TWI683290B - Spoken language teaching auxiliary method and device - Google Patents

Abstract

A spoken language teaching auxiliary device stores a piece of spoken language learning data including a training sentence and a weighted information, the training sentence is composed of a plurality of standard characters arranged in sequence, the weighted information has a weighted character corresponding to one of the standard characters and A weighted value. The device outputs the training sentence, and parses the speech when a speech is received to obtain a plurality of texts to be determined, and among the standard texts, those that do not match the text to be determined as at least one unmatched text, and use a The language recognition model generates an original score related to the at least one unmatched text, and when it is determined that the at least one unmatched text corresponds to the weighted text, the output unit is controlled to output according to the original score and the target weighted value A comment related to the voice.

Description

Spoken language teaching auxiliary method and device

The invention relates to a teaching auxiliary method and device, in particular to an oral teaching auxiliary method and device.

In recent years, the learned language teaching methods and methods mainly include audio-visual teaching, software teaching, and real person (on-site or remote) teaching.

The disadvantage of live-teaching is that it is expensive, and some users will not be able to speak for embarrassment when faced with live-teacher. The disadvantage of audiovisual teaching is that it cannot interact with users. The disadvantage of software teaching is that the user experience is not easy to realize the presence of spoken language. In particular, users can only follow the content of audiovisual teaching or software teaching, and it is difficult to determine whether the content they have repeated is correct.

Therefore, one of the objects of the present invention is to provide an oral teaching aid method, which can overcome at least one disadvantage of the prior art.

Therefore, the present invention provides an oral teaching assistant method, which is executed by an oral teaching assistant device. The oral teaching assistant device includes a storage unit, a voice input unit, an output unit, and an electrical connection to the storage unit, the voice input unit, And the processing unit of the output unit, the storage unit stores a plurality of spoken language learning data, each spoken language learning data includes a training sentence, a plurality of standard syllable groups and a piece of weighted information, the training sentence is composed of a plurality of standard characters arranged in sequence Constituted, the standard syllable groups correspond to the texts and are obtained by pre-analysing the texts, the weighted information has a weighted text and a weighted value corresponding to the weighted text, the weighted text corresponds to For one of the words included in the training sentence, the spoken language teaching aids include:

(A) The processing unit controls the output unit to output the training sentence;

(B) When receiving a voice from the voice input unit, the processing unit parses the voice to obtain a sentence to be determined composed of a plurality of characters to be determined and a plurality of syllables to be determined corresponding to the characters to be determined respectively group;

(C) When determining that at least one syllable group that does not match the standard syllable group and the pending syllable group, the processing unit regards the at least one standard text corresponding to the at least one syllable group that does not match as at least one Matching text

(D) The processing unit uses a language recognition model to generate an original score related to the at least one unmatched word based on the training sentence and the at least one unmatched word;

(E) when the processing unit determines that the at least one unmatched character corresponds to the weighted character, the weighted value corresponding to the weighted character is used as a target weighted value; and

(F) The processing unit controls the output unit to output an evaluation related to the speech based on the original score and the target weighted value.

In some embodiments, the output unit includes a display screen and a speaker. In step (E), the processing unit controls the display screen or the speaker to output the evaluation, and controls the display screen to display the at least one Match text.

In some embodiments, step (C) includes:

(c1) When the processing unit determines that the at least one syllable group that does not correspond to the standard syllable group and the waiting determination syllable group, determines that the at least one syllable group that does not correspond to the standard syllable group and the waiting determination syllable group Does not match; and

(c2) When the processing unit determines that the standard syllable groups respectively correspond to the waiting determination syllable group, it determines whether the standard syllable groups are respectively the same as the waiting determination syllable group, and determines that the standard syllable groups and When the at least one syllable group that is different from the waiting determination syllable group is determined, it is determined that the at least one syllable group that is different from the standard syllable groups does not match the waiting determination syllable group.

In some implementation forms, each learning material further includes a multimedia material, the multimedia material has a virtual reality image related to the training sentence, and before step (A), the processing unit according to the learning material contains the Multimedia data, controlling the display screen to display the virtual reality image of the multimedia data.

In some embodiments, the spoken language training device is a head-mounted display device, the output unit further includes a speaker, and the voice input unit includes a microphone.

In some implementation aspects, the language recognition model is an N-gram language recognition model.

Therefore, another object of the present invention is to provide a spoken language teaching aid that can overcome at least one of the shortcomings of the prior art.

Therefore, the present invention provides a spoken language teaching aid including a storage unit, a voice input unit, an output unit, and a processing unit electrically connected to the storage unit, the voice input unit, and the output unit. The storage unit stores a plurality of spoken language learning data. Each spoken language learning data includes a training sentence, a plurality of standard syllable groups, and a piece of weighted information. The training sentence is composed of a plurality of standard characters arranged in sequence. These standard syllable groups Corresponding to the standard texts respectively and obtained by parsing from the standard texts in advance, the weighted information has a weighted text and a weighted value corresponding to the weighted text, the weighted text corresponding to the training sentence One of these words.

The processing unit controls the output unit to output the training sentence, and when receiving a speech from the speech input unit, parses the speech to obtain a sentence to be composed of a plurality of characters to be decided and a plurality of words corresponding to the sentence The syllable group to be judged waiting for the judged text.

When determining that at least one syllable group that does not match the standard syllable group and the pending syllable group, the processing unit uses the at least one standard character corresponding to the at least one syllable group that does not match as at least one unmatched character.

The processing unit uses a language recognition model to generate an original score related to the at least one unmatched word based on the training sentence and the at least one unmatched word.

When determining that the at least one unmatched character corresponds to the weighted character, the processing unit uses the weighted value corresponding to the weighted character as a target weighted value.

The processing unit controls the output unit to output an evaluation related to the speech based on the original score and the target weighted value.

In some implementations, the output unit includes a display screen and a speaker, and the processing unit controls the display screen or the speaker to output the evaluation, and further controls the display screen to display the at least one inconsistent text.

In some implementation aspects, when the processing unit determines whether the standard syllable groups are consistent with the waiting determination syllable group, it is when determining that the standard syllable groups and the waiting determination syllable group do not correspond to at least one syllable group , It is determined that the at least one syllable group that does not correspond to the standard syllable groups does not match the waiting determination syllable group, and when it is determined that the standard syllable groups respectively correspond to the waiting determination syllable group, the standard syllables are also determined Whether the groups are respectively the same as the waiting determination syllable group, and when it is determined that the at least one syllable group of the standard syllable group is different from the waiting determination syllable group, it is determined that the at least one syllable group of the standard syllable group is different It does not match the waiting determination syllable group.

In some implementation forms, each learning material further includes a multimedia material, the multimedia material has a virtual reality image related to the training sentence, and the processing unit controls the output unit to output the training sentence before According to the multimedia data contained in the learning data, the display screen is controlled to display a virtual reality image of the multimedia data.

In some embodiments, the spoken language training device is a head-mounted display device, the output unit further includes a speaker, and the voice input unit includes a microphone.

In some implementation aspects, the language recognition model is an N-gram language recognition model.

The effect of the present invention is to use the language recognition model to obtain an original score based on the training sentence and the at least one inconsistent text, and output an evaluation related to the original score and weighted value, and the user can judge what he said based on the evaluation The degree of difference between the outgoing speech and the semantic meaning of the training sentence. And can avoid the situation that the user is embarrassed by the face of a real teacher and cannot speak.

Before the present invention is described in detail, it should be noted that in the following description, similar elements are denoted by the same numbers.

Referring to FIG. 1, an embodiment of the oral teaching aid device 100 of the present invention includes a storage unit 1, a voice input unit 2, an output unit 3 and an electrical connection to the storage unit 1, the voice input unit 2 and the output unit 3的处理单元4。 The processing unit 4.

In this embodiment, the storage unit 1 stores evaluation information 11 and multiple spoken language learning materials 12. The evaluation information 11 includes a plurality of evaluations respectively corresponding to different score intervals. For example, the evaluation information 11 includes three evaluations, which respectively correspond to "to be strengthened" evaluations with a score interval of 0 to less than 0.4, corresponding to "Not bad" evaluations with a score range of 0.4 to 0.8, and "Excellent" evaluations with a score range of more than 0.8 and not less than 0.1, but not limited to this.

Each spoken language learning material 12 includes a multimedia data 121 and at least one training sentence 122. Taking the spoken language learning material 12 only including a training sentence 122 as an example, the training sentence 122 corresponds to a plurality of standard syllable groups 123 and a weighted information 124. The multimedia data 121 is related to a virtual reality image and the situation of the virtual reality image is related to the training sentence 122. The training sentence 122 is a sentence composed of a plurality of characters arranged in sequence, and the standard syllable groups 123 corresponding to the training sentence 122 respectively correspond to the texts of the training sentence 122 and are pre-parsed by the And other texts. The weighted information 124 corresponding to the training sentence 122 has a weighted text 124A and a weighted value 124B corresponding to the weighted text 124A. The weighted text 124A corresponds to one of the standard texts included in the training sentence 122. For example, the virtual reality image presents a scene composed of three-dimensional realistic scenes and characters and related to an invited dialogue. The dialogue includes the training sentence 122 and other sentences. The training sentence 122 is, for example, "Let's go to the three o'clock show", these standard texts are "Let's", "go", "to", "the", "three", "o'clock", "show", corresponding to the standard The standard syllable group 123 of the text "o'clock" contains text type data of two syllables such as "o" and "clock", and the weighted text 124A is "o'clock", and the corresponding weighted value is 0.7.

It is worth noting that the number of spoken language learning data 12 stored in the storage unit 1 and the number of training sentences 122 and corresponding weighted information 124 contained in each spoken language learning data 12 are not limited to this embodiment, in other In an implementation form, the storage unit 1 can store a plurality of spoken language learning materials 12, and each spoken language learning data 12 can include a plurality of training sentences 122 and a corresponding plurality of weighted information 124.

In this embodiment, the voice input unit 2 includes a microphone 21, which allows the user to input voice by speaking. In this embodiment, the output unit 3 includes a display screen 31 and a speaker 32, and the virtual reality image can be displayed through the display screen 31, and the sound related to the virtual reality image can be output through the speaker 32, namely The above training sentence 122. In this embodiment, the spoken language teaching aid device 100 is, for example, implemented as a head-mounted display device (not shown in the figure), which is suitable for the user to wear on the head, and can be displayed through the display when worn by the user The screen 31 sees the three-dimensional realistic scenes and characters in the virtual image and hears the sounds related to the scenes and characters through the speaker 32, that is, the training sentence 122 mentioned above, and the content displayed on the display screen 31 can be followed to use The adaptability of the rotation of the person's head allows the user to input a voice into the spoken language teaching aid 100 by speaking into the microphone 21 when needed. However, the spoken language teaching aid device 100 of this embodiment is not limited to a head-mounted display device. It can also use a personal computer, a notebook computer, a tablet computer, or a smartphone with built-in or external speakers and microphones. achieve.

The following describes how the spoken language teaching aid device 100 of this embodiment performs an embodiment of the teaching aid method of the present invention.

First, in step S1, the processing unit 4 controls the display screen 31 to display the virtual reality image according to the multimedia data 121 included in the spoken language learning material 12. Continuing the previous example, the virtual reality image displayed on the display screen 31 is a dialogue situation about an invitation, so that a user wearing the spoken language teaching aid device 100 (ie, a head-mounted display device) can freely rotate his head The three-dimensional realistic scene and characters can be seen from the display screen 31 at an angle.

Then, in step S2, the processing unit 4 controls the output unit 3 to output training information related to the training sentence 122, and when receiving a voice from the voice input unit 2, in step S3, the processing unit 4 The speech is parsed to obtain a sentence to be determined composed of a plurality of characters to be determined and a plurality of groups of syllables to be determined corresponding to the characters to be determined. Continuing the previous example, the virtual reality image displayed on the display screen 31 is a dialogue situation about an invitation, and the speaker 32 emits the voice of the dialogue, which includes the training sentence 122 "Let’s go to the three o’ clock show". Then, the user speaks a voice into the microphone 21, for example, "Let's go to the show", the processing unit 4 parses the waiting judgment text to include "Let's", "go", "to", " "", "show", the sentence to be judged is "Let's go to the show" and multiple syllable groups to be judged corresponding to the character to be judged are parsed out.

Then, in step S4, the processing unit 4 determines whether the standard syllable groups 123 and the waiting determination syllable group match, and when it is determined that the standard syllable groups 123 and the waiting determination syllable group do not match at least one syllable group , The process proceeds to step S5, the processing unit 4 uses the at least one standard text corresponding to the at least one syllable group as at least one unmatched text, otherwise, the process proceeds to step S6, in step S6, because the input voice The corresponding sentence to be determined completely matches (same) the training sentence 122, so the processing unit 4 controls the display screen 31 to display a "very good" evaluation.

In this embodiment, in step S4, when the processing unit 4 determines whether the standard syllable groups 123 are consistent with the waiting determination syllable group, it determines that the standard syllable groups 123 and the waiting determination syllable group are not In a manner corresponding to at least one syllable group, it is determined that the uncorresponding at least one syllable group of the standard syllable groups does not match the pending syllable group. Continuing the previous example, because the waiting judgment syllable group corresponding to the waiting judgment text (including "Let's", "go", "to", "the", "show") is missing these standard characters (including "Let's", The order of ``go'', ``to'', ``the'', ``three'', ``o'clock'', ``show'') is the two of the fifth and sixth standard characters ``three'' and ``o'clock'' The standard syllable group 123, that is, the standard syllable group 123 corresponding to two standard characters such as "three" and "o'clock" does not correspond to the waiting determination syllable group, so "three" and "o'clock" and other two A standard character was judged as two non-matching characters. In addition, when the processing unit 4 determines that the standard syllable groups 123 respectively correspond to the waiting determination syllable group, it also determines whether the standard syllable groups 123 are respectively the same as the waiting determination syllable group, and determines the standards When the syllable group 123 is at least one syllable group that is different from the waiting determination syllable group, it is determined that the at least one syllable group that is different from the standard syllable group 123 does not match the waiting determination syllable group. For convenience of explanation, it is assumed here that the sentence spoken by the user is "Let's go to the nine o'clock show", and the processing unit 4 judges that the sentence to be judged "Let's go to the nine o'clock show" The waiting judgment texts respectively correspond to the standard texts in the training sentences (that is, the sentence to be judged is not missing text), and then the processing unit 4 judges that "nine" is not the same as the fifth in the training sentence 122 Standard text "three", then "nine" will be treated as an unmatched text.

Then, in step S7, the processing unit 4 uses a language recognition model to generate an original score related to the sentence to be determined according to the training sentence 122 and the at least one unmatched text. In this embodiment, the language recognition model is an N-gram language recognition model, which is not limited to this. In other implementations, the language recognition model can also use a recurrent neural network (Recurrent neural networks, RNN) or long short-term memory model (LSTM) model. The N-gram language recognition model can determine the importance of the at least one unmatched text according to the semantic meaning of the training sentence 122 and at least one unmatched text in the sentence to be determined, and determine a A score between 0 and 1 is used as the original score, because the N-gram language recognition model determines the importance of the at least one unmatched word based on the semantic meaning of the training sentence 122 and at least one unmatched word in the sentence to be determined It is an existing technology, so it will not be described here. Continuing the previous example, the processing unit 4 uses the N-gram language recognition model, according to the semantic meaning of the training sentence "Let's go to the three o'clock show" and the two unmatched words "three" and "o'clock" in the sentence to be determined ", judge the importance of "three" and "o'clock", and determine a score of 0.5 according to the importance, so the original score of the sentence to be judged is 0.5. Because "three" and "o'clock" are important words in the semantics of the training sentence 122, in other words, if the user does not speak correctly or the two words are missing in the spoken voice, it means that the user said And the training sentence 122 are very different, so the original score is only half of the full score (full score is 1).

Next, in step S8, the processing unit 4 determines whether the at least one unmatched text of the sentence to be determined corresponds to the weighted text. If so, the flow proceeds to step S9, otherwise, the flow proceeds to step S11. In step S9, the processing unit 4 uses the weighted value corresponding to the weighted text corresponding to the at least one unmatched text as a target weighted value. Continuing the previous example, since one of the two unmatched characters "o'clock" corresponds to a weighted character ("o'clock"), 0.7 is used as the target weighted value.

In step S10, the processing unit 4 calculates a weighted score based on the original score and the target weighted value, and controls the output unit 3 to output an evaluation related to the voice based on the weighted score and display the at least one disagreement Text. Continuing the previous example, the processing unit 4 takes the product of the original score and the target weighted value as the weighted score, so the weighted score is 0.35, and the corresponding evaluation of the weighted score (0.35) is "to be strengthened." , So the display screen 31 will display the evaluation of "to be strengthened" and words that the user such as "three" and "o'clock" said is wrong or unspoken, so that the user can know the voice he speaks The words "three" and "o'clock" are missing. In other embodiments, the evaluation may be output by the speaker 32 in the form of sound. It is worth noting that the weighting information 124 is preset by the manufacturer according to its own teaching purpose in advance, so that the deduction of "o’clock" by the target weighting value can be closer to its teaching purpose.

It is added that in other implementations, the spoken language learning material 12 includes a plurality of weighted information, and the weighted text of each weighted message corresponds to a standard text. Thus, when it is determined in step S8 that the text does not match When there are multiple and corresponding to multiple weighted characters, the weighted score in step S10 is the product of the original score and the target weighted values.

In step S11, since the at least one unmatched text of the sentence to be determined does not correspond to the weighted text 124A, the processing unit 4 only controls the output unit 3 to output an evaluation related to the speech based on the original score And display the at least one inconsistent text.

In summary, the above embodiment uses the language recognition model to generate at least one text in the to-be-decided sentence corresponding to the training sentence from the training sentence and the voice input by the user, and according to the at least one character The importance of obtains an original score, and the processing unit 4 multiplies the original score by the weighted value to increase the deduction of points, and outputs an evaluation related to the original score or the weighted score after the increased deduction to let The user can judge the degree of difference between the spoken speech and the semantics of the training sentence according to the evaluation and learn the missing or incorrectly pronounced words in the spoken speech. And it can realize the presence of the spoken language usage situation, and can also prevent the user from being unable to speak because of embarrassment for the real teacher.

However, the above are only examples of the present invention, and should not be used to limit the scope of the present invention. Any simple equivalent changes and modifications made according to the scope of the patent application of the present invention and the content of the patent specification are still classified as This invention covers the patent.

100················································································································································ Multimedia material · Training sentences 123 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ···· Microphone 3········ Output unit 31······· Display screen 32······· Speaker 4········ Processing unit S1~S11

Other features and functions of the present invention will be clearly presented in the embodiment with reference to the drawings, in which: FIG. 1 is a block diagram illustrating an embodiment of the oral teaching aid device of the present invention; and FIG. 2 is a flowchart Illustrates an embodiment of the oral teaching aid method of the present invention.

100················································································································································ Multimedia material · Training sentences 123 · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ···· Microphone 3········ Output unit 31······· Display screen 32······· Speaker 4········· Processing unit

Claims (12)

A spoken language teaching auxiliary method is executed by a spoken language teaching auxiliary device. The spoken language teaching auxiliary device includes a storage unit, a voice input unit, an output unit, and a process of electrically connecting the storage unit, the voice input unit and the output unit Unit, the storage unit stores a plurality of spoken language learning data, each of which includes a training sentence and a plurality of standard syllable groups corresponding to the training sentence and a weighted information, the training sentence is composed of a plurality of standard characters arranged in sequence Constituted, the standard syllable groups correspond to the texts respectively and are obtained by parsing the texts in advance, the weighted information has a weighted text and a weighted value corresponding to the weighted text, the weighted text corresponding to the text One of the characters included in the training sentence, the spoken language teaching aid includes: (A) the processing unit controls the output unit to output the training sentence; (B) the processing unit receives a word from the speech input unit When speaking, parse the speech to obtain a sentence to be judged consisting of a plurality of characters to be judged and a plurality of syllable groups to be judged corresponding to the characters waiting to be judged; (C) The processing unit judges the standard syllable groups When at least one syllable group that does not match the waiting determination syllable group, the at least one standard character corresponding to the at least one syllable group that is not matched is taken as at least one unmatched character; (D) the processing unit uses a language recognition model , According to the training sentence and the at least one unmatched text, generate an original score related to the at least one unmatched text; (E) When the processing unit determines that the at least one unmatched text corresponds to the weighted text, The weighted value corresponding to the weighted text is used as a target weighted value; and (F) the processing unit generates a weighted score based on the original score and the target weighted value, and controls the output unit to output a The evaluation of the voice. The assisting method for spoken language teaching according to claim 1, wherein the output unit includes a display screen and a speaker, and in the step (E), the processing unit controls the display screen or the speaker to output the evaluation and controls the The display screen displays the at least one unmatched text. The method for supporting spoken language teaching according to claim 1 or 2, wherein step (C) includes (c1) when the processing unit determines that at least one syllable group that does not correspond to the standard syllable group and the pending syllable group, Determining that the at least one syllable group that does not correspond to the standard syllable groups does not match the waiting determination syllable group; and (c2) when the processing unit determines that the standard syllable groups respectively correspond to the waiting determination syllable group, Whether the standard syllable groups are respectively the same as the waiting determination syllable group, and when it is determined that the standard syllable group is at least one syllable group different from the waiting determination syllable group, it is determined that the standard syllable group is different At least one syllable group does not match the waiting syllable group. The auxiliary method for oral teaching according to claim 3, wherein each learning material further includes a multimedia material, the multimedia material has a virtual reality image related to the training sentence, and before step (A), the processing unit The multimedia data contained in the learning data controls the display screen to display virtual reality images of the multimedia data. The oral teaching auxiliary method according to claim 4, wherein the oral teaching auxiliary device is a head-mounted display device, and the voice input unit includes a microphone. The auxiliary method for teaching spoken language according to claim 1, wherein the language recognition model is an N-gram language recognition model. A spoken language teaching auxiliary device includes a storage unit for storing a plurality of oral learning materials, each oral learning data includes a training sentence, a plurality of standard syllable groups and a weighted information, the training sentence is composed of multiple standards arranged in sequence Consisting of text, the standard syllable groups correspond to the standard text and are obtained by pre-analyzing the standard text, the weighted information has a weighted text and a weighted value corresponding to the weighted text, the weighted The text corresponds to one of the texts included in the training sentence; a voice input unit; an output unit; and a processing unit, electrically connected to the storage unit, the voice input unit, the output unit, and the voice input unit; Wherein, the processing unit controls the output unit to output training sentences; wherein, when receiving a speech from the speech input unit, the processing unit parses the speech to obtain a sentence to be judged consisting of a plurality of characters to be judged and more A syllable group to be determined corresponding to the waiting determination text; wherein, when the processing unit determines at least one syllable group that does not match the standard syllable group and the waiting determination syllable group, the at least one syllable group that does not match The at least one standard text corresponding to the group is regarded as at least one unmatched text; wherein, the processing unit uses a language recognition model to generate a text related to the at least one unmatched text according to the training sentence and the at least one unmatched text The original score; wherein, when the processing unit determines that the at least one unmatched text corresponds to the weighted text, the weighted value corresponding to the weighted text is used as a target weighted value; and, the processing unit is based on the original score and The target weighted value generates a weighted score, and controls the output unit to output an evaluation related to the speech according to the weighted score. The spoken language teaching support device according to claim 7, wherein the output unit includes a display screen and a speaker, and the processing unit controls the display screen or the speaker to output the evaluation, and controls the display screen to display the at least one Match text. The spoken language teaching aid according to claim 7 or 8, wherein the processing unit is determining whether the standard syllable groups are consistent with the standard syllable group when determining whether the standard syllable groups are consistent with the waiting determination syllable group When waiting to determine at least one syllable group that does not correspond to the syllable group, it is determined that the at least one syllable group that does not correspond to the standard syllable group does not match the waiting determination syllable group, and the processing unit determines that the standard syllable groups are respectively Corresponding to the waiting determination syllable group, it is also determined whether the standard syllable groups are respectively the same as the waiting determination syllable group, and when it is determined that at least one syllable group is different from the waiting determination syllable group, It is determined that the at least one syllable group that is different from the standard syllable groups does not match the pending syllable group. The spoken language teaching auxiliary device according to claim 9, wherein each learning material further includes a multimedia material having a virtual reality image related to the training sentences, and the processing unit is controlling the output unit to output the Before the one of the training sentences, according to the multimedia data contained in the learning data, the display screen is controlled to display the virtual reality image of the multimedia data. The spoken language teaching auxiliary device according to claim 10, wherein the spoken language training device is a head-mounted display device, the output unit further includes a speaker, and the voice input unit includes a microphone. The spoken language teaching auxiliary device according to claim 7, wherein the language recognition model is an N-gram language recognition model.

Images