JP2020160388A - Scoring support device, method thereof, and program - Google Patents

Abstract

To provide a scoring support device, a method thereof, and a program, which are applicable also to reading voices even if any correct reply to a question is not uniquely known, different from prior art applied only to the reading voice replying to a sentence only when a correct answer to the question is uniquely known.SOLUTION: The scoring support device has a scoring support unit that calculates the evaluation of a voice reply using a voice recognition result of voice reply to question sentence and a piece of correct reply information corresponding to a question sentence including at least one correct reply sentence to the correct answer sentence list.SELECTED DRAWING: Figure 1

Description

The present invention relates to a scoring support device, a method thereof, and a program that support scoring of voice answers to question sentences.

In Non-Patent Document 1, speech recognition is performed on the speech recognition result of the non-native speaker model by a grammar that replaces phonemes with the native speaker model, and pronunciation error candidates are output.

Zhang Hao, Daisuke Saito, Nobuaki Minematsu, Satoshi Kobashigawa, "Modeling of Japanese English Pronunciation Diversity and Application to Automatic Phoneme Error Detection", Proceedings of the Acoustical Society of Japan, 2-Q-4, 2018

A scoring support device that uses conventional technology to score whether there is a pronunciation error in the spoken answer (of the learning language) of a non-native speaker can be considered.

However, in the prior art, since it is necessary to uniquely know the correct answer (correct sentence) to the question, it can be applied only to the reading voice corresponding to the correct sentence.

An object of the present invention is to provide a scoring support device, a method thereof, and a program that can be applied to other than the reading voice with a correct sentence.

In order to solve the above problems, according to one aspect of the present invention, the scoring support device corresponds to a voice recognition result of a voice answer to a question sentence and a correct sentence list including at least one correct answer sentence corresponding to the question sentence. Includes a scoring support department that requests evaluation of voice responses using correct answer information.

According to the present invention, since it can be applied to other than the reading voice with the correct sentence, the range of learning materials is expanded.

The functional block diagram of the scoring support device which concerns on 1st Embodiment. The figure which shows the example of the processing flow of the scoring support apparatus which concerns on 1st, 3rd to 6th Embodiment. The figure which shows the example of data. The functional block diagram of the scoring support device which concerns on 2nd Embodiment. The figure which shows the example of the processing flow of the scoring support apparatus which concerns on 2nd Embodiment. The functional block diagram of the scoring support device which concerns on 3rd Embodiment. The figure which shows the example of the processing flow of the scoring support apparatus which concerns on 4th to 6th Embodiment.

Hereinafter, embodiments of the present invention will be described. In the drawings used in the following description, the same reference numerals are given to the components having the same function and the steps for performing the same processing, and duplicate description is omitted. Unless otherwise specified, the processing performed for each element of a vector or matrix shall be applied to all the elements of the vector or matrix.

<Points of each embodiment>
In the first embodiment, only the keywords included in the correct sentence and the keywords included in the voice recognition result are narrowed down, and the matching rate is used as the score.

In the case of a match rate based on keywords, a small error that some characters in the keywords included in the speech recognition result are different will greatly reduce the overall evaluation. In order to solve this problem, in the second embodiment, the matching rate for each character is used as a score.

In the first embodiment, it is difficult to control the keyword selection method. Therefore, in the third embodiment, in order to perform the evaluation according to the intention of the questioner, the correct sentence is expressed as a regular expression, the speech recognition result and the regular expression are compared, and the comparison result is used as a score.

In the fourth embodiment, in order to reduce the preparation cost of keywords, regular expressions, etc. corresponding to the correct answer sentence, the correct answer accuracy is scored by comparing the speech recognition result with the correct answer sentence.

Considering a scene used in a classroom or the like, even if the speaker speaks correctly, there may be an adverse effect due to ambient voice noise. Therefore, in the fifth embodiment, the recognition rate that does not consider the insertion error is used as the score. In addition, since there is a possibility of mistakes in insertion errors in sentences other than the beginning and end of sentences, the insertion errors should be ignored only at the beginning and end of sentences. In addition, in the case of a long sentence, there is a possibility that there is a gap between the phrases in the sentence, so the insertion error between the end of the phrase and the beginning of the phrase may be ignored.

When the score itself is presented to the learner, if an unreasonably bad point is given due to the influence of noise or the like, the scoring support device may have a bad impression. Therefore, in the seventh embodiment, classification is performed according to the range of the score, and the classification result is presented as the evaluation result.

<First Embodiment>
In the first embodiment, only the keywords included in the correct sentence list including at least one correct sentence for the question sentence are narrowed down, the matching rate with the speech recognition result is obtained, and the obtained matching rate is used as a score.

FIG. 1 is a functional block diagram of the scoring support device according to the first embodiment, and FIG. 2 is a diagram showing an example of the processing flow.

The scoring support device according to the first embodiment includes a voice recognition unit 110, a keyword creation unit 120, and a scoring support unit 130.

The scoring support device according to the first embodiment inputs a correct answer sentence list and a voice signal including a reply voice, evaluates the reply voice, and outputs a scoring result.

The scoring support device is a special program configured by loading a special program into a known or dedicated computer having, for example, a central processing unit (CPU), a main storage device (RAM: Random Access Memory), and the like. It is a device. The scoring support device executes each process under the control of the central processing unit, for example. The data input to the scoring support device and the data obtained by each process are stored in the main storage device, for example, and the data stored in the main storage device is read out to the central processing unit as needed. It is used for processing. At least a part of each processing unit of the scoring support device may be configured by hardware such as an integrated circuit. Each storage unit included in the scoring support device can be configured by, for example, a main storage device such as RAM (Random Access Memory) or middleware such as a relational database or a key-value store. However, each storage unit does not necessarily have to be provided inside the scoring support device, and is composed of an auxiliary storage device composed of semiconductor memory elements such as a hard disk, an optical disk, or a flash memory, and is a scoring support device. It may be configured to be provided outside the above.

<Scoring support method>
In this embodiment, scoring support is provided as follows.

(i) Present the question sentence to a speaker whose native language is not a certain language (hereinafter, also referred to as a learning language) (hereinafter, also referred to as a non-native speaker or a learner) in some way. For example, a paper on which the problem statement is described is presented, or the problem statement is displayed on an output device such as a display.

(ii) The learner answers the question sentence by utterance in the learning language and picks up the answer voice.

(iii) Scoring is performed based on the matching rate between the correct answer sentence for the question sentence and the voice recognition result of voice recognition of the voice signal including the answer voice, and the scoring result is presented to the learner in some way.

For example, the question sentence is "Please answer the English translation of'I like you.'" And the correct sentence is "I love you". A list containing at least one correct sentence is called a correct sentence list. Since the number of correct sentences for a question sentence is not limited to one, the correct sentence list includes one or more correct sentences (see FIG. 3). For example, "I like you" may be added as a correct sentence to the correct sentence list in addition to the above-mentioned correct sentence. If there are multiple question sentences, prepare a correct answer sentence list for each question sentence.

Hereinafter, the processing of each part for realizing the above-mentioned scoring support will be described.

<Voice recognition unit 110>
Input: Voice signal output including answer voice: Voice recognition result (sentence or voice recognition processing unit)
Processing content:
The voice recognition unit 110 performs voice recognition on the voice signal (S110), and outputs the voice recognition result as text.

Various methods can be considered for voice recognition. For example, a robust method for non-native speaker voice may be used by using the method of Reference 1.
(Reference 1) Ryo Masumura, Yu Kabashima, Takashi Moriya, Satoshi Kobashikawa, Yoshikazu Yamaguchi, Yuji Aono, "Study of neural acoustic model for Japanese English using voice data of native Japanese and native English", Acoustical Society of Japan Proceedings, 1-2-2, 2018

The voice recognition result output here is a part of a sentence or a sentence corresponding to a processing unit of voice recognition. In this example, the voice recognition result is "I love you" (see FIG. 3).

<Keyword creation unit 120>
Input: Correct sentence list Output: Keyword list for correct sentence list Processing content:
The keyword creation unit 120 extracts keywords from the correct sentences included in the correct sentence list and creates a keyword list. Create one keyword list for one correct sentence, and one keyword list contains one or more keywords. In this example, the keyword list is a list that includes the keyword "love" (see FIG. 3).

Various methods can be considered as the keyword extraction process. For example, keywords may be extracted based on rules. Two rules will be illustrated below.
(Rule 1) A rule is set in advance to extract predetermined part of speech such as nouns, prepositions, and verbs as keywords, and keywords are extracted from correct sentences according to this rule. In addition to the part of speech, it may be preset as a rule to have at least one predetermined language attribute.
(Rule 2) Keywords to be extracted are set in advance, and those that match the set keywords are extracted from the correct sentence.

For example, in the case of rule 1, morphological analysis is performed on the correct sentence, and only a predetermined part of speech is extracted as a keyword based on the morphological analysis result. Keywords may be extracted by combining the above rules 1 and 2.

The keyword creation unit 120 lists the extracted keywords and creates a keyword list. If the total number of keywords included in the keyword list is small, the extracted keyword and the synonym may be used as a new keyword and added to the keyword list by using a thesaurus or a thesaurus. From the learner's point of view, homophones that are easily confused as speech recognition may be treated as new keywords. For example, the extracted keyword or a homonym for the synonym thereof may be used as a new keyword and added to the keyword list. In this case, phonetic symbols and the like are added to the correct answer sentences, the extracted keywords, and their synonyms, and the homonyms can be obtained by these phonetic symbols and the like.

If a correct answer sentence list corresponding to the question sentence is given in advance, keywords may be extracted from the correct answer sentence in advance and a key third list may be created before performing the voice recognition process. By creating the keyword list in advance in this way, the processing time for scoring support can be reduced.

<Scoring Support Department 130>
Input: Speech recognition result, keyword list output for each correct sentence included in the correct sentence list: Score result processing content:
The scoring support unit 130 obtains a scoring result using the voice recognition result and the keyword list for each correct sentence (S130). In the present embodiment, the scoring support unit 130 calculates the matching rate between the voice recognition result and the keyword list for each correct sentence, and obtains a score based on the matching rate. For example, in the example of FIG. 3, since the voice recognition result includes all the keywords included in the keyword list (1), the calculated match rate is 100%, and the score is 100 out of 100 points.

In this embodiment, in order to prevent the learner's motivation from being lowered due to the deduction of points, the order of appearance of the keywords in the speech recognition result and the score are irrelevant.

For example, the scoring support unit 130 may calculate the matching rate for a plurality of keyword lists and output the best matching rate among the calculated matching rates as the scoring result.

<Effect>
In the case of the present embodiment, since it is sufficient to include the keyword corresponding to the correct answer sentence in the voice recognition result, it can be applied to other than the reading voice with the correct answer sentence, and the range of learning teaching materials is expanded.

Further, in the prior art, pronunciation information corresponding to the correct sentence is required, so that a cost for that is required, but it is not necessary in the configuration of the present embodiment. In addition, in the prior art, the learner's motivation tends to decrease because it is close to the deduction method that is not evaluated unless it is pronounced correctly, but with the configuration of this embodiment, even if it is a little wrong, it is evaluated with a high score. It can be expected to maintain the motivation of learners. Furthermore, evaluation can be performed even when the correct sentence list contains a plurality of correct sentences.

<Modification example>
In the present embodiment, the keyword creation unit 120 extracts keywords from the correct answer sentences included in the correct answer sentence list and creates a keyword list. However, the keyword list is separately prepared for each correct answer sentence in advance regardless of the keyword creation unit 120. May be prepared and given to the scoring support unit 130.

In the present embodiment, the order of appearance of the keywords in the voice recognition result and the score are irrelevant, but the score may be obtained in consideration of the order of appearance of the keywords in the voice recognition result. In this case, the keyword list contains information in the order of appearance. For example, the keywords included in the keyword list are listed in the order of appearance (the order of the keywords included in the keyword list corresponds to the order of appearance of the keywords in the correct sentence), and the scoring support unit 130 has one or more keywords included in the keyword list. The evaluation is based on one or more of the prefix match and the suffix match between the keyword and the speech recognition result. Further, the scoring support unit 130 may request an evaluation based on either a partial match or an exact match in consideration of the order of appearance of the keyword and the voice recognition result.

In this embodiment, the keyword extraction is not performed on the speech recognition result, but after applying the keyword extraction to the speech recognition result, the keyword extraction result for the speech recognition result and the keyword extraction result for the correct sentence are compared. You may. For example, keyword extraction may be performed using the part of speech information included in the word information of the speech recognition result. In this case, the scoring support device includes the answer sentence keyword creation unit 140 (indicated by a broken line in FIG. 1). For example, the answer sentence keyword creation unit 140 extracts a keyword from the voice recognition result and creates an answer sentence keyword list. At this time, keywords may be extracted based on preset rules. As the preset rule, rule 1, rule 2, and the like described in the keyword creation unit 120 of the present embodiment can be considered. The scoring support unit 130 requests an evaluation based on the answer sentence keyword list and the keyword list corresponding to the correct answer sentence. For example, the match rate between the keyword included in the answer sentence keyword list and the keyword included in the keyword list for each correct sentence is calculated, and the score is obtained based on the match rate. At this time, the order of appearance may be irrelevant to the score, or may be related.

Since there are cases where the voice recognition result is not capitalized, all the correct sentences may be converted to lowercase for matching. Similarly, since there are cases where the speech recognition result does not recognize symbols such as punctuation marks, commas, periods, exclamation marks, question marks, and apostrophes, the above symbols may be removed from the correct sentence. In addition, in order to improve the motivation of learners, the keywords of the answer sentence / correct answer sentence and their synonyms and homophones are treated the same. You may calculate the match rate with. For example, the scoring support device includes a character processing unit 150 (indicated by a broken line in FIG. 1). The character processing unit inputs the voice recognition result and the correct sentence list, and processes the voice recognition result and the correct sentence list based on the preset character processing rules. For example, the above-mentioned rule can be considered as the character processing rule. In other words, the character processing rule includes at least one of (i) converting uppercase letters to lowercase letters and (ii) removing pre-specified symbols.

For speech recognition, the learner's vocal motivation may be increased by using a speech recognition model corresponding to the pronunciation of a non-native speaker. For example, prepare a database (hereinafter, also referred to as a non-native speaker database) that stores a large number of sets of transcribed text (correct text) corresponding to voice data by non-native speakers, and refer to the non-native speaker database. , Learn a voice recognition model corresponding to the pronunciation of a non-native speaker using non-native speaker data (voice data by a non-native speaker and the corresponding transcript). The speech recognition model consists of a non-native speaker acoustic model learned from the speech-text pair and a language model learned from the text of the corresponding language. You may learn the speech recognition model corresponding to the pronunciation of the non-native speaker from the non-native speaker database, or you can input the speech recognition model by the native speaker and tune the speech recognition model by the native speaker. A speech recognition model corresponding to the pronunciation of a non-native speaker may be created.

In order to increase the motivation of the learner, the scoring support unit 130 may be configured to allow a difference of the number of characters or less specified in advance in the evaluation of the keyword and the voice recognition result. For example, the difference between the voice recognition result corresponding to the pronunciation error and the correct sentence in 2 to 3 character units may be allowed and regarded as the correct answer.

In this embodiment, the voice response of a non-native speaker is scored and the evaluation result is output. However, it is not limited to scoring non-native speakers' voice responses. If the question sentence is set, the correct answer sentence list including the correct answer sentence corresponding to the question sentence is given, the answer voice for the question sentence is input, the answer voice is evaluated, and the score result is output. It may be a good score.

<Second embodiment>
The part different from the first embodiment will be mainly described.

In the case of the first embodiment, since the match rate is calculated for each keyword, the evaluation is greatly lowered due to a subtle error that the voice recognition result includes erroneous recognition and some characters of the keyword are different. In the present embodiment, the matching rate for each character is used as a score.

FIG. 4 is a functional block diagram of the scoring support device according to the second embodiment, and FIG. 5 is a diagram showing an example of the processing flow.

The scoring support device according to the first embodiment includes a voice recognition unit 110, a keyword creation unit 120, a scoring support unit 230, and a character unit division unit 240.

<Character unit division 240>
Input: Voice recognition result, keyword output included in keyword list: Character list of voice recognition result, character list for each keyword included in keyword list Processing content:
The character unit division unit 240 divides the voice recognition result and the keywords included in the keyword list into character units (S240), and creates a character unit list (character list).

<Scoring Support Department 230>
Input: Character list of voice recognition result, character list output for each keyword: Score result processing content:
The scoring support unit 230 obtains the scoring result by using the character list of the voice recognition result and the character list for each keyword (S230). In the present embodiment, the scoring support unit 230 collates the character list of the voice recognition result with the character list for each keyword, and calculates the matching rate for each character. In addition, in the character list of the voice recognition result, all the matching parts of the character list of the keyword are collated. For example, the match rate may be calculated by a method such as dynamic programming (DP matching) or prefix match or suffix match (Reference 2), and the best match rate may be used as the scoring result.
(Reference 2) Seiichi Nakagawa, Ritsuji Ito, "Evaluation by Continuous Numeric Speech Recognition of Extended Continuous DP Method", IEEJ Journal C, 1988, Vol. 108, No. 10, p.834-841

The scoring support unit 230 uses the calculated matching rate as the scoring result. The match rate is calculated for each of the correct sentences included in the correct sentence list and each keyword included in the keyword list for the correct sentence, and the best match rate among the calculated match rates may be used as the scoring result.

For example, the voice recognition result is "have", the character list of the voice recognition result is "h", "a", "v", "e", the keyword is "love", and the character list of the keyword is. In the case of "l", "o", "v", and "e", 2 out of 4 letters match, the match rate is 50%, and the score is 50 out of 100 points.

<Effect>
With such a configuration, the same effect as that of the first embodiment can be obtained. In addition, finer scoring can be done that reflects subtle errors.

<Modification example>
In the present embodiment, the input of the character unit division unit 240 is a keyword included in the keyword list, but instead, the correct answer sentence included in the correct answer sentence list may be input (indicated by a broken line in FIG. 4). .. In that case, the character unit division unit 240 divides the correct answer sentence included in the correct answer sentence list into character units, creates a character unit list (character list), and outputs the correct answer sentence. Further, in this case, the scoring support unit 230 inputs the character list of the voice recognition result and the character list for each correct sentence, collates the character list of the voice recognition result with the character list for each correct sentence, and matches the match rate for each character. Is calculated, and the best match rate is used as the scoring result. In the case of this configuration, the scoring support device does not have to include the keyword creation unit 120.

By combining this modification and the second embodiment, the matching rate for each correct sentence and the matching rate for each keyword may be obtained, and the best matching rate may be used as the scoring result.

<Third Embodiment>
The part different from the first embodiment will be mainly described.

With the method of extracting keywords from correct sentences, it may be difficult to evaluate according to the intention of the questioner. In the present embodiment, the correct answer sentence is expressed as a regular expression, the voice recognition result is compared with the regular expression of the correct answer sentence, and the score when they match is used as the score.

As the regular expression, any regular expression may be defined and used.

FIG. 6 is a functional block diagram of the scoring support device according to the third embodiment, and FIG. 2 is a diagram showing an example of the processing flow.

The scoring support device according to the third embodiment includes a voice recognition unit 110 and a scoring support unit 330.

The scoring support device according to the third embodiment inputs a list of correct sentences of regular expressions and a voice signal including a reply voice, evaluates the reply voice, and outputs a scoring result.

<Scoring Support Department 330>
Input: Speech recognition result, correct sentence list including regular expression of correct sentence Output: Score result Processing content:
The scoring support unit 330 obtains a scoring result using the voice recognition result and the correct sentence list including the regular expression of the correct sentence (S330). In the present embodiment, the scoring support unit 330 obtains the scoring result based on whether or not the voice recognition result matches the regular expression of the correct answer sentence included in the correct answer sentence list (correct answer or incorrect answer). For example, if the regular expression of the correct sentence is "I love *" and the voice recognition result is "I love you, too", the voice recognition result matches the regular expression of the correct sentence (correct answer), so the score is 100 points. The score is 100 points.

<Fourth Embodiment>
The part different from the first embodiment will be mainly described.

In this embodiment, in order to reduce the preparation cost of the keywords of the first embodiment and the second embodiment and the regular expression of the third embodiment, the correct answer sentence itself included in the correct answer sentence list is compared with the voice recognition result. Therefore, the accuracy of the correct answer is calculated and used as a score.

FIG. 7 is a functional block diagram of the scoring support device according to the fourth embodiment, and FIG. 2 is a diagram showing an example of the processing flow.

The scoring support device according to the fourth embodiment includes a voice recognition unit 110 and a scoring support unit 430.

<Scoring Support Department 430>
Input: Speech recognition result, correct sentence list output: Score result Processing content:
The scoring support unit 430 obtains the scoring result by using the voice recognition result and the correct sentence list including the correct sentence (S430). The scoring support unit 430 matches the voice recognition result for each correct answer sentence included in the correct answer sentence list, obtains the correct answer accuracy, and uses it as a score. Matching is performed by a method such as dynamic programming (DP matching) so as to match the correct sentence as much as possible (see Reference 2).

The correct answer accuracy consists of the correct answer rate and the recognition accuracy, and may be in word units or character units. However, since the accuracy of the correct answer is higher in the character unit, the character unit is preferable in consideration of the learner's motivation. For example, if the correct sentence is "I love you" and the voice recognition result is "I have you", 2 out of 3 words match in word units, so the score is 66 out of 100 and the letters. In the unit, 6 out of 8 characters match, so the score can be 75 out of 100.

Also, a table of characters with similar pronunciation (for example, sh → s) may be prepared and the difference may be allowed.

<Fifth Embodiment>
The part different from the fourth embodiment will be mainly described.

Considering the scene used in the classroom, even if the speaker speaks correctly, there is an adverse effect due to the surrounding voice noise, so the recognition rate that does not consider the insertion error is used as the score.

FIG. 7 is a functional block diagram of the scoring support device according to the fifth embodiment, and FIG. 2 is a diagram showing an example of the processing flow.

The scoring support device according to the fifth embodiment includes a voice recognition unit 110 and a scoring support unit 530.

<Scoring Support Department 530>
Input: Speech recognition result, correct sentence list output: Score result Processing content:
The scoring support unit 530 obtains the scoring result by using the voice recognition result and the correct sentence list including the correct sentence (S530). The scoring support unit 530 matches the voice recognition result for each correct answer sentence included in the correct answer sentence list without considering the insertion error, obtains the correct answer accuracy, and sets the score. As in the fourth embodiment, matching is performed by a method such as dynamic programming (DP matching) so as to match the correct sentence as much as possible, and the score may be in word units or character units. For example, if the correct sentence is "I love you" and the voice recognition result is "ah I la have you", "ah" and "la" are ignored as insertion errors. The accuracy of the correct answer is 100% for both words and letters, and the score is 100 out of 100.

<Modification example>
Insertion errors are likely to occur at the beginning and end of sentences, and the voice in the utterance sentence may be a mere error, not an insertion error. Therefore, in this modification, the insertion error is ignored only at the beginning and end of the sentence. Also, in the case of a long sentence (for example, 5 words or more), there is a possibility that there is a gap between the phrases in the sentence, so the insertion error between the phrase end and the phrase beginning may be ignored.

<Scoring Support Department 530>
Input: Speech recognition result, correct sentence list output: Score result Processing content:
The scoring support unit 530 obtains the scoring result by using the voice recognition result and the correct sentence list including the correct sentence (S530). The scoring support unit 530 matches the voice recognition result for each correct sentence included in the correct sentence list without considering the insertion error at the beginning and end of the sentence, obtains the accuracy of the correct answer, and uses it as a score. The positions of the beginning and end of the sentence may be before the first word of the correct sentence and after the last word. For example, if the correct sentence is "I love you" and the speech recognition result is "ah I la have you", the "ah" at the beginning of the sentence is ignored as an insertion error and is located in the sentence excluding the beginning and end of the sentence. "la" is judged as an incorrect answer. For example, in the case of word units, "ah" is ignored and the score is 50 out of 100.

If the insertion error occurs only at either the beginning or the end of the sentence, the matching may be performed without considering only one of the beginning and the end of the sentence.

<Sixth Embodiment>
The part different from the fourth embodiment will be mainly described.

When the score itself is presented to the learner, the learner has a bad impression on the scoring support device when an unreasonably bad point is given due to the influence of noise or the like. Therefore, in the present embodiment, not the score itself but the score obtained internally is classified according to a predetermined rule, and the classification result is presented as the evaluation result.

FIG. 7 is a functional block diagram of the scoring support device according to the sixth embodiment, and FIG. 2 is a diagram showing an example of the processing flow.

The scoring support device according to the sixth embodiment includes a voice recognition unit 110 and a scoring support unit 630.

<Scoring Support Department 630>
Input: Speech recognition result, correct sentence list, classification table output: Grade result processing content:
The scoring support unit 630 obtains the scoring result by using the voice recognition result and the correct sentence list including the correct sentence (S630). The scoring support unit 630 matches the voice recognition result for each correct sentence included in the correct sentence list, classifies the internally obtained matching result according to the classification table, and uses the classification result as the evaluation result. Output.

For example, classification is performed based on the following classification table according to the range of correct answer accuracy.

Further, for example, the regular expression is classified based on the following classification table according to the matched regular expression.

<Other variants>
The present invention is not limited to the above embodiments and modifications. For example, the various processes described above may not only be executed in chronological order according to the description, but may also be executed in parallel or individually as required by the processing capacity of the apparatus that executes the processes. In addition, changes can be made as appropriate without departing from the spirit of the present invention.

<Programs and recording media>
Further, various processing functions in each device described in the above-described embodiment and modification may be realized by a computer. In that case, the processing content of the function that each device should have is described by a program. Then, by executing this program on the computer, various processing functions in each of the above devices are realized on the computer.

The program describing the processing content can be recorded on a computer-readable recording medium. The computer-readable recording medium may be, for example, a magnetic recording device, an optical disk, a photomagnetic recording medium, a semiconductor memory, or the like.

Further, the distribution of this program is performed, for example, by selling, transferring, renting, or the like a portable recording medium such as a DVD or a CD-ROM in which the program is recorded. Further, the program may be distributed by storing the program in the storage device of the server computer and transferring the program from the server computer to another computer via a network.

A computer that executes such a program first, for example, first stores a program recorded on a portable recording medium or a program transferred from a server computer in its own storage unit. Then, when the process is executed, the computer reads the program stored in its own storage unit and executes the process according to the read program. Further, as another embodiment of this program, a computer may read the program directly from a portable recording medium and execute processing according to the program. Further, each time the program is transferred from the server computer to this computer, the processing according to the received program may be executed sequentially. In addition, the above processing is executed by a so-called ASP (Application Service Provider) type service that realizes the processing function only by the execution instruction and result acquisition without transferring the program from the server computer to this computer. May be. In addition, the program shall include information used for processing by a computer and equivalent to the program (data that is not a direct command to the computer but has a property of defining the processing of the computer, etc.).

Further, although each device is configured by executing a predetermined program on a computer, at least a part of these processing contents may be realized by hardware.

Claims (17)

Includes a scoring support unit that requests evaluation of the voice answer using the voice recognition result of the voice answer to the question sentence and the correct answer information corresponding to the correct answer sentence list including at least one correct answer sentence corresponding to the question sentence.
Scoring support device. The scoring support device according to claim 1.
The voice recognition result is a voice recognition result of the voice response by a voice recognition model corresponding to the pronunciation of a non-native speaker.
Scoring support device. The scoring support device according to claim 1 or 2.
Includes a character processing unit that processes the voice recognition result and the correct answer information based on preset character processing rules.
The character processing rule is
(i) Convert uppercase letters to lowercase letters,
(ii) Delete the pre-specified symbol,
Including at least one of
Scoring support device. A scoring support device according to any one of claims 1 to 3.
The scoring support unit requests the evaluation of the voice response without considering the insertion error included in the voice recognition result.
Scoring support device. The scoring support device according to claim 4.
The insertion error is an insertion error at one or more of the beginning of the sentence, the end of the sentence, and the end of the phrase and the beginning of the phrase.
Scoring support device. A scoring support device according to any one of claims 1 to 5.
The correct answer information includes a keyword list composed of a list of at least one word for one of the correct answer sentences in the correct answer sentence list, and the keyword list includes words extracted from the one correct answer sentence and words. Contains at least one of the synonyms or homophones corresponding to the extracted words as a keyword.
The scoring support unit is a part of a keyword and the voice recognition result for each keyword list set in advance for the correct answer sentence included in the correct answer sentence list or created from the correct answer sentence included in the correct answer sentence list. Find an evaluation based on one or more of matches, exact matches, prefix matches, and suffix matches,
Scoring support device. The scoring support device according to claim 6.
The keyword list of the correct answer information is a list including information in the order of appearance of the keywords in one correct answer sentence.
The scoring support unit evaluates the matching rate of the voice recognition result and the keyword with respect to the appearance order of the keyword for each keyword list of the correct answer information.
Scoring support device. The scoring support device according to claim 6 or 7.
The keyword has at least one predetermined linguistic attribute.
Scoring support device. A scoring support device according to any one of claims 6 to 8.
Includes an answer sentence keyword creation unit that extracts keywords from the voice recognition result and creates an answer sentence keyword list.
The scoring support unit requests an evaluation based on the answer sentence keyword and the keyword list of the correct answer information.
Scoring support device. A scoring support device according to any one of claims 1 to 5.
The scoring support unit requests an evaluation in which the correct answer information and the voice recognition result are collated in character units, and the correct answer information is a correct answer sentence list or a keyword list.
Scoring support device. A scoring support device according to any one of claims 1 to 5.
The correct answer information includes one or more regular expressions corresponding to the one correct answer sentence, and the scoring support unit requests matching between the correct answer information and the voice recognition result as an evaluation.
Scoring support device. A scoring support device according to any one of claims 1 to 5.
The correct answer information is the correct answer sentence list.
Scoring support device. A scoring support device according to any one of claims 1 to 12.
The scoring support unit allows a difference of the number of characters or less specified in advance in the evaluation of the correct answer information and the voice recognition result.
Scoring support device. A scoring support device according to any one of claims 1 to 13.
The scoring support unit outputs the best evaluation among the obtained evaluations as an evaluation result.
Scoring support device. A scoring support device according to any one of claims 1 to 14.
The scoring support unit classifies the obtained evaluation according to a predetermined classification table, and outputs the classification result as the evaluation result.
Scoring support device. It includes a scoring support step for requesting the evaluation of the voice answer by using the voice recognition result of the voice answer to the question sentence and the correct answer information corresponding to the correct answer sentence list including at least one correct answer sentence corresponding to the question sentence.
Scoring support method. A program for operating a computer as a scoring support device according to any one of claims 1 to 15.

Images