JP4669988B2 - Language learning device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a language learning device capable of effectively presenting a stimulation sound for learning whose acoustic features are operated to a learner. <P>SOLUTION: In the foreign language learning device 300, a hard disk 324 stores a model speech of a foreign language in the form of an original speech database 3242 and also stores the degree of attainment of a learner and amplification rages of acoustically featured parts included in the model speech in the form of an amplification degree database 3246 while making them correspond to each other. A learning control section 3220 synthesizes a stimulation speech by selectively amplifying acoustically featured parts at a designated amplification rate and updates the amplification rate into a less one when a correct answer rate when the stimulation speech is imparted to the learner meets level-up conditions predetermined according to the degree of attainment. <P>COPYRIGHT: (C)2006,JPO&amp;NCIPI

Description

  The present invention relates to a configuration of a language learning apparatus for performing language learning.

  In recent years, there have been many attempts to use a character or speech input / output system implemented by a computer, a relational database system, or a speech recognition system for learning a foreign language. That is, in such a foreign language learning apparatus, the learner inputs an answer to the personal computer by a keyboard operation, a mouse operation, or a voice in accordance with a problem to be asked by the computer. The database system and the speech recognition system corresponding to the foreign language installed in the personal computer determine the learner's understanding and correctness of pronunciation and display the result as feedback to the learner. It is.

  Also, the learner's pronunciation of the sentence is indicated for each word, and the pronunciation evaluation result for the learner's pronunciation of the foreign language sentence is efficiently used for the learner's pronunciation of the foreign language. A foreign language learning apparatus for feedback is also proposed (see, for example, Patent Document 1).

  By the way, many learning experiments have been conducted so far, in order to enable effective learning, operating voices used as learning stimuli, enhancing acoustic features, controlling speech speed, etc. As a result, it is difficult to obtain a good result.

  On the other hand, voice stimuli are converted into high-quality sound (CD quality), streaming broadcast quality sound (Internet radio, etc.) and mobile phone quality sound, respectively. First, these intelligibility is converted into Japanese native speakers and American native speakers. After measuring the above, the results of conducting a learning experiment using Japanese native speakers as subjects, and using a speech format (and phonology) that was clear to American native speakers as a learning stimulus. However, it was reported that the training effect was low or was not recognized in the group in which learning was performed in a speech format with low clarity (for example, Non-Patent Document 1 and Non-Patent Document 2). See).

Considering these things together, it is highly likely that some conventional acoustic manipulations have been performed acoustically, but as a result, clarity has not been achieved, resulting in training effects. It is possible that there is no significant difference for.
JP 2001-265 211 A Adachi, Yamada, “Examination of Validity of Second Language Speech Learning Using Compressed Speech”, Acoustical Society of Japan, 3-5-19, pp. 503-504 (20033.3) Adachi, Yamada, “Effect of speech quality on phonological perception learning in second language: Comparison between QCELP and PCM formats”, 2004 Spring Meeting (ASJ) Proceedings, pp. 437-438

  Furthermore, acoustic differences and clues at the time of listening generally differ due to phoneme conflicts. Therefore, for example, / r / and / l / (hereinafter abbreviated as RL), / b / and / v / (hereinafter abbreviated as BV), / s / and / θ / (hereinafter abbreviated as STH) are Japanese native languages. The phoneme is difficult to perceive for the speaker, but the effect of adding noise may be different. Thus, for a speaker having a certain native language, a phoneme pair having difficulty in perceptual discrimination will be referred to as a “phoneme pair having a phoneme conflict”.

  For this reason, for example, even when noise masks acoustic features with the same intensity, the acoustic features used for phonological perception differ depending on the subject's native language, so the clarity for the learner may vary There is also. Therefore, a phoneme pair with such a phoneme conflict is exactly what the learner should be listening to, but the acoustic operation is controlled by the learner's native language and by the type of the phoneme that conflicts. There may be a need to change.

  However, conventionally, in learning a foreign language, there has been a problem that the method is not always clear as to how to present a learning stimulus sound with an acoustic feature manipulated to a learner.

  The present invention has been made to solve the above-described problems, and its purpose is to effectively present a learning stimulus sound in which an acoustic feature is manipulated to a learner. It is to provide a language learning device.

In order to achieve such an object, the language learning device of the present invention stores a model speech of a language and associates a learner's achievement with an enhancement rate of an acoustic feature portion included in the model speech. storage means for storing Te, the model voice, a stimulus sound synthesizing means for synthesizing selectively emphasized stimulate speech acoustic characteristic portion of the model voice given emphasis ratio, stimuli The synthesizing unit includes a detecting unit for identifying a position on the time axis where the acoustic feature portion exists in each of the model sounds input from the storage unit, and the stimulation sound synthesizing unit is in accordance with the identified position. Then, the acoustic features are selectively emphasized, and a sound signal for giving the stimulus sound to the learner as a task is output to the sound output device that reproduces the stimulus sound, and an answer from the learner is received. And interface means for, by calculating the correct rate of the learner with respect to a given stimulus speech as a problem by specifying the enhancement factor, and controls the stimuli synthesis means, when stimulated speech for the learner correct answer rate of, in response to the level-up condition is satisfied, which is predetermined according to achievement, further comprising a learning control means for updating to reduce the enhancement rate specified for stimuli synthetic means.
Preferably, the storage unit stores positional information indicating in which position on the time axis of the model voice a phonological component having an acoustic characteristic of the model voice exists in association with the model voice. The detecting means specifies the position on the time axis where the acoustic feature portion exists based on the position information.
Preferably, the stimulus sound synthesis means further includes frequency analysis means for performing frequency analysis on the model voice and parameterizing the acoustic characteristics of the phonemes in the model voice, and the detection means is analyzed by the frequency analysis means. Based on the result, the position on the time axis where the acoustic feature portion exists is specified.

  Preferably, the learning control unit updates the amplification rate so that the correct answer rate when the stimulus is given to the learner satisfies a level-down condition that is determined in advance depending on the degree of achievement. To do.

Preferably, the acoustic feature portion corresponds to the opposite phoneme portion, the storage means classifies and stores the degree of achievement for each opposite phoneme, and the learning control means sends the specified conflict to the stimulus sound synthesis means. for each phoneme, with reference to the stored achievement in the storage means to generate a stimulus sound, interface means outputs the audio signal for presenting the generated stimuli voice for the learner.

  Embodiments of the present invention will be described below with reference to the drawings.

  In the following description, the present invention will be described by taking as an example a case where a learner whose native language is Japanese learns English. However, the present invention can also be applied to a case where a learner whose mother tongue is other than Japanese learns a language other than his / her native language, and more generally can be applied to a foreign language learning apparatus.

  The present invention can also be applied as a more general language learning device when a learner who is not sufficiently familiar with the native language learns about his / her own native language. In this case, in the following description, the learner's native language proficiency may correspond to the foreign language native speaker for the learner.

  As will be explained later, the effects of noise type and intensity on clarity when multiple noises with different characteristics are added to English speech and presented to native Japanese speakers and native American speakers Experiments have shown that the percentage of correct answers is not an absolute value but a relative trend. This suggests that, even when noise masks the acoustic features with the same intensity, the clarity is different because the acoustic features used when performing phonological perception differ depending on the mother tongue of the subject. For example, with regard to plosives, experimental results suggested that clarity can be manipulated by selectively emphasizing plosive components and manipulating volume.

  The foreign language learning apparatus of the present invention uses such knowledge to present a clear voice to the primary learner to enhance the training effect, and to acquire the acoustic characteristics after having acquired a certain ability Is used to reduce clarity, gradually learn more difficult tasks, and finally acquire the ability close to the native speaker.

(System configuration of the present invention)
FIG. 1 is a conceptual diagram showing an example of a system 1000 using the foreign language learning apparatus of the present invention.

  Referring to FIG. 1, the system 1000 sequentially presents learning tasks of a certain difficulty level to the learner 2, and trains the learner for the step based on answers to the tasks from the learner. The computer 100 for performing the process of giving or presenting the learning subject of the next step to a learner further is provided.

  The computer 100 includes a disk drive 108 for reading information on a recording medium such as a CD-ROM (Compact Disc Read-Only Memory) and an FD drive 106 for reading / writing information on a flexible disk (FD) 116. A computer main body 102, a display 104 as a display device connected to the computer main body 102, a keyboard 110 and a mouse 112 as input devices also connected to the computer main body 102, and a microphone 132 as a voice input device. And a speaker 134 as an audio output device.

  Note that the microphone 132 and the speaker 134 may be a headphone and a microphone worn by the user 2 using a headset.

  The system 1000 further includes a server computer 300 for presenting a task to be given to the learner 2, and a network 400 for connecting the server computer 300 and the computer 100. As the network, for example, the Internet or the like can be used. That is, in the following description, the computer 102 main body will be described as functioning as a client machine. However, the computer 100 itself may hold the function of presenting a problem in the server 300.

  Therefore, in the present invention, the main part is the server computer 300 which is a foreign language learning problem presentation device having a problem presentation function.

  FIG. 2 is a block diagram showing the hardware configuration of the server computer 300. As shown in FIG.

  The server computer 300 includes a computer main body 302, a display 304 as a display device connected to the computer main body 302, and a keyboard 310 and a mouse 312 as input devices also connected to the computer main body 302.

  As shown in FIG. 2, in addition to the disk drive 308 and the FD drive 306, the computer main body 302 includes a CPU (Central Processing Unit) 320, ROM (Read Only Memory), and RAM (Random) connected to the bus BS, respectively. A memory 322 including an Access Memory, a direct access memory device, such as a hard disk 324, and an interface 328 for communicating with the network 400. For example, a CD-ROM 318 is mounted on the disk drive 108. An FD 316 is attached to the FD drive 306.

  As will be described later, when the foreign language learning apparatus of the present invention operates, a database that stores information that is the basis of the operation will be described as being stored in the hard disk 324.

  The CD-ROM 118 may be another medium, such as a DVD-ROM (Digital Versatile Disc) or a memory card, as long as it can record information such as a program installed in the computer main body. In this case, the computer main body 302 is provided with a drive device that can read these media.

  The main part of the foreign language learning apparatus of the present invention is composed of computer hardware and software executed by the CPU 320. Generally, such software is stored and distributed in a storage medium such as a CD-ROM 318 or FD 316, read from the storage medium by the disk drive 308 or FD drive 306, and temporarily stored in the hard disk 324. Alternatively, when the device is connected to the network 400, it is temporarily copied to the hard disk 324 from another server on the network. Then, the data is further read from the hard disk 324 to the RAM in the memory 322 and executed by the CPU 320. In the case of network connection, the program may be directly loaded into the RAM and executed without being stored in the hard disk 324.

  The computer hardware itself and its operating principle shown in FIGS. 1 and 2 are general. Therefore, the most essential part of the present invention is software stored in a storage medium such as the FD 316, the CD-ROM 318, and the hard disk 324.

  As shown in FIG. 1, when the computer 100 is connected to the server 300 via the network 400, a database that stores information that is the basis of the operation is stored in a storage device such as the hard disk 324 in the server 300. When the computer 100 is operated, the computer 100 operates as a client machine, receives the assignment data given to the learner as described below from the server 300, displays or outputs it, and returns the answer data to the server. Do.

  FIG. 3 is a functional block diagram showing the configuration of the foreign language learning apparatus of the present invention.

  As shown in FIG. 3, in the CPU 320, as a functional block, a frequency analysis unit 3202 that performs frequency analysis based on a speech processing program, a clarification processing unit 3204 that performs speech clarification processing, and a foreign language learning A learning control unit 3220 for controlling a problem presentation flow as will be described later is included based on the program.

  In the following description, it is assumed that “phoneme enhancement” is performed by selectively amplifying the phoneme portion in the speech clarification process. However, “phoneme enhancement” can also be performed on the phoneme by selectively attenuating unnecessary portions other than the phoneme.

  Also, in the hard disk 324 connected to the CPU 120 by the bus BS, original voice data by a speaker whose native language is English recorded in an anechoic room or the like is model voice to be presented to the learner 2. The stored original speech database 3242, the arrival level database 3244 storing a correspondence table shown for each phoneme for learning the correspondence between the correct answer rate and the arrival level prepared in advance, the arrival level and acoustic feature portion prepared in advance An amplification degree database 3246 that stores a correspondence table of amplification amounts with respect to sound, and a phonological acoustic model database 3248 that records a phonological acoustic model used when detecting the presence of an acoustic feature such as an opposite phoneme are recorded. It shall be. The phonological acoustic model is not particularly limited. For example, a hidden Markov model can be used.

  Here, the stimulating sound (model voice for learning) to be presented to the learner 2 is selectively amplified with respect to the acoustic feature portion by the clarification processing unit 3204 as will be described later on the data in the original voice database 3242. Generated by doing.

  Further, the data stored in the achievement level database 3244 may be divided into 10% correct answer ratios, for example, 10% up to 70%, and 5% for more than 70%. Setting is also possible. Note that the reaching level is set in advance for each learning phoneme. At this time, the division of the reaching level may be different for each phoneme (acoustic feature portion, for example, an opposite phoneme portion) to be learned.

(Enhancement of acoustic features)
FIG. 4 is a block diagram for explaining the operations of the frequency analysis unit 3202 and the clarification processing unit 3204 described in FIG. 3 in more detail.

Referring to FIG. 4, frequency analysis unit 3 202 receives digitally quantized original voice data, converts it using an algorithm such as FFT (Fast Fourier Transform) or wavelet transform, and performs frequency analysis. The intensity of each frequency component contained in the speech signal is divided and analyzed in time series, and the acoustic features of the phoneme are parameterized. In this part, in addition to the method of performing the sequential calculation at the time of learning as described above, for example, data that has been calculated in advance may be stored.

  Subsequently, the conflict phoneme detection unit 3206 detects an acoustic feature as follows and selectively sends the feature portion to the amplification processing unit 3208. In other words, it is said that the acoustic features used for perception vary depending on phonemes, and a plurality of features are used in combination. In this apparatus, a representative thing is emphasized among a plurality of acoustic features. The following is an example.

      1) In the case of / r /-/ l / conflict, it is considered that the transition of the third formant is used for discrimination. Therefore, formant tracking is performed after frequency analysis, and / r / or / l / phoneme. The frequency processing component 3208 amplifies the frequency component of the frequency band detected as the third formant in the region where is uttered.

      2) In the case of / b / − / v / conflict, it is considered that a pulsed acoustic component called buzz bar existing in / b / is used for discrimination. Therefore, the amplification processing unit 3208 amplifies the portion where the plosive sound exists and all the frequency components around it.

      3) It is considered that / s / − / th / is discriminated by a frequency band (upper limit, lower limit, width, etc.) in which a frictional sound component appears. Therefore, the amplification processing unit 3208 amplifies all frequency components in the region where the frictional sound component exists.

  First, as a premise, in the original voice database 3242, in relation to the digital data of the original voice, where in each original voice data the phoneme component having the acoustic characteristics as described above exists on the time axis. It is assumed that data is also stored.

  That is, the confrontation phoneme detection unit 3206 receives position information of phonological components from the original speech database 3242, and performs amplification processing on the amplification processing unit 3208 when there is no acoustic feature component in the original speech data. It is instructed to output as it is without performing. On the other hand, the confrontation phoneme detection unit 3206 amplifies the signal within a specified range (time) to the amplification processing unit 3208 when any acoustic feature component exists in the original audio data. Give instructions to do. The intensity of amplification in the amplification processing unit 1208 is performed at an amplification factor corresponding to the calculated amplification amount. However, since the volume of the original voice differs for each voice file, an attempt is made to present the average sound pressure of all word sounds and the current presentation. The amount of amplification is weighted by the difference in sound pressure of the word speech, and amplification is performed with the amplification factor obtained by calculation.

  The output audio data may be further subjected to a predetermined compression process as long as the audio quality is not sacrificed, if necessary.

  On the computer 102 side, a digital-to-analog conversion device is used for audio reproduction and is reproduced from the speaker 134.

  FIG. 5 is a diagram showing an example of a sound waveform when sound data stored in the original sound database is reproduced.

  In FIG. 5, the English word speech “LAB” spoken by an American whose native language is English is shown as a waveform.

  FIG. 6A is a diagram showing the result of frequency analysis of the waveform of FIG.

  That is, when the waveform shown in FIG. 5 is subjected to frequency analysis, a voiceprint pattern as shown in FIG. 6A is obtained. In FIG. 6 (a), a thin portion vertically extending around 500 ms is an acoustic component called “buzz bar”. Thus, when the power is weak (in the figure, the intensity of the power is indicated by the darkness of black), “B” may not be perceived and “V” may be perceived.

  FIG. 6B is a diagram showing a voiceprint pattern of a voice in which a plosive sound component is detected and only the plosive sound component portion is amplified.

  In FIG. 6B, the intensity of amplification is moderately amplified in accordance with the preceding voice, and is amplified with an envelope in order to improve the connection with the preceding and following voices. That is, the gain is gradually increased as the plosive portion is approached, and the gain is gradually decreased after the maximum gain.

  In FIG. 6B, the “LA” portion is not amplified, and the volume is increased as a whole by amplifying with the envelope, so that it does not sound harshly loud. However, since the plosive component is greatly amplified, it becomes easier for the listener to perceive “B” and to easily determine “LAB” as a word.

  FIG. 7 is a flowchart for explaining processing performed by the learning control unit 3220 shown in FIG.

  With reference to FIG. 7, first, the learning control unit 3220 performs a test on the learner using, for example, the confrontation phoneme designated by the learner 2 using the original speech, and the learner's achievement (expertise / (Capability) is confirmed (step S100).

  Subsequently, the learning control unit 3220 refers to the achievement level database 3244 and derives a feature enhancement amplification amount that is considered appropriate for the learner from the test result (correct answer rate) (S102). .

  The learning control unit 3202 dynamically generates a learning stimulus in which the acoustic feature is amplified by the amplification amount with respect to the sound data (naturally-spoken speech) in the original sound database 3242 (S104).

  The learner 2 is trained using such training voice stimulation, and the average correct answer rate at that time is obtained (S106).

  If the average correct answer rate satisfies the level-up condition (S108), the learning control unit 3202 reduces the feature enhancement amplification amount by, for example, a predetermined level (S110). On the other hand, if the average correct answer rate satisfies the level down condition (S112), the learning control unit 3202 increases the feature enhancement amplification amount by, for example, a predetermined level (S114).

  When the learning control unit 3202 determines that sufficient results have been achieved by stimulation of the original voice (S116), the learning control unit 3202 ends the learning.

  Here, the “level-up condition” is a condition that, for example, the current achievement level of the learner 2 continuously exceeds the target correct answer rate determined for each phoneme a predetermined number of times, and the “level-down condition” The condition is, for example, that the current achievement level of the learner 2 falls below a prescribed correct answer rate determined for each phoneme a predetermined number of times. The predetermined number of times may be one time, or the predetermined number of times may be set to be different depending on the degree of achievement.

  Here, the method for deriving the feature enhancement amplification amount in step S102 will be described in more detail.

  As described above, after performing the listening test using the original voice or the voice that emphasizes the acoustic features with an appropriate amplification amount, the correspondence table of the correct answer rate and the arrival level prepared in advance in the amplification degree database 3246 Map test results and identify learner achievement levels. The arrival level is calculated for each learning phoneme. After that, it is applied to the correspondence table of the achievement level and amplification amount prepared in advance in the achievement level database 3244, and the amplification amount appropriate for the learner is calculated.

  By giving the learning task as described above, it is possible to present the learning stimulus sound with enhanced acoustic features to the learner so that the learning effect can be effectively improved.

  In the following, experimental conditions and experimental results that suggest the influence of the enhancement of acoustic features on the listening results will be described in more detail for words and the like that have phonemic conflicts as described above.

[Experimental result]
Due to phonological conflict, acoustic differences and clues at the time of listening generally differ among listeners of different native languages. Therefore, for example, / b / and / v / (hereinafter abbreviated as BV) and / s / and / θ / (hereinafter abbreviated as STH) are phonemes that are difficult to perceive for Japanese native speakers. It may be different.

  Therefore, in the following experiments, Japanese native speakers (hereinafter abbreviated as JA) and American native speakers (hereinafter abbreviated as AE) are targeted, and the characteristics are different with respect to US word speech that opposes RL, BV, and STH. An experiment was conducted to add clarity and measure clarity. In addition, as a result of a preliminary experiment conducted with native speakers of American language, it was confirmed that the phoneme was affected by the sound pressure presented, so this was also verified.

(1 Experimental method)
(1.1 Stimulation)
Experiments were performed using English word pairs of the smallest phoneme pairs that oppose each other in three phonemes: RL pairs (right-1ight etc.), BV pairs (base-base etc.), and STH pairs (mouse-mouth etc.). For each confrontation, a total of 220 words of 50, 30, and 30 pairs (a total of 110 pairs) spoken by two American native speakers (one male and one female) was used as a stimulus voice. The voice recorded in the anechoic room was saved as a file in PCM (Pulse Code Modulation) format with an accuracy of 44.1 kHz and 16 bits for each word.

  As a stimulus for adding noise, the average of the sound pressure level (A characteristic) peak value when each word is output through headphones is 59 dB in the RL and STH conflicts, and 65 dB in the BV conflict. The amplitude was adjusted as follows.

  Adjusting the amplitude so that the peak value of the sound pressure level (A characteristic) when white noise and pink noise generated by the noise generator are output through headphones is in accordance with the SN of each condition, added to the sound used in this experiment did. Noises with a duration longer by 200 ms before and after the voice were superimposed.

  FIG. 8 is a diagram showing the SN ratio used as an experimental condition.

  Further, as a stimulus for measuring the influence of the presented sound pressure on the clarity, the average of the sound pressure level (A characteristic) peak when each word is output through the headphones is from 39 dB to 69 dB in each phoneme confrontation. The amplitude was adjusted in 5 dB steps.

(1.2 Experiment participants)
In the JA experiment, 11 university students who were native speakers of Japanese and had no experience of staying abroad for more than 3 months participated in the experiment. In the AE experiment, three American native speakers from the age of 23 to 43 participated in the experiment. All confirmed that they had normal hearing.

(1.3 Procedure)
The actual saponification was carried out in a soundproof room for 3 days. Two English words that form the smallest phoneme pair were visually presented on the computer screen, and at the same time, either word was presented in both ears via headphones. Participants in the experiment judged and selected which word was the word pair on the screen.

(Noise-added voice session)
It was divided into two days according to the type of added noise. Each section consisted of two sections for each sound, and the order of the speakers was constant. Each section is composed of blocks for each phoneme confrontation including all SN-speech sounds, and presented in the order of RL, BV, and STH confrontation. All voice stimuli were presented in a random order within each block, and no feedback was given regarding the correctness of the answers.

(Sound pressure fluctuation session)
A sound pressure fluctuation session was conducted after the noise-added speech session. Except for the different stimuli, the structure and method were the same as for the noisy speech session.

(2 results)
(JA experiment)
FIG. 9 is a diagram showing the result of a noise-added voice session in the JA experiment.

  In any phonological conflict, it was shown that the correct answer rate tends to decrease when the SN ratio decreases.

  A two-factor ANOVA was performed for each phoneme confrontation, with the noise type and the signal-to-noise ratio as factors within the subject, and the correct answer rate as a dependent variable. In the BV conflict, the -9 dB condition of the white noise condition was excluded from the analysis. As a result, the main effect of the S / N ratio factor is significant in any phoneme conflict ([F (6,60) = 24.950, p <0.01], [<0.01] in the RL, BV, and STH phoneme conflicts, respectively). F (7,70) = 18.641, P <0.01], [F (6,60) = 32.15, P <0.01]) is the main effect and interaction of noise type factors Both were not significant.

  Next, FIG. 10 is a diagram showing a result of a sound pressure fluctuation session in the JA experiment.

  A two-factor ANOVA was performed with phonological confrontation and presented sound pressure as factors within the subject, and the correct answer rate as the dependent variable. As a result, the main effect of the presenting sound pressure factor was significant [F (6,60) = 10.503, P <0.01]. Although the main effect and interaction of phonological confrontation factors were not significant, when the correct answer rates of the 39 dB condition and the 63 dB condition were compared, the BV confrontation showed a greater change in the correct answer rate than the other phonological confrontation.

(AE experiment)
FIG. 11 is a diagram illustrating a result of a noise-added speech session in each phoneme conflict in the AE experiment. In any phoneme conflict, it was shown that the correct answer rate tends to decrease as the SN ratio decreases.

  Next, FIG. 12 is a diagram showing the results of the sound pressure fluctuation session in the AE experiment. The RL and STH conflicts showed little change in the presented sound pressure range used in the experiment, but the BV conflict showed that the correct answer rate was susceptible to the presented sound pressure.

  Summarizing the above analysis results, the correct answer rate decreased with a decrease in the S / N ratio in all phoneme confrontations for both Japanese native speakers and American native speakers. Furthermore, the following relations were clarified in relation to the influence of noise addition on the mother tongue, phoneme pair, and presented sound pressure.

(Native and non-native)
For American native speakers, there is a signal-to-noise ratio range that is unlikely to be affected by noise addition in phoneme pairs other than BV, whereas for Japanese native speakers, the correct answer rate tends to decrease with slight noise addition. Indicated.

  In addition, the influence of the noise type may differ depending on the native language of the experiment participant (example: comparison between AE-15 dB condition and JA-9 dB condition of RL conflict). This suggests that the acoustic features used for perception differed depending on the mother tongue.

(Phoneme pair)
The effect of noise was different for each phoneme pair. The RL conflict was relatively resistant to the noise used in the actual sapon, but the BV conflict was greatly affected by the addition of a little noise, and the correct answer rate decreased at an almost constant rate in the STH conflict. This indicates that the acoustic features used for discrimination differ depending on phoneme pairs, and even when the same noise is added, different effects are exerted.

(Present sound pressure)
In both cases of Japanese native speakers and American native speakers, the correct answer rate greatly decreased when the sound pressure presented was low in the BV confrontation. It shows that the perception of BV sound is greatly affected by the presented sound pressure.

(Modification of the configuration of the clarification processing unit 3204)
The configuration of the clarification processing unit 3204 has been described with reference to FIG. 4, but may have other configurations as described below.

  FIG. 13 is a diagram illustrating a modification of the configuration of the clarification processing unit 3204.

  That is, in FIG. 4, the position information of the characteristic phoneme component is received from the original speech database 3242. However, as described below, the feature phoneme component is selectively amplified while detecting the feature phoneme component as needed. Also good.

  That is, when no acoustic feature component is detected, the conflict phoneme detection unit 3206 shown in FIG. 13 does not perform any processing as it is and sends data to the signal selection unit 1210 that performs subsequent processing. In contrast, when the acoustic feature component is detected, the allelic phoneme detection unit 3206 sends a signal to the amplification processing unit 3208, and the amplification processing unit 3208 amplifies the signal within the detected range (time). The data is sent to the signal selection unit 3210 that performs the subsequent processing. Also in the example of FIG. 13, the amplification intensity in the amplification processing unit 1208 is performed at an amplification factor corresponding to the calculated amplification amount. However, since the volume of the original voice differs for each voice file, the average sound of all the word voices The amplification amount may be weighted by the difference between the pressure and the sound pressure of the word speech to be presented at present, and the amplification may be performed with the amplification factor obtained by calculation.

  The signal selection unit 3210 selectively synthesizes the non-amplified data sent from the allelic phoneme detection unit 3206 and the amplified data from the amplification processing unit 1208, and outputs the resultant as voice data with enhanced features. . Further, in the example of FIG. 13, the output audio data may be subjected to a predetermined compression process within a range that does not sacrifice the quality of the sound product, if necessary.

  With such a configuration, the same processing as shown in FIG. 4 can be performed.

  As described above, in the language learning device or the foreign language learning device of the present invention, by applying an acoustic operation based on the clarity of speech as an evaluation criterion, the primary learner is presented with an easily perceptible problem. Since a difficultly perceptible task is dynamically generated or created in advance and presented to a learner who has mastered the above, an effective learning task can be presented.

  The embodiment disclosed this time should be considered as illustrative in all points and not restrictive. The scope of the present invention is defined by the terms of the claims, rather than the description above, and is intended to include any modifications within the scope and meaning equivalent to the terms of the claims.

It is a conceptual diagram which shows an example of the system 1000 using the foreign language learning apparatus of this invention. It is a figure which shows the hardware constitutions of the server computer 300 in a block diagram format. It is a figure which shows the structure of the foreign language learning apparatus of this invention with a functional block. FIG. 4 is a block diagram for explaining in more detail the operations of the frequency analysis unit 3202 and the clarification processing unit 3204 described in FIG. 3. It is a figure which shows an example of the audio | voice waveform at the time of reproducing | regenerating the audio | speech data stored in an original audio | voice database. It is a figure which shows the result of having frequency-analyzed the waveform, and the result of having selectively amplified. 4 is a flowchart for explaining processing performed by a learning control unit 3202 illustrated in FIG. 3. It is a figure which shows the S / N ratio used as experiment conditions. It is a figure which shows the result of the noise addition voice session in JA experiment. It is a figure which shows the result of the sound pressure fluctuation | variation session in JA experiment. It is a figure which shows the result of the noise addition audio | voice session in each phoneme confrontation in AE experiment. It is a figure which shows the result of the sound pressure fluctuation | variation session in AE experiment. FIG. 18 is a diagram illustrating a modification of the configuration of the clarification processing unit 3204.

Explanation of symbols

  100 computer, 102 computer main body, 104 display, 106 FD drive, 108 disk drive, 110 keyboard, 112 mouse, 118 CD-ROM, 132 microphone, 134 speaker, 300 server computer, 302 computer main body, 304 display, 306 FD drive, 308 disk drive, 310 keyboard, 312 mouse, 318 CD-ROM, 320 CPU, 322 memory, 324 hard disk, 328 interface, 1000 system, 3202 frequency analysis unit, 3204 clarification processing unit, 3220 learning control unit, 3242 original speech database , 3244 reaching level database, 3246 amplification degree database, 3248 phonological acoustic model Database.

Claims (5)

A language learning device,
Storage means for storing the model voice and storing the learner's reach and the enhancement rate of the acoustic feature portion included in the model voice in association with each other;
For the model voice, a stimulus sound synthesizing means for synthesizing selectively emphasized stimulate speech said acoustic characteristic portion of the model voice given emphasis ratio,
The stimulating sound synthesizing unit includes a detecting unit for specifying a position on the time axis where the acoustic feature portion exists in each of the model sounds input from the storage unit, and the stimulating sound synthesizing unit includes Selectively accentuate the acoustic feature according to the identified location;
An interface means for outputting an audio signal for giving the stimulating sound to the learner as a task to an audio output device for reproducing the stimulating sound and receiving an answer from the learner;
To calculate the correct answer rate of the learner with respect to the stimulus speech given as the problems by specifying the enhancement factor if, by controlling the stimuli combining means, giving said stimulus voice to the learner the percentage of correct answers, the in response to reaching levels up condition is satisfied, which is predetermined according to, further a learning control means for updating to reduce the enhancement rate specified for the stimuli combining means A language learning device. The storage means stores, in association with the model sound, position information indicating in which position on the time axis of the model sound a phonological component having the acoustic feature for the model sound exists. And
The language learning apparatus according to claim 1, wherein the detection unit specifies a position on a time axis where the acoustic feature portion exists based on the position information. The stimulating sound synthesizing unit further includes a frequency analyzing unit that performs frequency analysis on the model voice and parameterizes an acoustic characteristic of a phoneme in the model voice,
The language learning apparatus according to claim 1, wherein the detection unit specifies a position on a time axis where the acoustic feature portion exists based on an analysis result of the frequency analysis unit. The learning control means, the correct answer rate when given the stimulus voice to the learner, the in response to reach level down condition is satisfied is determined in advance according to increase the emphasis rate The language learning apparatus according to claim 2 , which is updated as follows. The acoustic feature portion corresponds to an opposing phoneme portion;
The storage means classifies and stores the degree of achievement for each of the opposing phonemes,
The learning control means causes the stimulation sound synthesizing means to generate the stimulation sound with reference to the degree of achievement stored in the storage means for each of the designated allophones,
Said interface means, said outputs an audio signal for presenting against the generated the stimuli voice the learner, language learning device according to claim 2 or 3.