TWI876580B - System and method for obtaining accurate surname - Google Patents

Abstract

A system and a method for obtaining an accurate surname are provided. The method includes following steps: when an accurate-surname-confirmation-module judges that a text does not include a precise-surname-statement and does not include a confirmation-surname-sentence, extracting, by the accurate-surname-confirmation-module, a surname from the text; using, by a similar-sound-acquisition-module, to obtain a similar-note-text-sequence; using, by a TOP-N-surname-generation-module, the similar-note-text-sequence to obtain a TOP-N-surname; and generating, by a precise-surname-answer-sentence-generation-module, a precise-surname-querying-back-sentence corresponding to the TOP-N-surname.

Description

System and method for obtaining accurate surname

The present invention relates to a system and method for obtaining an accurate surname.

When a customer connects/calls a general existing conversation service system, if the conversation service system needs to confirm the customer's surname at this time, the existing text recognition technology often misjudges because there is a high probability of homophones in Chinese, resulting in a bad impression for the customer.

The system for obtaining an accurate surname of the present invention includes a storage medium and a processor. The storage medium stores a plurality of modules, wherein the plurality of modules include an accurate surname confirmation module, a similar sound acquisition module, a TOP N surname generation module, and an accurate surname answer sentence generation module. The processor is coupled to the storage medium, wherein the processor accesses and executes the plurality of modules to execute the following steps: when the accurate surname confirmation module determines that the text does not include an accurate surname statement and does not include a surname confirmation sentence, the accurate surname confirmation module extracts the surname from the text; the similar sound acquisition module uses the surname to obtain a similar musical note text sequence; the TOP N surname generation module uses the similar musical note text sequence to obtain the TOP N surnames; and the accurate surname answer sentence generation module generates an accurate surname answer sentence corresponding to the TOP N surnames.

The method for obtaining an accurate surname of the present invention comprises the following steps: when the accurate surname confirmation module determines that the text does not include an accurate surname statement and does not include a surname confirmation sentence, the accurate surname confirmation module extracts the surname from the text; the similar sound acquisition module uses the surname to obtain a similar musical note text sequence; the TOP N surname generation module uses the similar musical note text sequence to obtain the TOP N surnames; and the accurate surname question answer sentence generation module generates an accurate surname question answer sentence corresponding to the TOP N surnames .

Fig. 1 is a schematic diagram of a system 1 for obtaining an accurate surname according to an embodiment of the present invention. The system 1 may include a storage medium and a processor.

The storage medium is, for example, any type of fixed or removable random access memory (RAM), read-only memory (ROM), flash memory, hard disk drive (HDD), solid state drive (SSD) or similar components or a combination of the above components, and is used to store multiple modules or various applications that can be executed by the processor. In this embodiment, the storage medium can store multiple modules, wherein the multiple modules may include an accurate surname confirmation module 20, a similar sound acquisition module 30, a TOP N surname generation module 40, and an accurate surname question sentence generation module 50. In other embodiments, the multiple modules may also include a voice recognition and text conversion module 10. In other embodiments, the storage medium can also store the Chinese character database 11, the surname database 21, and the Hundred Family Surnames database 41.

The processor is, for example, a central processing unit (CPU), or other programmable general-purpose or special-purpose micro control unit (MCU), microprocessor, digital signal processor (DSP), programmable controller, application specific integrated circuit (ASIC), graphics processing unit (GPU), image signal processor (ISP), image processing unit (IPU), arithmetic logic unit (ALU), complex programmable logic device (CPLD), field programmable gate array (FPGA), or other similar components or combinations of the above components. The processor may be coupled to the storage medium and access and execute a plurality of modules and various applications stored in the storage medium.

FIG2 is an operation flow chart of the system 1 shown in FIG1. Please refer to FIG1 and FIG2 at the same time.

In step S100, the speech recognition and text conversion module 10 can convert speech into text. This will be further described below with reference to FIG.

FIG3 is a detailed flow chart of step S100 shown in FIG2. Please refer to FIG1, FIG2 and FIG3 at the same time.

In step S101, the speech recognition and text conversion module 10 can receive speech from a client device. In detail, the system 1 shown in FIG1 may further include a transceiver (not shown in FIG1 ). The speech recognition and text conversion module 10 can receive speech from a client device (not shown in FIG1 ) via the transceiver. The client device is, for example, a telephone or a smart speaker device, but the present invention is not limited thereto. For example, when a customer operates a client device to connect to the system 1, the system 1 may ask, "What is your last name?" The following embodiment will be described assuming that the speech received by the speech recognition and text conversion module 10 from the client device is "My last name is Wang".

In step S102, the speech recognition and text conversion module 10 can perform speech recognition operation on the speech to obtain recognized speech. In detail, the speech recognition and text conversion module 10 can perform pre-processing on the speech. Since different customers are in different environments, the distances between the customers and the sound receiving device/microphone of the customer equipment are also different, and various noises will affect the sound receiving results. Therefore, the speech recognition and text conversion module 10 can remove the noise of the environment and recognize the usable recognized speech (human voice frequency).

In step S103, the speech recognition and text conversion module 10 can use the recognized speech and the Chinese character library 11 to obtain text corresponding to the speech. Specifically, the speech recognition and text conversion module 10 can use the Chinese character content in the Chinese character library 11 to generate text corresponding to the recognized speech. Based on the above embodiment, the speech recognition and text conversion module 10 can obtain the text "My surname is Wang".

Please return to FIG. 2. In step S200, when the precise surname confirmation module 20 determines that the text does not include the precise surname statement and does not include the surname confirmation sentence, the precise surname confirmation module 20 can extract the surname from the text. This will be further explained below with reference to FIG. 4.

FIG4 is a detailed flow chart of step S200 shown in FIG2. Please refer to FIG1, FIG2 and FIG4 at the same time.

In step S201, the accurate surname confirmation module 20 may use the surname saying database 21 to determine whether the text includes an accurate surname saying.

When the precise surname confirmation module 20 determines that the text includes the precise surname statement (the determination result of step S201 is "yes"), in step S202, the precise surname confirmation module 20 can adopt the precise surname statement. For example, if the text includes the precise surname statement "My surname is Zhang, and my bow is Zhang" or the precise surname statement "My surname is Huang, and my grass is Huang", the precise surname confirmation module 20 can adopt the precise surname statement. Then, the precise surname confirmation module 20 can execute step S203 to end the dialogue with the client device. In other words, the system 1 will not execute steps S300, S400 and S500 at this time.

When the precise surname confirmation module 20 determines that the text does not include the precise surname statement, the precise surname confirmation module 20 can determine whether the text includes a surname confirmation sentence. In detail, when the precise surname confirmation module 20 determines that the text includes a surname confirmation sentence (the judgment result of step S204 is "yes"), in step S203, the precise surname confirmation module 20 can end the dialogue with the customer. For example, assume that after multiple rounds of dialogue with a specific customer, the text has included a surname confirmation sentence (such as "Yes, that's right"). Since the text does not include the precise surname statement but has included the surname confirmation sentence, the precise surname confirmation module 20 can execute step S203 to end the dialogue with the customer device. In other words, at this time, the system 1 will not execute steps S300, S400 and S500.

On the other hand, when the precise surname confirmation module 20 determines that the text does not include a surname confirmation sentence (the determination result of step S204 is "No"), in step S205, the precise surname confirmation module 20 can extract a surname from the text. Based on the above embodiment, since the text "My surname is Wang" does not include a surname confirmation sentence, the precise surname confirmation module 20 can execute step S205. In other words, the precise surname confirmation module 20 can extract a possible surname "Wang" from the text "My surname is Wang".

Please return to FIG. 2. In step S300, the similar sound acquisition module 30 can use the surname to obtain a similar phonetic character sequence. FIG. 5 will be used to further illustrate this.

FIG5 is a detailed flow chart of step S300 shown in FIG2. Please refer to FIG1, FIG2 and FIG5 at the same time.

In step S301, the similar sound acquisition module 30 can remove the Chinese tones in the surname to obtain the retained phonetic notation. The Chinese tones can be the 1234 tones of Chinese. Based on the above embodiment, the similar sound acquisition module 30 can remove the Chinese tones in the surname "王" to obtain the retained phonetic notation "ㄨㄤ".

In step S302, the similar sound acquisition module 30 can use the phonetic consonant vowel table and the reserved phonetic notation (as shown in Table 1 below) to obtain a similar musical note text sequence. Based on the above embodiment, the similar sound acquisition module 30 can use the phonetic consonant vowel table and the reserved phonetic notation "ㄨㄤ" to obtain a similar musical note text sequence "ㄨㄤ、ㄅㄤ、ㄈㄤ、ㄓㄤ、ㄔㄤ、ㄏㄨㄤ、ㄎㄨㄤ...".

Table 1 Example of a phonetic consonant and vowel table consonant ㄅㄆㄇㄈㄉㄊㄋㄌㄍㄎㄏㄐㄑㄒ Consonants can exist independently ㄓㄔㄕㄖㄗㄘㄙ vowel ㄧㄨㄩㄚㄛㄜㄝㄞㄟㄠㄡㄢㄣㄤㄥㄦ

Please return to FIG. 2. In step S400, the TOP N surname generation module 40 can obtain the TOP N surnames using similar phonetic character sequences. This will be further described below with reference to FIG. 6.

FIG6 is a detailed flow chart of step S400 shown in FIG2. Please refer to FIG1, FIG2 and FIG6 at the same time.

In step S401, the TOP N surname generation module 40 can generate a possible surname sequence using similar phonetic character sequences and the Hundred Family Surnames database 41. Based on the above-mentioned embodiment, the TOP N surname generation module 40 can compare the similar phonetic character sequences "ㄨㄤ、ㄅㄤ、ㄈㄤ、ㄓㄤ、ㄔㄤ、ㄏㄨㄤ、ㄎㄨㄤ..." and the Hundred Family Surnames database 41 to generate a possible surname sequence "汪、王、邦、方、房、张、常、黄、匡".

In step S402, the TOP N surname generation module 40 can select TOP N surnames from the possible surname sequence. In the above embodiment, the TOP N surname generation module 40 can select the TOP N surnames "王" and "黄" from the possible surname sequence "汪、王、邦、方、房、张、常、黄、匡".

Please return to FIG. 2. In step S500, the precise surname answer sentence generation module 50 can generate precise surname answer sentence corresponding to the TOP N surnames. FIG. 7 will be used to further illustrate the above.

FIG7 is a detailed flow chart of step S500 shown in FIG2. Please refer to FIG1, FIG2 and FIG7 at the same time.

In step S501, the precise surname answer sentence generation module 50 can use the surname saying database 21 to select the precise surname sayings corresponding to the TOP N surnames. Based on the above embodiment, the precise surname answer sentence generation module 50 can use the surname saying database 21 to select the precise surname sayings "三橫一豎王" and "草头黄" corresponding to the TOP N surnames ("王" and "黄") .

In step S502, the precise surname answer sentence generation module 50 can use the precise surname utterance to generate a precise surname answer sentence corresponding to the TOP N surnames. Based on the above embodiment, the precise surname answer sentence generation module 50 can use the precise surname utterance "三橫一竖王" and "草头黄" to generate a precise surname answer sentence "Excuse me, is it 三橫一竖王, or 草头黄?".

After generating the precise surname query sentence, the system 1 can transmit the precise surname query sentence to the client device via a transceiver (not shown in FIG. 1 ). After the client responds, the next round of dialogue can be entered (i.e., returning to step S100 again). If the user responds with the voice "It is three horizontal and one vertical king", then in step S201, the precise surname confirmation module 20 can determine that the text "It is three horizontal and one vertical king" includes the precise surname statement "three horizontal and one vertical king", and can execute steps S202 and S203 (ending multiple rounds of dialogue with the client).

On the other hand, if the user responds with the voice "Neither, it's Lang, Lang in Lang" at this time, then in step S201, the precise surname confirmation module 20 can determine that the text "Neither, it's Lang, Lang in Lang" includes other precise surname statements, and can execute steps S202 and S203 (ending multiple rounds of dialogue with the customer).

On the other hand, if the user responds with the voice "neither, my surname is Hang", then in step S201, the precise surname confirmation module 20 can determine that the text "neither, my surname is Hang" does not include the precise surname statement. Then, the precise surname confirmation module 20 can determine in step S204 that the text "neither, my surname is Hang" does not include the surname confirmation sentence, and can execute step S205. Then, the system 1 will continue to execute steps S300, S400 and S500, and after executing step S500, start the next round of dialogue (i.e. return to step S100).

On the other hand, if the user responds with the voice message "Not Su La, my surname is Huang La (the user with the surname Fang pronounces 'Fang' as 'Huang' due to his accent)" at this time, and if after multiple rounds of dialogue, System 1 still cannot obtain the accurate surname, when the number of multiple rounds of dialogue is equal to the threshold value, System 1 will end the multiple rounds of dialogue with the customer.

The present invention further provides a method for obtaining an accurate surname, wherein the method can be implemented by the system 1 shown in FIG. 1 . The method comprises the following steps:

(a) When the precise surname confirmation module determines that the text does not include a precise surname statement and does not include a surname confirmation sentence, the precise surname confirmation module extracts the surname from the text.

(b) The similar sound acquisition module uses the surname to obtain a similar phonetic character sequence.

(c) The TOP N surname generation module obtains the TOP N surnames using similar phonetic character sequences.

(d) The precise surname answer sentence generation module generates precise surname answer question sentences corresponding to the TOP N surnames.

The method has been described in the above embodiments and will not be described again here.

In summary, the system and method for obtaining an accurate surname of the present invention can obtain the accurate surname of a customer through accurate surname sayings, similar phonetic character sequences, and accurate surname answer sentences. In addition, the accuracy of the surname obtained can be improved through multiple rounds of dialogue with the customer. Based on this, for a dialogue service system that requires a customer's surname, the accurate surname of the customer can be obtained more effectively.

1: System for obtaining accurate surnames 10: Voice recognition and text conversion module 20: Accurate surname confirmation module 30: Similar sound acquisition module 40: TOP N surname generation module 50: Accurate surname question sentence generation module 11: Chinese character library 21: Surname saying database 41: Hundred Family Surname Database S100~S500, S101~S103, S201~S205, S301~S302, S401~S402, S501~S502: Steps

FIG. 1 is a schematic diagram of a system for obtaining an accurate surname according to an embodiment of the present invention. FIG. 2 is an operation flow chart of the system shown in FIG. 1 . FIG. 3 is a detailed flow chart of step S100 shown in FIG. 2 . FIG. 4 is a detailed flow chart of step S200 shown in FIG. 2 . FIG. 5 is a detailed flow chart of step S300 shown in FIG. 2 . FIG. 6 is a detailed flow chart of step S400 shown in FIG. 2 . FIG. 7 is a detailed flow chart of step S500 shown in FIG. 2 .

S100~S500: Steps

Claims (8)

A system for obtaining an accurate surname comprises: a storage medium storing a plurality of modules, wherein the plurality of modules comprises an accurate surname confirmation module, a similar sound acquisition module, a TOP N surname generation module and an accurate surname answer sentence generation module; and a processor coupled to the storage medium, wherein the processor accesses and executes the plurality of modules to perform the following steps: when the accurate surname confirmation module determines that a text does not include an accurate surname statement and does not include a surname confirmation sentence, the accurate surname confirmation module extracts a surname from the text; the similar sound acquisition module obtains a similar musical note text sequence using the surname; the TOP N surname generation module obtains a TOP N surname using the similar musical note text sequence; and the accurate surname answer sentence generation module generates a sentence corresponding to the TOP An accurate surname answer question sentence for the surname N, wherein the accurate surname answer question sentence includes a first surname and a second surname, and the first surname and the second surname have similar pronunciations. A system as described in claim 1, wherein the storage medium further stores a surname statement database, wherein the precise surname confirmation module uses the surname statement database to determine whether the text includes the precise surname statement; When the precise surname confirmation module determines that the text includes the precise surname statement, the precise surname confirmation module ends the dialogue with the client device; when the precise surname confirmation module determines that the text does not include the precise surname statement, the precise surname confirmation module determines whether the text includes the confirmation surname sentence; when the precise surname confirmation module determines that the text includes the confirmation surname sentence, the precise surname confirmation module ends the dialogue with the client device. A system as described in claim 1, wherein the similar sound acquisition module removes the Chinese tones in the surname to obtain the retained Zhuyin; the similar sound acquisition module uses the Zhuyin consonant vowel table and the retained Zhuyin to obtain the similar phonetic character sequence. A system as described in claim 1, wherein the storage medium further stores a database of Hundred Family Surnames, wherein the TOP N Surname generation module uses the similar phonetic character sequence and the Hundred Family Surnames database to generate a possible surname sequence; the TOP N Surname generation module selects the TOP N surnames from the possible surname sequence. The system as described in claim 1, wherein the storage medium further stores a surname statement database, wherein the precise surname answer sentence generation module uses the surname statement database to select the precise surname statement corresponding to the TOP N surnames; The precise surname answer sentence generation module uses the precise surname statement to generate the precise surname answer sentence corresponding to the TOP N surnames. A system as described in claim 1, wherein the processor repeatedly executes each of the steps; when the number of times the processor repeatedly executes each of the steps is equal to a number threshold value, the processor stops executing each of the steps. The system as described in claim 1 further includes a speech recognition and text conversion module, wherein the storage medium further stores a Chinese character library, wherein the speech recognition and text conversion module receives speech from a client device; the speech recognition and text conversion module performs a speech recognition operation on the speech to obtain recognized speech; the speech recognition and text conversion module uses the recognized speech and the Chinese character library to obtain the text corresponding to the speech. A method for obtaining an accurate surname is suitable for being executed by a system including an accurate surname confirmation module, a similar sound acquisition module, a TOP N surname generation module and an accurate surname answer sentence generation module, wherein the method comprises the following steps: when the accurate surname confirmation module determines that a text does not include an accurate surname statement and does not include a surname confirmation sentence, the accurate surname confirmation module extracts a surname from the text; the similar sound acquisition module uses the surname to obtain a similar musical note character sequence; the TOP N surname generation module uses the similar musical note character sequence to obtain the TOP N surnames; and the accurate surname answer sentence generation module generates an accurate surname answer sentence corresponding to the TOP N surnames, wherein the accurate surname answer sentence includes a first surname and a second surname, and the first surname and the second surname have similar pronunciations.

Images