection signals for repetition are recorded. Good too.

[Effect]

Based on the above configuration, the language learning system of the present invention uses a standard pronunciation recording medium such as a video disk that records standard pronunciation created by a teacher or the like, or standard pronunciation that is input from now on, and uses standard pronunciation. Using a visual recognition display means, spectrum analysis is performed to obtain the frequency characteristics (formant) of the vocal tract, and this waveform is visually recognized and displayed on a monitor.
After that, the learner is asked to make an audio pronunciation that imitates the standard pronunciation, and the audio pronunciation is analyzed using a visual recognition display means to obtain the frequency characteristics of the vocal tract, in the same way as the standard pronunciation. After visually recognizing and displaying this waveform on a monitor, the learner visually compares the image shape signal using a signal comparison means and learns the language while visually grasping the degree of learning. act so that it can.

As described above, the language learning system of the present invention spectrally analyzes the standard pronunciation and the learner's pronunciation, and
The feature is that the formant is detected, converted into a speech shape signal, displayed on a monitor, etc., and superimposed so that language learning can be visually compared and identified. This system has a special effect in that it is possible to understand whether or not there is similarity between the standard pronunciation and the learner's pronunciation using not only the target but also visual means, and learning similar to that performed by infants and young children.

〔Example〕

Hereinafter, embodiments embodying the present invention will be described with reference to the drawings.

Here, Figs. 1 to 4 show an apparatus embodying the system of the present invention, Fig. 1 is a schematic system diagram, Fig. 2 is a schematic circuit configuration diagram, and Fig. 3 is a schematic configuration of a spectrum analyzer. figure,
FIG. 4 is an explanatory diagram of the monitor screen displaying the standard pronunciation shaping signal and the learner's vocalization shaping signal, and FIG.
) is an explanatory diagram of the shaping signal on the monitor screen when the number of samplings for the a sound is controlled to one, and FIG. 5(b)
FIG. 2 is an explanatory diagram of a shaping signal on a monitor screen when the number of samplings for the a sound is controlled to 10 times.

In summary, the language learning system of this embodiment has four stages: (1) standard pronunciation creation means, (2) standard pronunciation visual recognition display means, (2) learner's utterance pronunciation visual recognition and display means, and (4) shaped signal comparison means. .

- Standard pronunciation creation means This means includes means for recording standard pronunciations on a recording medium such as a video disc by a teacher or the like, or for directly inputting standard pronunciations by a teacher or the like without going through a recording medium.

Here, the standard pronunciation is, as mentioned above, the pronunciation that corresponds to teaching materials for language learning. Then, the standard pronunciation is memorized and recorded by the teacher on a recording medium such as a video disk, and the words and sentences corresponding to the text are composed.
■Character/symbol frames such as standard pronunciation spellings, pronunciation letters and symbols, and videos for learning progress, ■Standard pronunciation audio during repetition, standard pronunciation control signals, and sampling signals, ■
Consists of explanatory audio including standard pronunciation that corresponds to the video. The disc (laser disc) has video channels and
Two audio channels can be used, and the video for learning progress and the character/symbol frames for standard pronunciation are recorded as video on the video channel, and the video channel is used on one of the audio channels (channel A). When repeating (in other words, when the video is a still image), the explanatory audio including the standard pronunciation corresponding to the video is sent to the other channel (B channel) along with the standard pronunciation control signal and the sampling detection signal. , record the standard pronunciation to be repeated, and create a standard pronunciation recording medium.And,
Images of frames for learning progress are played on the monitor, and audio corresponding to the images is played from the audio A channel in conjunction with the learning frame images. When the image reaches the end of the image of the frame for learning progress, it becomes a still image and the character/symbol frame is displayed.

The standard pronunciation control signal is a signal for controlling the word length of a word or sentence in standard pronunciation, and has a start signal and a stop signal. This allows the word length to be identified. As a result, when the video channel is a still image, the standard pronunciation consisting of the above-mentioned words or sentences recorded on the audio B channel can be changed to ) can be displayed on one screen of the monitor. In addition, the sampling number signal is a control signal that allows the learner to easily view the standard pronunciation and vocalized pronunciation shaped signals on the monitor, and is the number of extraction times (sampling) of feature parameters for each pronunciation.
This is a signal that can be changed.

Standard pronunciation visual recognition display means This means uses the standard pronunciation obtained by the standard pronunciation creation means, analyzes the spectrum to detect the power spectrum, and extracts characteristic parameters (formants) related to phonology from the power spectrum. This is a means for obtaining a standard pronunciation shaping signal and displaying the standard pronunciation visual recognition signal.

This method is implemented by using the device shown in Figure 1.2 for boat fishing. To explain this device, monitor TV 1
It consists of three devices: a pronunciation identification device 2, and a laser disc player 3. Pronunciation identification device 2
Specifically, as shown in FIG. The spectrum is analyzed by the spectrum analyzer 8 via the AD converter 5, DA converter 6, and LPF sinchia connected in series, and the formant is detected to obtain a standard oscillation waveform, which is then sent to the formant generation circuit 9 and written into the memory. It is connected to a standard oscillation waveform storage circuit 11 via a circuit 10 so as to be stored therein. A standard pronunciation storage circuit (usually RAM) 12 is connected to the AD converter 5, and the standard pronunciation digital signal of the audio B channel/channel in the disc audio AD converted by the AD converter 5 is stored in the standard pronunciation storage circuit. 12
It has been stored in memory. Further, a control circuit 14 is connected to the standard pronunciation storage circuit 12 via a memory counter 3. In addition, input terminal 4 of audio B channel
The standard pronunciation control signal and the sampling number signal of the audio B channel are connected to the control circuit 13 via the audio signal detection circuit 15 so that the standard pronunciation control signal and the sampling number signal can be used to control the standard pronunciation. The standard sounding digital signal stored in the memory circuit 12 is repeated by the control circuit 13 and the memory counter 2,
The signal can be transmitted to the LPF switch 7 via the DA converter 6. Here, the word length of the standard pronunciation is recognized by the standard pronunciation control signal, and accordingly, the standard pronunciation can be repeatedly reproduced within the range of the word length. The control circuit 13 is also connected to an operation switch 16, a spectrum analyzer 8, a standard utterance waveform storage circuit 11, a learner's utterance waveform storage circuit 17, and counters 18 and 19, respectively. Further, the signals of the standard utterance waveform storage circuit 11 and the learner's utterance waveform storage circuit 17 are mixed together by a disk video input terminal 20 and a mixing circuit 21, and the signals are combined and displayed on the monitor 1 as a video signal. are connected like this. Further, the LPF switch 7 is connected via an amplifier 25 to an audio signal input terminal 22 for video progression synchronized with the video of the video channel recorded on the audio A channel, a receiver 23, and a microphone 24.

With this circuit, the standard pronunciation is analyzed by the spectrum analyzer 8 to detect the power spectrum, and the characteristic parameters (formants) related to phonology are extracted from the power spectrum to generate the standard pronunciation shaping signal. The waveform signal is displayed on the monitor 1 via the storage circuit 11 and the mixing circuit 21 according to instructions from the operation switch 16 and the control circuit 13.

The spectrum analyzer 8 is constituted by a filter bank as shown in FIG. 3, detects formants, and obtains a standard pronunciation shaping signal for visually recognizing and displaying the standard pronunciation. Specifically, bandpass filters 26, 26, etc. with slightly different center frequencies are connected in parallel.
Moreover, the filters 26, 26, . . . are arranged so that there is no gap between the passbands of adjacent filters, and the averaging circuits 28, 28, .
are connected, and the outputs of the bandpass filters 26, 26... are squared by the squarer 27, 27..., and the averaging circuit 2
8.28... is the average value, that is, each filter 26.
The power of the audio signal included in the pass band of 26... is determined, and the power of the audio signal included in the passband of 26 is determined, and then the power of the audio signal included in the pass band of
．． 28..., understand how much power is included in each frequency band, and add up all these powers to obtain a power spectrum. Then, for example, a formant creation circuit 9 composed of an integrating circuit can obtain a formant consisting of a curve connecting the leading ends of the power spectrum.

Note that although the spectrum analyzer used here has been described as using a filter bank, a method using a linear prediction method may also be used.

- Learner's vocal pronunciation visual recognition display means This means spectrally analyzes the learner's vocal pronunciation to detect a power spectrum, extracts characteristic parameters (formants) related to phonology from the power spectrum, and extracts feature parameters (formants) related to phonology from the power spectrum. This means obtains a pronunciation shaping signal and visually recognizes and displays the pronunciation produced by the learner. This means visually recognizes and displays pronunciation in the same manner as the standard pronunciation visual recognition display means. That is, the pronunciation identification device used in the standard pronunciation visual recognition display means is used to visually recognize and display pronunciation.

Shaped signal comparison means This means visually compares the standard pronunciation shaped signal visually recognized and displayed by each means with the learner's uttered pronunciation shaped signal displayed by visual recognition of the learner's uttered pronunciation. It is a means to do so. That is, in the pronunciation recognition circuit shown in FIG. 2, the standard pronunciation shaping signal and the learner's utterance pronunciation shaping signal are superimposed (for example, placed in parallel on the monitor at top and bottom positions, or superimposed). It is a means of making it possible to visually identify it.

Next, to explain how to use the language learning system of this embodiment described above, the video channel includes learning videos (videos explaining how to practice standard pronunciation and videos explaining practice words and sentences) and the explanatory videos. At the end, there is a frame of standard pronunciation spelling, pronunciation letters, and symbols that becomes a still image, and audio A
The channel corresponds to the learning video of the video channel (
Using a laser disc that has a synchronized video and audio, and also has a standard pronunciation, a standard pronunciation control signal, and a sampling number signal on the audio B channel, when the laser disc is played by the operation switch by playing the laser disc on the laser disc player 2, the pronunciation identification device The video/audio signal of the audio A channel is input to the video/audio input terminal 22 of the disc, and the learner can reproduce the video/audio with the receiver 23, and the standard pronunciation signal of the audio B channel is input to the disc audio input terminal. A sound generation control signal and a sampling number signal are inputted, and a standard sounding signal and three other signals are detected by an audio signal detection circuit 15, and a disc video input terminal 20
A video channel signal is input to. Then, the learning video is played back on the monitor using the video channel 2 signal and the audio A channel signal, and the video and audio corresponding thereto are also played back. When the explanation of the standard pronunciation is finished in the learning video and the character/symbol frame of the standard pronunciation is entered, the laser disc player 2 converts it into a still image, while it is recorded on the audio B channel, and the audio signal detection circuit 15 According to the detected standard pronunciation control signal, it is recorded on the audio B channel and output to the disc audio input terminal 4.
The standard pronunciation signal inputted from the standard pronunciation memory circuit 12 is repeated by the control circuit 14 and the memory counter 13, and then sent to the LPF via the DA converter 6.
The signal is transmitted to the switch 7, the spectrum is analyzed by the spectrum analyzer 8, and a power spectrum is detected. From the power spectrum, the formant in the standard pronunciation is detected, a standard pronunciation shaping signal is obtained, and the standard pronunciation is stored. The character/symbol frame of the video channel is superimposed on the standard pronunciation shaping signal by the mixing circuit 21 via the circuit 11, and the image is displayed on the monitor 1. Here, according to the standard pronunciation control signal, the standard pronunciation shaping signal to be displayed as an image on the monitor 1 can be displayed on the monitor screen regardless of the length of the word of the standard pronunciation. That is, detecting the signal,
By controlling the time for writing the standard pronunciation shaping signal into the standard pronunciation storage circuit 12, the screen of the monitor 1 can be used effectively. Further, the standard pronunciation shaping signal is voice B
With the sampling number signal recorded in the channel, as shown in Fig. 5, on the monitor 1, the shaped signal of each word is set in advance according to the words that make up the standard pronunciation so that it can be easily identified visually. Then, the spectrum analyzer 8 samples a predetermined number of samples and displays the images. The learner can repeatedly hear the standard pronunciation on the listener 23 via the LPF audio switching circuit 7.

Next, if learners want to learn a language according to standard pronunciation,
When voice pronunciation is input through the microphone 24, the voice pronunciation signal is amplified by the amplifier 25, transmitted to the LPF switch 7, and subjected to spectrum analysis by the spectrum analyzer 8 in the same manner as the standard pronunciation described above. From the power spectrum, the formant in the voice pronunciation is detected and a voice pronunciation shaping signal is obtained. Monitor 1 on which the symbol frame and the above standard pronunciation shaping signal are displayed overlappingly
An image is displayed above (see Figure 4). Here, similarly to the standard pronunciation, the vocal pronunciation produced by the learner is also shaped in the same way as the standard pronunciation by the sampling number signal transmitted to the control circuit 14. This is because the learner's pronunciation is a true version of the standard pronunciation. Further, the learner's vocal pronunciation is also processed in the same way as the standard pronunciation by the audio control signal (audio B channel). Then, the learner visually checks whether the standard pronunciation shaping signal and the audio pronunciation shaping signal match (match/disagree). Able to identify and recognize whether someone is using a computer or not. Here, whether the two signals match or not can be checked visually by the learner himself or by using other visual identification means such as a meter, an LED, a light emitting device such as a lamp (not shown), etc. There is.

After learning a predetermined standard pronunciation, by operating the operation switch, the laser disc player 2 plays the laser disc, and the user can learn the next standard pronunciation in the same way.

It should be noted that the present invention is not limited to the above-described embodiments, and includes modifications that can be made without departing from the gist of the present invention. Incidentally, in the above-mentioned embodiment, the standard pronunciation was previously recorded on a video disk. However, it is of course possible that the standard pronunciation may be directly uttered by the teacher, for example.

〔Effect of the invention〕

As is clear from the above explanation, according to the language learning system of the present invention, the standard pronunciation and the learner's uttered pronunciation are each subjected to spectrum analysis, formants are detected, a speech shape signal is generated, and this is transmitted to the monitor, etc. In addition to displaying the pronunciation on the screen, it is also superimposed to visually compare and identify, so it is possible to check whether or not the standard pronunciation and the learner's pronunciation are compatible not only audibly but also visually. It has the effect of being able to understand and perform learning similar to that performed by infants and young children.

[Brief explanation of drawings]

1 to 4 show an apparatus that embodies the system of the present invention. FIG. 1 is a schematic system diagram, FIG. 2 is a schematic circuit diagram, and FIG. 3 is a schematic diagram of a spectrum analyzer. Figure 4 is an explanatory diagram of the monitor screen displaying the standard pronunciation shaping signal and the learner's vocalization shaping signal, and Figure 5 (a) is a
FIG. 5(b) is an explanatory diagram of the shaping signal on the monitor screen when the number of samplings for the sound is controlled to 1. FIG. 5(b) is an explanatory diagram of the shaping signal on the monitor screen when the number of samplings for the sound a is controlled to 10 times. . 1... Monitor TV, 2... Pronunciation identification device, 3...
・Laser disc player, 4...B channel audio input terminal, 5...AD converter, 6...DA converter, 7...LPF switch, 8...spectrum analyzer, 9... Formant creation circuit, 10... Memory writing circuit, 11... Standard oscillation waveform storage circuit, 12...
Standard pronunciation memory circuit, 13...Memory counter, 14.
...Control circuit, ]5...Write signal detection circuit, 16...
・Operation switch, 17...Learner's utterance sound waveform memory circuit, 18.19...Counter 20...Input terminal for disk video, 21...Mixing circuit, 22...Audio signal input terminal for video progression , 23... Lenher, 2
4...Microphone, 25...Amplifier, 26...Band pass filter, 27...Squarer, 28...Averaging circuit patent

Claims (7)

[Claims] (1) A standard pronunciation is spectrally analyzed to detect a power spectrum, and feature parameters related to phonology are extracted from the power spectrum to obtain a standard pronunciation shaping signal, and this signal can be displayed for standard pronunciation visual recognition. The standard pronunciation visual recognition display means that spectrally analyzes the learner's spoken pronunciation to detect the power spectrum, extracts feature parameters related to phonology from the power spectrum, and generates the learner's vocal pronunciation shaping signal. a learner's uttered pronunciation visual recognition display means for visually recognizing and displaying the signal as the learner's uttered pronunciation, the standard pronunciation shaped signal visually recognized and displayed by each of the means, and the learner's uttered Pronunciation Visual Recognition The present invention has a signal comparison means for visually comparing the displayed learner's utterance pronunciation shaping signal, and the signal comparison means allows language learning to be carried out by grasping the degree of similarity between both signals. A language learning system characterized by the following. (2) The standard pronunciation and the learner's spoken pronunciation are subjected to spectrum analysis using a common spectrum analyzer to detect the power spectrum of the pronunciation and extract feature parameters related to phonology. language learning system. (3) The standard pronunciation shaping signal and the learner's utterance pronunciation shaping signal are displayed as images on the monitor, and the signals can be visually compared by superimposing both signals. Language learning system. (4) The signal comparison means that compares the standard pronunciation shaping signal and the learner's utterance pronunciation shaping signal visually displays signal coincidence, mismatch, signal similarity, and dissimilarity using light emitting elements such as LEDs and lamps. The language learning system according to claim 1. (5) Create a standard pronunciation record by recording an audio pronunciation of a word or sentence and a standard pronunciation control signal that specifies the word length of the word or sentence that makes up the audio pronunciation, and 2. The language learning system according to claim 1, wherein the medium is standard pronunciation. (6) The standard pronunciation recording medium is formed of a video disc, and the video channel of the disc records standard pronunciation learning videos as well as character/symbol frames such as standard pronunciation spellings and phonetic symbols. On one channel, audio such as explanatory audio and standard pronunciation audio corresponding to the learning video of the video channel is recorded, and on the other channel, standard pronunciation for repetition and standard pronunciation control that detects and controls word length. The language learning system according to claim 1, wherein the language learning system records signals. (7) A standard pronunciation for repetition, a standard pronunciation control signal for detecting and controlling word length, and a power spectrum sampling frequency control signal are recorded in the other audio channel. language learning system.