TWI846240B - Speech training method, speech training system and user interface thereof - Google Patents

Abstract

A speech training method for intelligently generating several training words for pronunciation training, comprising the following steps. Several word factors are received through an input search device, and several search words are obtained by searching on the cloud platform according to the word factors. Performing recognition processing, analysis processing and translation processing on the search words, through a computing device, to generate training words. The training words is presented according to a bar form or a graphic form, through a feedback device.

Description

Voice training method, voice training system and user interface thereof

The present disclosure relates to a method and system for data processing, and more particularly to a method, system and user interface thereof for speech training.

"Arthria" is the incorrect position of articulation, inaccurate airflow direction and speed, and uncoordinated oral movements that cause speech errors and unclear speech. It may lead to stuttering and unfluent speech, or delayed language development. "Arthria" is an indicator for assessing the growth and development of preschool children.

In the existing treatment methods for articulation disorders, trainers (e.g., speech therapists) conduct pronunciation training for trainees (e.g., children) by presenting training materials (i.e., training vocabulary) on physical paper and using games. Before the training course, the trainer assesses the trainee to confirm which pronunciation the trainee has an abnormality. During the training course, the trainer provides the trainee with appropriate training vocabulary for pronunciation training, and one course of treatment is about 6 times.

During the interval between each course of treatment, the trainee can train at home. However, if the interval is too long, the trainee can only use the existing training vocabulary to repeatedly train, which is easy to produce fatigue and reduce the training effect. If the trainee (or the trainee's parents) can generate and update the training vocabulary by themselves, it can increase the trainee's interest and improve the training effect.

Therefore, there is an urgent need in the art for an improved speech training method and speech training system that can adaptively provide training vocabulary to facilitate trainees to perform pronunciation training at any time.

According to one aspect of the present disclosure, a speech training system is provided for intelligently generating a plurality of training words for performing a pronunciation training, including an input search device, a computing device, and a feedback device. The input search device is used to receive a plurality of vocabulary factors, and to search a plurality of search words on a cloud platform according to the vocabulary factors. The computing device is used to perform a recognition process, a parsing process, and a translation process on the search words to generate the training words. The feedback device is used to present each of the training words in a list form or a graphic form.

According to another aspect of the present disclosure, a speech training method is provided for intelligently generating a plurality of training words for performing a pronunciation training, comprising the following steps. A plurality of vocabulary factors are received by an input search device, and a plurality of search words are searched on a cloud platform according to the vocabulary factors. A computing device performs a recognition process, a parsing process, and a translation process on the search words to generate the training words. A feedback device presents each of the training words in a list form or a graphic form.

Other aspects and advantages of the present disclosure will become apparent by reading the following drawings, detailed description, and claims.

The technical terms in this specification refer to the customary terms in this technical field. If this specification explains or defines some terms, the interpretation of these terms shall be subject to the explanation or definition in this specification. Each embodiment of this disclosure has one or more technical features. Under the premise of possible implementation, a person with ordinary knowledge in this technical field can selectively implement part or all of the technical features in any embodiment, or selectively combine part or all of the technical features in these embodiments.

FIG. 1A is a functional block diagram of a speech training system 1000 according to an embodiment of the present disclosure. As shown in FIG. 1A , the speech training system 1000 includes: an input search device 100, a computing device 200, a feedback device 300, and a local database 400. The input search device 100 is connected to the computing device 200, and the computing device 200 is connected to the feedback device 300 and the local database 400. The speech training system 1000 is communicatively connected to a cloud platform 500 and a terminal device 600. Among them, the input search device 100 is communicatively connected to the cloud platform 500, and the feedback device 300 is communicatively connected to the terminal device 600.

The speech training system 1000 intelligently and automatically generates a plurality of training vocabularies tw, which form a training vocabulary cluster TW. The training vocabulary tw is provided to a trainee with articulation abnormality. The trainee, as the user U20, performs pronunciation training according to the training vocabulary tw to correct the symptoms of articulation abnormality. For example, if the trainee often has articulation abnormality in the pronunciation of the Chinese character "ㄆㄧㄥ", the training vocabulary cluster TW includes training vocabularies tw related to the pronunciation of "ㄆㄧㄥ", such as "apple", "bottle", etc.

On the other hand, the trainer operates the speech training system 1000 as the user U10 to generate a training vocabulary cluster TW. The trainer is, for example, the parent of the trainee or a speech therapist at the trainee's treatment institution. The trainer examines the trainee's pronunciation with abnormal articulation, and the trainer analyzes the vowels, consonants, prepositions, rhymes and combined rhymes of the pronunciation as vocabulary factors wf. For example, the trainer analyzes the initial consonant "ㄆ" and the vowels "ㄧ" and "ㄥ" of the pronunciation "ㄆㄧㄥ". These vowels, consonants (or rhymes, etc.) are used as vocabulary factors wf.

The user U10 (i.e., the trainer) inputs the vocabulary factor wf via the input search device 100. The input search device 100 has a text input function, and the user U10 inputs the vocabulary factor wf as "ㄆ", "ㄧ", and "ㄥ" in text form. Alternatively, the input search device 100 has a voice input function, and the user U10 inputs the vocabulary factor wf in a spoken form.

The input search device 100 searches on the cloud platform 500 according to the vocabulary factor wf. The cloud platform 500 is an open platform on the Internet, such as a web search engine, an online video platform, or an online blog. According to the vocabulary factor wf, a plurality of search terms sw are searched from the cloud platform 500, which are common or popular terms circulating on the cloud platform 500. For example, the search terms sw related to the vocabulary factors wf of "ㄆ", "ㄧ", and "ㄥ" are "apple", "bottle", "drama", "parallel space", and "screen", etc. These search terms sw constitute a search term cluster SW.

In another example, the user U10 can also directly input the source language through the input search device 100, for example, the user U10 directly inputs "apple". On the other hand, the input search device 100 searches for auxiliary images sg related to the vocabulary factor wf in the cloud platform 500. The auxiliary images sg are, for example, real photos or hand-drawn patterns of "apple" or "bottle".

The computing device 200 performs recognition processing, parsing processing and translation processing on multiple search terms sw in the search term cluster SW to generate multiple training terms tw. These training terms tw are selected from these search terms sw, and the training terms tw are suitable for the trainees to perform pronunciation training. These training terms tw constitute the training term cluster TW. More specifically, the computing device 200 analyzes the data obtained from the cloud platform 500 (i.e., the searched search terms sw), distinguishes and recognizes the product information and user information in these data, determines the semantics contained in these data, and integrates data in different languages. For example, the data searched from the cloud platform 500 include search terms sw such as "apple", "bottle", "Ping opera", "parallel space" and "screen". When the trainee is a young child (e.g., 3 to 10 years old), the search terms sw that the trainee can understand are "apple" and "bottle". Accordingly, the computing device 200 filters out "Ping opera", "parallel space" and "screen", and selects "apple" and "bottle" as training terms tw. The training terms of "apple" and "bottle" The training vocabulary cluster TW is composed of the vocabulary tw. On the other hand, the computing device 200 analyzes the searched auxiliary images sg to determine whether they are copyrighted by others and only selects auxiliary images sg without infringement suspicion. Then, the training vocabulary cluster TW and the auxiliary images sg are transmitted to the feedback device 300 and stored in the local database 400.

The local database 400 stores the currently generated training vocabulary clusters TW and auxiliary images sg, and the training vocabulary clusters TW that the trainee has used in the past. In addition, the local database 400 stores the historical records of the trainee's pronunciation training, including: the schedule of pronunciation training and the training results. This historical record can reflect the trainee's learning and growth experience. The local database 400 establishes a training vocabulary question bank TWL based on the currently generated and previously used training vocabulary clusters TW, auxiliary images, and the historical records of the trainee's pronunciation training.

The feedback device 300 processes the training vocabulary tw according to the training vocabulary question bank TWL and the historical record of the pronunciation training of the trainee, so that the training vocabulary tw is presented in a suitable presentation manner. For example, the feedback device 300 performs arrangement processing so that the training vocabulary tw has a list form. The arrangement order of the training vocabulary tw can be specified by the trainee, or the feedback device 300 determines it according to the historical record of the trainee. In addition, the feedback device 300 performs visualization processing so that the training vocabulary tw has a graphic form (for example, a cute font suitable for children).

The feedback device 300 is communicatively connected to the terminal device 600. The feedback device 300 transmits the training vocabulary tw in a list form or a graphic form to the terminal device 600. The terminal device 600 presents the training vocabulary tw to the user U20 (i.e., the trainee) in a list form or a graphic form, and the terminal device 600 also presents the searched auxiliary image sg to the user U20.

The terminal device 600 is the device currently used by the trainee. The terminal device 600 may be a portable mobile computing device, such as a smart phone, a tablet computer, or a laptop computer. The terminal device 600 may also be a fixed computing device, such as a desktop computer or a smart TV.

In this embodiment, the speech training system 1000 is independent of the terminal device 600. For example, the speech training system 1000 is disposed in another terminal device 700 other than the terminal device 600.

The input interface of the terminal device 700 (e.g., a touch screen or microphone of the terminal device 700) can be used as the input search device 100 in the voice training system 1000. The central processing unit (CPU) or graphics processing unit (GPU) of the terminal device 700, or the application program executed by the terminal device 700, can be used as the computing device 200 and the feedback device 300 in the voice training system 1000. The memory or storage medium of the terminal device 700 can be used as the local database 400 in the voice training system 1000.

When the voice training system 1000 is independent of the terminal device 600, the user U10 (i.e., the trainer) operates the voice training system 1000 via the terminal device 700 other than the terminal device 600 to generate training vocabulary tw. The terminal device 700 is communicatively connected to the terminal device 600, and the terminal device 700 transmits the training vocabulary tw to the terminal device 600 via wired communication or wireless communication. The user U20 (i.e., the trainee) reads the training vocabulary tw and the auxiliary image sg in the form of a list or a graphic via the terminal device 600 to perform pronunciation training.

Alternatively, the speech training system 1000 can be integrated into the terminal device 600. FIG. 1B is a functional block diagram of a speech training system 1000b of another embodiment of the present disclosure. As shown in FIG. 1B, the speech training system 1000b is disposed in the terminal device 600, and each component of the speech training system 1000b (i.e., the input search device 100, the computing device 200, the feedback device 300, and the local database 400) respectively corresponds to the central processing unit, the graphics processing unit, the memory or the storage medium of the terminal device 600, or is an application executed by the terminal device 600.

When the voice training system 1000b is integrated into the terminal device 600, the terminal device 600 is first delivered to the user U10 (i.e., the trainer), and the user U10 operates the voice training system 1000b via the terminal device 600 to generate training vocabulary tw. Then, the terminal device 600 is delivered to the user U20 (i.e., the trainee), and the user U20 reads the training vocabulary tw and the auxiliary image sg in the form of a list or a graphic via the terminal device 600.

In one example, the trainee can also play the role of the user U10, and the trainee can operate the voice training system 1000b of the terminal device 600 to generate training vocabulary tw. Accordingly, when the trainee is away from the medical institution and staying at home (for example, between two visits to the doctor), the trainee can operate the voice training system 1000b to generate training corpus for self-training.

Furthermore, the terminal device 600 of the trainee can be synchronized with the terminal device currently used by the trainer (not shown in the figure), and the feedback device 300 of the voice training system 1000b can provide a communication mechanism between the trainee and the trainer. The trainer can issue instructions to guide the trainee through the feedback device 300. Furthermore, the trainee can report the training results through the feedback device 300. Accordingly, during the interval between two consultations, the trainer does not need to wait for the trainee to generate training vocabulary tw, and the trainee can also grasp the training status of the trainer at any time to save time for both parties.

FIG. 1C is a functional block diagram of the computing device 200 of the speech training system 1000. As shown in FIG. 1C, the computing device 200 includes a recognition device 210, a parsing device 220, and a translation device 230.

The recognition device 210, the parsing device 220 and the translation device 230 are all connected to the input search device 100 to receive the search vocabulary cluster SW and the auxiliary image sg from the input search device 100, and process the search vocabulary cluster SW to generate the training vocabulary cluster TW. In addition, the recognition device 210, the parsing device 220 and the translation device 230 are all connected to the feedback device 300 to transmit the training vocabulary cluster TW and the auxiliary image sg to the feedback device 300.

In operation, the recognition device 210 can execute a detection program, which performs recognition processing on multiple search terms sw in the search term cluster SW to distinguish the product information and user information therein. In addition, the recognition device 210 can determine the relevant data of the homophones and pronounciations of the search terms sw. In one example, the recognition device 210 can detect whether these search terms sw are training corpora that meet the expectations of the user U10 (i.e., the trainer). The recognition device 210 filters out the search terms sw that do not meet the expectations of the user U10. For example, the search words sw include "Penghu" and "piqiu", which are pronounced differently from "ㄆ", "ㄧ" and "ㄥ", indicating that "Penghu" and "piqiu" do not meet the matching results and are therefore filtered out. For another example, the search words sw include "buttocks", which have an indecent meaning and are therefore filtered out. The input search device 100 further searches the cloud platform 500 for words more related to "ㄆ", "ㄧ" and "ㄥ".

The parsing device 220 can execute a parsing program to analyze sentences, strings or popular words related to the vocabulary factor wf. For example, a popular word commonly used on the cloud platform 500 is "ping-pong competition", which is related to the vocabulary factors wf of "ㄆ", "ㄧ" and "ㄥ". In one example, the parsing device 220 can be further connected to the local database 400 to obtain the historical records of the trainee's training from the local database 400 to analyze the trainee's personal background, including: what type of vocabulary the trainee is more interested in, etc. Accordingly, the analysis device 220 can cluster the multiple search terms sw through feature screening to select training terms tw suitable for the trainee from these search terms sw.

The translation device 230 can execute a translation program to translate similar words with different vowels, and can expand the training vocabulary tw to provide the trainee with more diversified training. For example, the vowel "ㄥ" in the vocabulary factors wf of "ㄆ", "ㄧ" and "ㄥ" can be translated into the vowel "ㄣ" to obtain similar words pronounced as "ㄆ", "ㄧ" and "ㄣ", such as "jigsaw" and "character". Accordingly, the trainee can obtain training vocabulary tw such as "apple", "bottle", "jigsaw" and "character" from the feedback device 300, making the pronunciation training more diversified.

FIG. 2A is a schematic diagram of the operation interface 31a of the input search device 100 of the voice training system 1000 or 1000b, and the voice training system 1000b is used as an example for explanation. The voice training system 1000b is set in the terminal device 600 (corresponding to the embodiment of FIG. 1B), and the terminal device 600 is, for example, a smart phone. The trainer acts as the user U10, and the trainer operates the input search device 100 through the operation interface 31a to establish the question type to be implemented on the trainee.

The operation interface 31a is, for example, a graphical user interface (GUI). The operation interface 31a includes a plurality of operation areas 311 to 313 and 321 to 326. The operation area 311 is a main menu, which includes a plurality of options, such as: "Home", "Medical Records" and "Home Training". When the option "Home" is selected, the operation areas 312 and 313 display the outline of the question types. For example, the operation area 312 displays "Standard Training" and the operation area 313 displays "Children's Training", which is suitable for younger trainees.

The operation areas 321 to 326 selectively display more detailed customized question types. For example, the customized question type displayed in the operation area 322 is "Children's Home Training", and the customized question type displayed in the operation area 325 is "Comprehensive Training Revised Edition".

FIG. 2B is a schematic diagram of another operation interface 31b of the input search device 100 of the speech training system 1000 or 1000b, and the speech training system 1000b is used as an example for explanation. After the trainer creates the outline of the question type and/or customizes the question type through the operation interface 31a of FIG. 2A, the trainer selects the training vocabulary tw through the operation interface 31b of FIG. 2B.

The trainer first inputs the vocabulary factor wf through the operation areas 341 and 342 of the operation interface 31b. For example, the initial consonant "ㄆ" is input through the operation area 341, and the final vowel "ㄣ" is input through the operation area 342. On the other hand, the operation area 343 displays the search words sw related to the initial consonant "ㄆ" and the final vowel "ㄣ" obtained from the cloud platform 500, including: "fountain", "pinyin", "wash basin" and "ping pong", etc.

Then, the computing device 200 performs recognition processing, analysis processing and translation processing on the search words sw to select training words tw suitable for the trainee, including: "fountain" and "ping pong". In addition, the operation area 345 displays the selected training words tw of "fountain" and "ping pong".

FIG. 3A is a schematic diagram of the operation interface 41a of the feedback device 300 of the speech training system 1000 or 1000b, and the speech training system 1000b is used as an example for explanation. The trainee plays the role of the user U20, and the trainee operates the feedback device 300 through the operation interface 41a to perform pronunciation training. As shown in FIG. 3A, the operation area 411 of the operation interface 41a displays the training vocabulary tw in the form of a list in sequence. For example, the training vocabulary tw currently displayed is "apple". In addition, the feedback device 300 can enhance the display of the training vocabulary tw in a graphical form, for example, displaying "apple" in the operation area 411 using a cute font 414 suitable for children.

On the other hand, the operation area 412 is a recording button. The trainee touches or slides the recording button to activate the recording function of the terminal device 600, and the trainee's pronunciation of "apple" is recorded in the local database 400 of the voice training system 1000b. The operation area 413 can display the recording time accordingly.

Figure 3B is a schematic diagram of another operation interface 41b of the feedback device 300 of the voice training system 1000 or 1000b. As shown in Figure 3B, the feedback device 300 displays an auxiliary image sg related to "apple" through the operation interface 41b. The operation area 421 of the operation interface 41b displays the text form of "apple", and the operation area 423 displays the auxiliary image sg of "apple". When the trainee is a young child, the auxiliary image sg helps to enhance the interest of the trainee, and the operation area 424 displays a guiding narrative to prompt the trainee to pay attention to the auxiliary image sg for training. Similar to the operation interface 41a of FIG. 3A , the operation area 422 of the operation interface 41b is a recording key.

FIG. 4 is a schematic diagram of another operation interface 51 of the feedback device 300 of the speech training system 1000 or 1000b. The operation interface 51 is operated by the trainer as the user U10. The trainer controls the feedback device 300 through the operation interface 51 to view the training results of the trainee. As shown in FIG. 4, the operation area 511 of the operation interface 51 shows that the current training vocabulary tw to be viewed is "apple", and the feedback device 300 can obtain the recorded content of the trainee's pronunciation of "apple" from the local database 400 and play it (i.e., recording playback).

According to the recording playback, the trainer can check the trainee's training results for "蘋果" to determine whether the trainee's pronunciation has articulation abnormalities. The trainer can record the training results in the operation areas 512~515. For example, if the trainee's pronunciation of the initial consonant "ㄆ" and the vowels "ㄧ" and "ㄥ" of the first character "蘋果" is correct, the trainer selects "T" (i.e., True) in the operation areas 512 and 513, indicating a positive training result.

Similarly, if the trainee pronounces the vowels "ㄨ" and "ㄛ" of the second character "果" correctly, the trainee selects "T" in the operation area 515 to indicate a positive training result. On the other hand, if the trainee pronounces the initial consonant "ㄍ" of "果" incorrectly, the trainee selects "F" (i.e., False) in the operation area 514 to indicate a negative training result. In one example, the negative training result "F" is displayed in a high-contrast color.

FIG. 5 is a flow chart of a speech training method according to an embodiment of the present disclosure. The speech training method according to the present embodiment can be implemented in conjunction with the speech training system 1000 or 1000b of the embodiments of FIG. 1A to FIG. 4. Referring to FIG. 5, first, in step S102, the input search device 100 receives a vocabulary factor wf, and searches the cloud platform 500 for a search vocabulary sw according to the vocabulary factor wf.

Then, in step S104, the search vocabulary sw is processed by the computing device 200 for recognition, analysis and translation to generate the training vocabulary tw. Step S104 includes the following sub-steps. In step S104-1, the homophones and broken pronunciations of the search vocabulary sw are determined by the recognition device 210. In step S104-2, the character strings or popular words related to the vocabulary factor wf are analyzed by the parsing device 220. In step S104-3, the similar words related to the vowel are translated by the translation device 230.

Then, in step S106, the training vocabulary tw and the history record of pronunciation training are stored in the local database 400. Then, in step S108, a training vocabulary question bank TWL is generated according to the training vocabulary tw and the history record. Then, in step S110, the training vocabulary tw is processed by the feedback device 300 according to the training vocabulary question bank TWL and the history record, so that the training vocabulary tw has a list form or a graphic form.

Then, in step S112, the training vocabulary tw is presented in a list form or a graphic form by the feedback device 300. Then, in step S114, the training vocabulary tw in a list form or a graphic form is transmitted to the terminal device 600 by the feedback device 300.

Although the present invention has been disclosed in detail with preferred embodiments and examples, it is understood that these examples are intended to be illustrative rather than restrictive. It is expected that a person with ordinary knowledge in the art can think of various modifications and combinations, which fall within the spirit of the present invention and the scope of the attached patent application.

1000,1000b: Voice training system 100: Input search device 200: Computing device 300: Feedback device 400: Local database 500: Cloud platform 600,700: Terminal device 210: Recognition device 220: Analysis device 230: Translation device 31a,31b,41a,41b,51: Operation interface 311~313,321~326,341~345: Operation area 411~414,421~424,511~515: Operation area U10,U20: User wf: Vocabulary factor SW: Search vocabulary cluster sw: Search vocabulary sg: auxiliary image TW: training vocabulary cluster tw: training vocabulary TWL: training vocabulary question bank S102~S114: steps

FIG. 1A is a functional block diagram of a speech training system of one embodiment of the present disclosure. FIG. 1B is a functional block diagram of a speech training system of another embodiment of the present disclosure. FIG. 1C is a functional block diagram of a computing device of the speech training system. FIG. 2A is a schematic diagram of an operation interface of an input search device of the speech training system. FIG. 2B is a schematic diagram of another operation interface of an input search device of the speech training system. FIG. 3A is a schematic diagram of an operation interface of a feedback device of the speech training system. FIG. 3B is a schematic diagram of another operation interface of a feedback device of the speech training system. FIG. 4 is a schematic diagram of another operation interface of a feedback device of the speech training system. Figure 5 is a flow chart of the speech training method of an embodiment of the present disclosure.

1000: Voice training system

100: Enter the search device

200: Computing device

300: Feedback device

400: Local database

500: Cloud Platform

600,700: terminal device

U10,U20:User

wf: Lexical factor

SW:Search term clusters

sw:Search vocabulary

sg: auxiliary images

TW: Training vocabulary clusters

tw: Vocabulary training

TWL: Vocabulary Training Question Bank

Claims (20)

A speech training system is used to intelligently generate a plurality of training words for a pronunciation training, comprising: an input search device for receiving a plurality of vocabulary factors and searching a plurality of search words on a cloud platform according to the vocabulary factors; an operation device for performing a recognition process, a parsing process and a translation process on the search words to generate the training words; and a feedback device for presenting each of the training words in a list form or a graphic form. The speech training system as described in claim 1 further includes: A local database for storing the training vocabularies and a history record of the pronunciation training, and generating a training vocabulary question bank based on the training vocabularies and the history record. A speech training system as described in claim 2, wherein the feedback device processes the training vocabulary according to the training vocabulary question bank and the historical record so that each of the training vocabulary has the list form or the graphic form. The speech training system as described in claim 1, wherein the vocabulary factors include at least one initial consonant and one final consonant with abnormal articulation. The speech training system as described in claim 4, wherein the computing device comprises: a recognition device for determining homophones and pronounciations of each of the search words; a parsing device for analyzing strings or popular words related to the vocabulary factors; and a translation device for translating similar words related to the vowels. The speech training system as described in claim 1, wherein the cloud platform is an Internet search engine, an Internet video platform or an Internet blog. A voice training system as described in claim 6, wherein the search terms are common terms or popular terms circulated on the cloud platform. A speech training system as described in claim 1, wherein the feedback device is communicatively connected to a terminal device, and the feedback device transmits each of the training words in the list form or the graphic form to the terminal device. A speech training method is used to intelligently generate a plurality of training words for performing a pronunciation training, comprising: Receiving a plurality of vocabulary factors by an input search device, and searching a plurality of search words on a cloud platform according to the vocabulary factors; Performing a recognition process, a parsing process and a translation process on the search words by a computing device to generate the training words; and Presenting each of the training words in a list form or a graphic form by a feedback device. The speech training method as described in claim 9 further includes: Storing the training vocabularies and a history record of the pronunciation training in a local database; and Generating a training vocabulary question bank based on the training vocabularies and the history record. The speech training method as described in claim 10 further includes: The feedback device processes the training vocabulary according to the training vocabulary question bank and the historical record, so that each training vocabulary has the list form or the graphic form. The speech training method as described in claim 9, wherein the vocabulary factors include at least one initial consonant and one final consonant with abnormal articulation. The speech training method as described in claim 12, wherein the steps of performing the recognition processing, the parsing processing and the translation processing on the search words by the computing device include: Determining homophones and pronounciations of each of the search words by a recognition device; Analyzing strings or popular words related to the vocabulary factors by a parsing device; and Translating similar words related to the vowels by a translation device. The voice training method as described in claim 9, wherein the cloud platform is an Internet search engine, an Internet video platform or an Internet blog. A voice training method as described in claim 14, wherein the search terms are common terms or popular terms circulated on the cloud platform. The voice training method as described in claim 9, wherein the feedback device is communicatively connected to a terminal device, and the voice training method further comprises: The feedback device transmits each of the training words in the list form or the graphic form to the terminal device. A user interface of a speech training system is used to operate the speech training system to intelligently generate a plurality of training words for a pronunciation training. The speech training system includes an input search device and a feedback device. The user interface includes: a first operation interface, associated with the input search device, and used to input a plurality of vocabulary factors; and a second operation interface, associated with the feedback device, and used to display each of the training words in a list form or a graphic form, The speech training system searches for multiple search terms on a cloud platform based on the vocabulary factors, and performs a recognition process, a parsing process, and a translation process on the search terms to generate the training terms. A user interface as described in claim 17, wherein the first operating interface displays the search terms correspondingly. A user interface as described in claim 17, wherein the speech training system searches the cloud platform for a plurality of auxiliary images based on the vocabulary factors, and the second operating interface displays the auxiliary images accordingly. A user interface as described in claim 17, wherein the feedback device is communicatively connected to a terminal device, and each training vocabulary displayed on the second operating interface is transmitted to the terminal device.

Images