Input support program and input support method

Abstract

To provide an input support program and input support method capable of reducing user's manual input operations in speech recognition.SOLUTION: An input support program for the present invention causes a computer to perform the steps of: recording product list data including a plurality of product data; recording user list data including user ID data; acquiring phonetic candidate list data based on voice data; acquiring and displaying product candidate list data based on the phonetic candidate list data; and recording unique candidate data for product data selected from the plurality of product candidate list data.SELECTED DRAWING: Figure 3

Description

[Technical field]

[0001] The present invention relates to an input support program and an input support method. [Background technology]

[0002] In manufacturing companies that directly handle products and businesses that provide services, inventory management of the products handled is an extremely important issue.

[0003] On the other hand, with the recent development of IT technology, IT technology is being actively adopted for inventory management of these products, i.e., inventory management is becoming increasingly computerized.

[0004] In computerized inventory management, when inputting products and their stock quantities, a user generally uses an input device such as a keyboard to record the information into an information processing device (a so-called computer).

[0005] However, when using an input device such as a keyboard that the user must operate with their hands, even if the user is using their hands to check products, they must first use those hands to operate the keyboard, resulting in reduced work efficiency.

[0006] In order to solve these work problems, a voice input device such as a microphone is sometimes used. The voice input device is generally connected to an information processing device that manages inventory, and can output the voice uttered by the user as voice data to the information processing device. Meanwhile, the information processing device compares this voice data with pre-recorded product data and identifies the product that the user has imagined, thereby solving the work problems such as operating a keyboard with the above-mentioned hands.

[0007] However, there are various issues in matching the voice data uttered by the user with the product data. Specifically, first, phonetic data, which is the reading of the voice data, is created from the voice data, and then the phonetic data is checked to see if it matches the reading data of the product data. If the phonetic data and the reading data of the product data match, there is no problem, but there are cases where they do not match due to the user's pronunciation environment, etc.

[0008] As a technique for dealing with such problems regarding voice input, for example, there is a technique described in Patent Document 1 below. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] Patent No. 6004502 Summary of the Invention [Problem to be solved by the invention]

[0010] The technology described in Patent Document 1 above discloses technology related to a POS terminal that displays in an identifiable manner when there is a mismatch between the voice-recognized product name and the product whose product button the operator (user) presses; however, this only indicates a mismatch and does not address the above-mentioned problem, i.e., the hassle of the user having to manually operate the keyboard when there is a mismatch between the voice-recognized product name and the product whose product button the user presses.

[0011] In view of the above, an object of the present invention is to provide an input assistance program and an input assistance method that can reduce manual input operations by a user in speech recognition. [Means for solving the problem]

[0012] An input support program according to one aspect of the present invention for solving the above problem causes a computer to execute the steps of recording product list data including multiple product data, recording user list data including user ID data, acquiring phonetic candidate list data based on voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and recording unique candidate data for product data selected from the multiple product candidate list data.

[0013] In addition, in this respect, although not limited thereto, it is preferable that the unique candidate data is recorded differently for each user ID data.

[0014] In addition, in this respect, although not limited thereto, it is preferable that the unique candidate data is created based on phonetic candidate list data.

[0015] In addition, in this respect, although not limited thereto, it is preferable that a plurality of unique candidate data are recorded and each unique candidate data includes frequency data.

[0016] In addition, in this aspect, although not limited thereto, it is preferable that the steps of acquiring phonetic candidate list data based on voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and recording unique candidate data for product data selected from the plurality of product candidate list data are omitted when one product data can be identified from the product list data based on the phonetic candidate list data, and are executed when one product data cannot be identified from the product list data based on the phonetic candidate list data.

[0017] In addition, in this respect, although not limited thereto, even if one product data cannot be identified from the product list data based on the phonetic candidate list data, if one unique candidate data can be identified, it is preferable to omit the steps of acquiring phonetic candidate list data based on voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and recording unique candidate data for product data selected from the multiple product candidate list data.

[0018] In addition, an input support method according to another aspect of the present invention executes the steps of recording product list data including a plurality of product data, recording user list data including user ID data, acquiring phonetic candidate list data based on voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and recording unique candidate data for product data selected from the plurality of product candidate list data. Effect of the Invention

[0019] As described above, according to the present invention, it is possible to provide an input assistance program and an input assistance method that can reduce the manual input operation of the user in speech recognition. [Brief description of the drawings]

[0020] [Figure 1] 1 is a diagram illustrating an outline of an example of an input support system according to an embodiment. [Diagram 2] 1 is a diagram illustrating an outline of an example of an input support system according to an embodiment. [Diagram 3] FIG. 4 is a diagram showing a flow of an input assistance program according to the embodiment. [Figure 4] 5A to 5C are diagrams showing an example of product data and product list data used in the input support program according to the embodiment. [Diagram 5] 11 is a diagram showing an image relating to user ID data used in the input support program according to the embodiment. FIG. [Figure 6] 1 is a diagram showing an image relating to phonetic candidate list data used in an input support program according to an embodiment; [Figure 7] 11 is a diagram showing an image of product candidate list data used in the input support program according to the embodiment. FIG. [Figure 8] 11 is a diagram showing an image of unique candidate data used in the input assistance program according to the embodiment; FIG. BEST MODE FOR CARRYING OUT THEINVENTION

[0021] Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings. However, the present invention can be embodied in many different forms, and is not limited to the specific examples shown in the following embodiments.

[0022] FIG. 1 is a diagram showing an outline of an input support system S according to this embodiment (hereinafter referred to as "this system").

[0023] As shown in the figure, the system is realized using an information processing device. More specifically, the system S includes a voice input device V, a manual input device H, a display device D, and an information processing device C connected to these input devices.

[0024] The information processing device C in this system S is preferably configured to include, but is not limited to, a central processing unit (CPU), which are components of a general information processing device, a non-volatile recording medium such as a hard disk or flash memory, a volatile recording medium such as a memory, a bus connecting these, and is configured to receive input of various data from a voice input device V and a manual input device H, perform predetermined processing, and display the results of the processing on a display device D. The information processing device C alone is generally called a computer, but in this system S, the voice input device V, voice output device M, manual input device H, and display device D are also expressed as a computer including the voice input device V, voice output device M, manual input device H, and display device D.

[0025] Moreover, the voice input device V in this system S is a device that can recognize the user's voice and output it as electronic data to the information processing device C. More specifically, the voice input device V accepts voice input from the user, converts the voice into an electrical signal, and outputs the voice data to the information processing device C. Note that the voice input device V is not limited to this, but can be exemplified by a so-called microphone or the like. As a result, the words uttered by the user can be processed as data by the voice input device V.

[0026] In addition, it is also preferable that the present system S has an audio output device M such as a speaker. The audio output device M has the opposite function to the audio input device V, and can receive input of audio data, process the audio data, and output it as audio that the user can recognize. As a result, if the necessary input instructions can be transmitted to the user as audio, rather than being displayed on the display device D as characters or the like, it becomes possible to perform input processing without looking at the display device D, which has the advantage of improving work efficiency.

[0027] As described above, the present system S is provided with a manual input device H such as a keyboard or a mouse that allows input processing to be performed using hands. As described above, the ultimate objective of the present system S is to have all data processing performed only by the voice input device V and the voice output device M, but at the present time, it is not realistic to completely not provide the manual input device H. For this reason, it is preferable to connect a manual input device H such as a keyboard or a mouse. In the case of a keyboard, a physical keyboard as a physical entity may be used, but it is also possible to use a so-called virtual keyboard in which, for example, a sensor that detects finger movements is arranged on the surface of the display device D to make it a touch screen and a keyboard layout is displayed on the display screen. In the case of a mouse, it is also possible to use not only a physical mouse but also the above-mentioned touch screen.

[0028] In addition, as described above, the present system S preferably includes a display device D such as a monitor. By providing the display device D, data created by the information processing device can be transmitted to the user as visual information. As described above, it is also a preferred example that a transparent electrode or the like can be arranged on the surface of the display device D to make it into a touch screen that recognizes the touch of the user.

[0029] The system S may be a so-called notebook type or desktop type, or may be a mobile information terminal, which has become increasingly popular in recent years, specifically a so-called smartphone or tablet terminal. A typical mobile information terminal is a device in which the above-mentioned CPU, display device, and other components are housed and integrated in a single cover, and a sensor is placed on the display device to make it a so-called touch screen, which can also function as an input device, making it very easy to handle and preferable. In the case of a mobile information terminal, a program for executing the method of the present invention is recorded and displayed as a so-called app on the mobile information terminal, and the method of the present invention can be easily executed by starting the app.

[0030] The present system S may also be configured via a telecommunication line, the so-called Internet I. Specifically, by connecting a plurality of information processing devices C through a telecommunication line via devices such as a modem or a router and transmitting and receiving necessary data, an environment for providing input support to a user can be prepared even in a remote location, and by connecting a plurality of information processing devices C, necessary data can be centrally managed and feedback can be provided appropriately in a franchise or the like that operates in a plurality of stores. In this case, the entire method can be implemented by executing the steps of the method executed by the present system S part by part on different computers. An example of the present system S in this case is shown in FIG. 2. In this case, one of the plurality of information processing devices C may be a centralized management information processing device MC, and the connection and management of these may be centralized.

[0031] As is clear from the above description, by using the present system S, it is possible to provide an input assistance program (hereinafter referred to as "the present program") and an input assistance method (hereinafter referred to as "the present method") according to this embodiment. Specifically, the present method can be realized by recording the present program on a recording medium of an information processing device and executing the present program. The present method and the present program will be described in detail below.

[0032] This program causes a computer to execute the steps of (S1) recording product list data including multiple product data, (S2) recording user list data including multiple user ID data, (S3) acquiring phonetic candidate list data based on voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and (S4) recording unique candidate data for product data selected from the multiple product candidate list data. By using this program, it is possible to provide an input assistance program and an input assistance method that can reduce the manual input operation of a user when using voice recognition technology for product management such as inventory management. Figure 3 shows a flow chart of the steps performed by this program.

[0033] As is clear from the above description, just to be clear, the method includes the steps of (S1) recording product list data including a plurality of product data, (S2) recording user list data including a plurality of user ID data, (S3) acquiring phonetic candidate list data based on voice data, and acquiring and displaying product candidate list data based on the phonetic candidate list data, and (S4) recording unique candidate data for product data selected from the plurality of product candidate list data.

[0034] First, the program has a step (S1) of recording product list data including multiple product data. In the program, "product list data" refers to data that includes multiple product data and is sometimes called a product master. Also, "product data" refers to data that includes information about products.

[0035] As described above, product data is data that contains information about products, such as the product name and the reading of the product. Figure 4 shows an image of product data and product list data. As shown in the figure, by adopting product data, it is possible to grasp the products that are the subject of inventory management. In the figure, the numbers and symbols such as No. 001 on the left side are product management number data, and each of these management number data is recorded in correspondence with a product name such as tomato or apple and its reading. Each of these numbers and product names is product data, and the entirety of this product data becomes product list data.

[0036] This program also has a step (S2) of recording user list data including multiple user ID data. Generally, when performing inventory management, the inventory management may be performed by a single user, but since inventory management needs to be performed periodically and is often performed by different employees at different times due to working hours and the like, it is preferable to have multiple users. When using voice recognition technology, each user's voice has its own characteristics, and it is important to link each user's unique characteristics to that person. This can be achieved by recording multiple user ID data to identify multiple users.

[0037] As is clear from the above description, "user ID data" is data that includes information about the user ID. Here, user ID refers to the user's ID (identification number). Figure 5 shows an image of user list data and user ID data. The number is the user ID, and the user's name and pronunciation, etc. are recorded in correspondence with this user ID. This makes it possible to create a user list, which becomes the user list data.

[0038] The program also includes a step of (S3) acquiring phonetic candidate list data based on the voice data, and acquiring and displaying product candidate list data based on the phonetic candidate list data. According to this step, the product candidate list data is displayed based on the voice data.

[0039] Here, "audio data" refers to data including information about the voice uttered by the user, and more specifically, data including information about the voice received by the audio output device M, preferably raw data of the voice. The audio data is not limited as long as it is possible to create the phonetic candidate list data, which will be described in detail later, based on the audio data, but for example, data recorded in an appropriate format based on information in which the horizontal axis corresponds to time and the vertical axis corresponds to the volume (intensity) of the voice can be exemplified. This makes it possible to handle information about the voice in an information processing device. Note that audio data is generally saved according to a standard, and for example, ".MP3" or ".WAV" can be adopted, but audio data does not necessarily need to be saved as long as it is possible to create the phonetic candidate list, and is not limited to this.

[0040] Furthermore, in this step, the "phonetic candidate list data" is data including information on the phonetic candidate list. Here, "phonetic candidate" refers to candidates for what the user actually said (whether it was pronounced), and the phonetic candidate list refers to a set (list) of multiple phonetic candidates. Each phonetic candidate in the phonetic candidate list data is referred to as "phonetic candidate data." Figure 6 shows an image of phonetic candidate list data. This figure shows an example of a case where, for example, when the user utters "akakakubi," the computer displays candidates for what the user pronounced from the corresponding voice data. This is an example where the most likely possibilities are displayed from the top.

[0041] Although voice data and phonetic candidate data are common in that they are based on the voice uttered by the user, they are different in that "voice data" is actual data (raw data) that records the voice that the user actually uttered, while phonetic candidate data is candidate data of what the user was actually going to say based on this voice data. For example, if the recording state of the voice data is perfect, and the user also speaks clearly and articulates well, and as a result, the user's intention can be completely understood without any other options, the phonetic candidate is uniquely determined, and there is no need to provide multiple phonetic candidates. However, it is not easy to uniquely determine the phonetic candidate due to various factors such as poor pronunciation of the user, problems with the voice input device used for recording, ambient sounds being picked up during recording, and unnecessary sounds being added to the beginning or end of a word due to a shift in recording timing. For example, even if a user thinks of radish and pronounces it as "daikon," if this is converted into voice data and converted into phonetic candidate data by an information processing device, the phonetic candidate may become "zaikon." This causes a discrepancy between the user's voice and the phonetic candidates, so it is important to create multiple phonetic candidates (phonetic candidate list) to eliminate this discrepancy. Also, to repeat, in the example shown above, when the user pronounces "akakakubi," "akakakubi" is displayed at the top, but considering the possibility that "mi" may have been pronounced as "bi," processing is performed to include "akakumi" as a candidate. The same applies to other examples.

[0042] In addition, in this step, the method of creating phonetic candidate data from the voice data can be realized by using a known technology, for example, a voice recognition technology for converting the voice data into text data can be used. The voice recognition technology is not limited as long as it can create phonetic candidate data from the voice data, and examples of the voice recognition technology include finem recognition technology, vocabulary recognition technology, and recognition technology using a neural network.

[0043] In addition, in this step, product candidate list data is acquired and displayed based on the phonetic candidate list data. Here, the "product candidate list data" is data including information about the product candidate list, and the product candidate list is a set (list) of products (product candidates) that are determined to be related to the acquired phonetic candidate list data from a pre-established product list. In other words, by acquiring and displaying the product candidate list, it is possible to identify which product in the product list the user was referring to. FIG. 7 shows an image diagram of the product candidate list data. As described above, the product candidate list data includes multiple product candidate data, but the "product candidate data" is selected from the product list data. Therefore, the product data and the product candidate data are the same, but the product candidate data selected from the product list data becomes the product candidate data. In the example shown in this figure, "red neck radish" is displayed at the top because it is determined to be the closest from the phonetic candidate list, but there is a possibility that it is "red miso" or "red shiso", and an image of the display in this list is shown.

[0044] In addition, this step is based on the phonetic candidate list data, and may be based on all phonetic candidate data, or may be based on a portion of the phonetic candidate list data, more specifically, one phonetic candidate data. If all phonetic candidate data is used, the amount of data processing increases accordingly, and the processing time becomes too long. Therefore, it is possible to balance the amount of processing and accuracy by performing the processing based on only a portion of the phonetic candidate data, for example, three or less, two or less, or even one phonetic candidate data. Furthermore, based on experience, sufficient accuracy can be ensured even with one phonetic candidate data, which is considered to have the highest priority.

[0045] In this case, if the product that the user has in mind is not in the product candidate list, it is preferable to display a list of all products. In this case, it is preferable to display a button for displaying all products on the display device, and when the user presses this button, to display all products.

[0046] In addition, in this step, the priority order and arrangement of the product candidate data in the product candidate list data is not limited, but it is preferable to perform a predetermined process on the phonetic candidate data and the product candidate data to find their closeness and display them in order of closeness. This can improve the accuracy of displaying products that the user expects. The predetermined process is not limited, but it is preferable to perform a process of finding the Levenshtein distance. The Levenshtein distance is a value for evaluating the degree to which two character strings differ. The algorithm for the Levenshtein distance is well known.

[0047] In addition, in this step, from the viewpoint of explanation, a voice recognition technique for product data is used, but this is not limited to product data, and can also be applied to data such as unit numbers, such as quantity, etc. (unit data).

[0048] The program also includes a step (S4) of recording unique candidate data for product data selected from a plurality of product candidate list data. Here, "unique candidate data" refers to data including information about the unique candidate, and "unique candidate" refers to a candidate that is uniquely identified to a user. More specifically, each user has a characteristic way of speaking, and if a uniform process was performed for all users, it would not be possible to match the characteristics of each user, and the product candidate list would have to be displayed at all times. However, displaying the product candidate list would result in a loss of time, so by recording unique candidate data for each user, it is possible to reduce the display of the product candidate list as much as possible.

[0049] As described above, the unique candidate data is recorded in correspondence with the product data. That is, each product data has unique candidate data. An image of this case is shown in FIG. 8. In this figure, for example, when a user selects "red neck radish", the phonetic candidate list data or phonetic candidate data created when selecting this product is associated with the product data of red neck radish. In this way, by providing the unique candidate data in correspondence with the product data, even if there is no corresponding product list data when the phonetic candidate list data is acquired, it is possible to specify the product that the user intended by referring to the unique candidate data in the product list data, and it is possible to reduce the number of times the product candidate list is displayed.

[0050] As is clear from the above description, the above-mentioned effects can be obtained by recording the unique candidate data in accordance with the characteristics of each user, so it is preferable that the unique candidate data be recorded differently for each user ID data.

[0051] As described above, the unique candidate data is not limited as long as it is recorded according to the characteristics of the user and can reduce the number of times the product candidate list is displayed, but it is preferable that the unique candidate data is created based on the phonetic candidate list data, and more specifically, it is preferable that it is any one of the phonetic candidate data from the phonetic candidate list data. Since the phonetic candidate list is a candidate list of words that the user is likely to have uttered, accuracy can be improved by characterizing all of the acquired phonetic candidates for the user. However, if all of the phonetic candidates are recorded, the amount of data held will increase and processing will take time. Therefore, it is possible to ensure a balance by narrowing down the phonetic candidate list data to phonetic candidate data with high priority.

[0052] In addition, in this step, although not limited thereto, it is preferable that a plurality of unique candidate data are recorded and each of the unique candidate data includes frequency data. As described above, by recording unique candidate data for each product data, it is possible to match the product data using a plurality of unique candidate data, and therefore it is possible to reduce the number of times the product list data is displayed. This point is clear from the above. Also, by including frequency data in each of the unique candidate data, it is possible to prioritize the plurality of unique candidate data. For example, when the frequency exceeds a certain level, when the unique candidate data is detected, it becomes easier to select the product without displaying the product list data. Also, by prioritizing, it is possible to change the order of data matching and improve the processing speed. In the case of the above figure, for example, there are "Akakumi", "Akakuri", "Akasumi", and "Sakakubi", each of which is unique candidate data, and each of which is displayed as the numbers "5", "3", "1", and "1", which are frequency data. According to this example, when "Akakubi Daikon" is selected, it is shown that the number of times that "Akakumi" and phonetic candidate data have been created is 5.

[0053] In the above example, the unique candidate data is generally selected from among product data whose phonetic candidates are thought to be similar to the sound uttered by the user. However, it is not uncommon for the phonetic candidates to be completely different from the sound of the selected product, such as an abbreviation or dialect. Therefore, the correspondence between the unique candidate data and the product data does not need to be determined by the Levenshtein distance or the like. To explain more specifically, while the product data is "onigiri", it is not rare for the product data to be "omusubi" depending on the region or user. In such a case, even if "omusubi" is used, it is difficult to determine that "onigiri" is close, but it is very inefficient to display the product list every time. Therefore, if the phonetic candidate is "omusubi" and "onigiri" is selected when the product list is displayed, it is preferable to record this "omusubi" as the unique candidate data. This makes it possible to provide a system customized for each user, region, etc.

[0054] In addition, in this program, although not limited thereto, it is preferable that the step of (S3) acquiring phonetic candidate list data based on the voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and the step of (S4) recording unique candidate data for product data selected from a plurality of product candidate list data are omitted when one product data can be identified from the product list data based on the phonetic candidate list data, and are executed when one product data cannot be identified from the product list data based on the phonetic candidate list data. As described above, when phonetic candidate list data is acquired, product data corresponding to the phonetic candidate data in the phonetic candidate list data is searched for, but when the phonetic candidate data and the product data match, it is not necessary to acquire and display the product candidate list data. Therefore, the above steps (S3) and (S4) can be omitted. In this case, although it is possible to record, since it is possible to identify the product data from the phonetic candidate data, it is not necessarily necessary to record the unique candidate data.

[0055] In addition, in this program, even if one product data cannot be identified from the product list data based on the phonetic candidate list data, if one unique candidate data can be identified, it is preferable to omit the steps of (S3) acquiring phonetic candidate list data based on the voice data and acquiring and displaying the product candidate list data based on the phonetic candidate list data. If one unique candidate data can be identified, it means that the product data corresponding to the unique candidate data can also be identified. As a result, sufficient accuracy can be maintained even if the product candidate list data is not displayed. For example, if the phonetic candidate data is created when the frequency data exceeds a certain value, and the phonetic candidate data is created again, the one product data can be identified immediately, thereby omitting unnecessary processing.

[0056] In addition, this program is explained using inventory management as an example because this is the initial purpose and for ease of explanation, but it can be applied to purposes other than inventory management. Since this program aims to link the individual user's pronunciation characteristics with multiple pre-recorded types of goods and services, it can be applied not only to inventory management, but also to order processing and attendance management in stores. [Industrial Applicability]

[0057] As described above, the present invention can provide an input assistance program and an input assistance method that can reduce the manual input operation of the user in speech recognition.

Claims

1. On the computer, recording product list data including a plurality of product data; recording user list data including user ID data; A step of acquiring phonetic candidate list data based on the voice data, and acquiring and displaying product candidate list data based on the phonetic candidate list data; a step of recording unique candidate data for the product data selected from the plurality of product candidate list data.

2. 2. The input support program according to claim 1, further comprising a step of recording user list data including a plurality of user ID data.

3. 2. The input support program according to claim 1, wherein the unique candidate data is recorded differently for each of the user ID data.

4. 2. The input support program according to claim 1, wherein the unique candidate data is generated based on the phonetic candidate list data.

5. 2. The input support program according to claim 1, wherein a plurality of said unique candidate data are recorded, and each of said unique candidate data includes frequency data.

6. The steps of acquiring phonetic candidate list data based on the voice data, acquiring and displaying product candidate list data based on the phonetic candidate list data, and recording unique candidate data for product data selected from the plurality of product candidate list data include: If one of the product data items can be identified from the product list data based on the phonetic candidate list data, the step is omitted.

2. The input support program according to claim 1, which is executed when one of said product data cannot be specified from said product list data based on said phonetic candidate list data.

7. Even if one of the product data cannot be identified from the product list data based on the phonetic candidate list data, When one of the unique candidate data is identified, 7. The input support program according to claim 6, further comprising: a step of acquiring phonetic candidate list data based on the voice data; acquiring and displaying product candidate list data based on the phonetic candidate list data; and a step of recording unique candidate data for product data selected from the plurality of product candidate list data.

8. recording product list data including a plurality of product data; recording user list data including user ID data; A step of acquiring phonetic candidate list data based on the voice data, and acquiring and displaying product candidate list data based on the phonetic candidate list data; a step of recording unique candidate data for product data selected from the plurality of product candidate list data.

Images