Information processing device, information processing method, and recording medium
Patent Information
- Authority / Receiving Office
- JP · JP
- Patent Type
- Applications
- Filing Date
- 2023-11-16
- Publication Date
- 2025-05-22
Abstract
Description
Information processing device, information processing method, and recording medium
[0001] The present disclosure relates to the technical fields of an information processing device, an information processing method, and a recording medium.
[0002] There are known devices that perform speech recognition processing to convert speech data into text. For example, Patent Literature 1 discloses a technology that, when performing speech recognition on an utterance in a speech sequence, uses information from the first utterance in the utterance sequence to the utterance immediately before the target utterance as context.
[0003] Japanese Patent Application Laid-Open No. 2020-129015
[0004] An object of this disclosure is to provide an information processing device, an information processing method, and a recording medium that aim to improve upon the techniques disclosed in prior art documents.
[0005] One aspect of the information processing device disclosed herein includes a conversion means for converting a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, and an output means for recognizing the recognition target using the vector sequence of the fixed token length and outputting a string that is the recognition result.
[0006] One aspect of the information processing method disclosed herein is to convert, by at least one computer, a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, recognize the recognition target using the vector sequence of the fixed token length, and output a string that is the recognition result.
[0007] One aspect of the recording medium of this disclosure is a computer program recorded on at least one computer that causes the computer to execute an information processing method, which converts a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, recognizes the recognition target using the vector sequence of the fixed token length, and outputs a string that is the recognition result.
[0008] 1 is a block diagram showing the hardware configuration of a first information processing device; 2 is a block diagram showing the functional configuration of the first information processing device; 3 is a flowchart showing the operation flow of the first information processing device; 4 is a block diagram showing the configuration of a second information processing device; 5 is a flowchart showing the operation flow of the second information processing device; 6 is a block diagram showing the configuration of a third information processing device; and 7 is a flowchart showing the operation flow of the third information processing device.
[0009] Hereinafter, embodiments of an information processing device, an information processing method, and a recording medium will be described with reference to the drawings.
[0010] First Embodiment A first information processing apparatus will be described with reference to FIGS. 1 to 3. FIG.
[0011] (Hardware Configuration) First, the hardware configuration of the first information processing apparatus will be described with reference to Fig. 1. Fig. 1 is a block diagram showing the hardware configuration of the first information processing apparatus.
[0012] 1, a first information processing device 10 includes a processor 11, a RAM (Random Access Memory) 12, a ROM (Read Only Memory) 13, and a storage device 14. The information processing device 10 may further include an input device 15 and an output device 16. The processor 11, RAM 12, ROM 13, storage device 14, input device 15, and output device 16 are connected to each other via a data bus 17. The data bus 17 may be an interface other than a data bus (for example, a LAN, a USB, etc.).
[0013] The processor 11 loads a computer program. For example, the processor 11 is configured to load a computer program stored in at least one of the RAM 12, the ROM 13, and the storage device 14. Alternatively, the processor 11 may load a computer program stored in a computer-readable storage medium using a storage medium reading device (not shown). The processor 11 may acquire (i.e., load) the computer program from a device (not shown) located outside the information processing device 10 via a network interface. The processor 11 controls the RAM 12, the storage device 14, the input device 15, and the output device 16 by executing the loaded computer program. In particular, in this embodiment, when the processor 11 executes the loaded computer program, functional blocks for performing recognition processing are realized within the processor 11. In other words, the processor 11 may function as a controller that executes each control in the information processing device 10.
[0014] The processor 11 may be configured as, for example, a central processing unit (CPU), a graphics processing unit (GPU), a field-programmable gate array (FPGA), a digital signal processor (DSP), an application-specific integrated circuit (ASIC), or a quantum processor. The processor 11 may be configured as one of these, or may be configured to use multiple processors in parallel.
[0015] The RAM 12 temporarily stores computer programs executed by the processor 11. The RAM 12 temporarily stores data that the processor 11 temporarily uses while it is executing the computer programs. The RAM 12 may be, for example, a dynamic random access memory (D-RAM) or a static random access memory (SRAM). Alternatively, other types of volatile memory may be used instead of the RAM 12.
[0016] The ROM 13 stores computer programs executed by the processor 11. The ROM 13 may also store fixed data. The ROM 13 may be, for example, a programmable read-only memory (PROM) or an erasable read-only memory (EPROM). Alternatively, other types of non-volatile memory may be used instead of the ROM 13.
[0017] The storage device 14 stores data that the information processing device 10 stores for a long period of time. The storage device 14 may operate as a temporary storage device for the processor 11. The storage device may store computer programs executed by the processor 11. The storage device 14 may include, for example, at least one of a hard disk device, a magneto-optical disk device, an SSD (Solid State Drive), and a disk array device.
[0018] The input device 15 is a device that receives input instructions from a user of the information processing device 10. The input device 15 may include, for example, at least one of a keyboard, a mouse, and a touch panel. The input device 15 may be configured as part of a smartphone, a tablet terminal, an earphone-type terminal, a watch-type terminal, an HMD (Head Mounted Display) terminal, etc. The input device 15 may be, for example, a device that includes a microphone and is capable of voice input.
[0019] The output device 16 is a device that outputs information related to the information processing device 10 to the outside. For example, the output device 16 may be a display device (e.g., a display or digital signage) that can display information related to the information processing device 10. The output device 16 may also be a speaker or the like that can output information related to the information processing device 10 as audio.
[0020] 1 may be configured to be included in a device external to the first information processing device 10. For example, the first information processing device 10 may be configured to include a processor 11, a RAM 12, and a ROM 13, and the other devices, such as a storage device 14, an input device 15, and an output device 16, may be configured as external devices. That is, the first information processing device 10 may be configured as an information processing system including a plurality of different devices. Furthermore, some of the calculation functions of the first information processing device 10 may be realized by an external server, a cloud, or the like.
[0021] (Functional Configuration) Next, the functional configuration of the first information processing device 10 will be described with reference to Fig. 2. Fig. 2 is a block diagram showing the functional configuration of the first information processing device.
[0022] 2, the first information processing device 10 is configured as a device that executes recognition processing on a recognition target and outputs a character string that is the recognition result. The first information processing device 10 is configured to include a conversion unit 110 and a recognition result output unit 120 as processing blocks for realizing its functions. Note that each of the conversion unit 110 and the recognition result output unit 120 may be realized by, for example, the above-mentioned processor 11 (see FIG. 1). The conversion unit 110 and the recognition result output unit 120 may be part of a model configured by a neural network.
[0023] The conversion unit 110 is configured to convert related strings or related sentence vectors into vector sequences of fixed token lengths. Related strings are strings related to the recognition target. More specifically, related strings are strings containing information that can be used when recognizing the recognition target. For example, related strings may be strings containing information that can improve the accuracy of recognizing the recognition target. More specific examples of related strings will be described in detail in other embodiments below. On the other hand, related sentence vectors are sentence vectors extracted from related strings. Since related sentence vectors are extracted from related strings, they contain information that can be used when recognizing the recognition target, just like related strings. The vector sequences of fixed token lengths converted by the conversion unit 110 are configured to be output to the recognition result output unit 120.
[0024] The recognition result output unit 120 is configured to be able to recognize a recognition target using a vector sequence of fixed token length input from the conversion unit 110. Specifically, the recognition result output unit 120 is configured to be able to execute a recognition process on the recognition target, taking into account the content indicated by the vector sequence of fixed token length. The recognition result output unit 120 is also configured to be able to output a character string that is the recognition result of the recognition target. The recognition result output unit 120 may be configured to execute the recognition process sequentially and output the characters that are the recognition result one by one. The recognition result output unit 120 may be configured as a decoder that receives the vector sequence of fixed token length and the recognition target as input, and outputs a character string that is the recognition result of the recognition target.
[0025] (Flow of Operation) Next, the flow of operation in the first information processing device 10 will be described with reference to Fig. 3. Fig. 3 is a flowchart showing the flow of operation of the first information processing device.
[0026] 3, when the operation of the first information processing device 10 is started, the conversion unit 110 first acquires a related character string or a related sentence vector (step S101). Then, the conversion unit 110 converts the acquired related character string or related sentence vector into a vector sequence of a fixed token length (step S102).
[0027] Next, the recognition result output unit 120 recognizes the recognition target using the vector sequence of the fixed token length input from the conversion unit 110 (step S103), and outputs a character string that is the recognition result of the recognition target (step S104).
[0028] The recognition result output unit 120 may acquire all recognition targets in advance and execute recognition processing for all of the recognition targets at once. In this case, the recognition result output unit 120 may output the recognition results for the recognition targets at once. Alternatively, the recognition result output unit 120 may acquire recognition targets sequentially and execute recognition processing for each acquired target. In this case, the recognition result output unit 120 may output the recognition results for each recognized target.
[0029] (Technical Effects) Next, technical effects obtained by the first information processing device 10 will be described.
[0030] As described with reference to FIGS. 1 to 3 , in the first information processing device 10, related character strings or related sentence vectors are converted into vector sequences of fixed token lengths. Then, the recognition target is recognized using the vector sequences of fixed token lengths. This makes it possible to perform highly accurate recognition processing while suppressing an increase in the amount of calculations. Specifically, by converting related character strings or related sentence vectors into vector sequences of fixed token lengths, the amount of calculations required for the recognition processing can be reduced. For example, compared to inputting related character strings directly to the recognition result output unit 120, the amount of calculations required for recognition can be reduced. Note that the value of the fixed token length is not particularly limited, but it is preferably short enough to achieve the above-described technical effect. Furthermore, by using related character strings related to the recognition target in recognizing the recognition target, the recognition accuracy of the related character string can be improved compared to when the related character strings are not used.
[0031] (Specific Application Examples) Here, specific application examples of the above-described first information processing apparatus 10 will be described.
[0032] The first information processing device 10 may be applied to a speech recognition device. Specifically, the first information processing device may be configured as a device that converts speech data to be recognized into text and outputs the text. In this case, the related character string may be acquired as a character string related to the speech data to be recognized. When the first information processing device 10 is applied to a speech recognition device, speech recognition can be performed with high accuracy while suppressing an increase in the amount of calculation required for speech recognition.
[0033] The first information processing device 10 may be applied to a translation device. Specifically, the first information processing device may be configured as a device that outputs text data obtained by translating text data to be recognized into another language. In this case, the related character string may be acquired as a character string related to the text data to be recognized (i.e., the data to be translated). When the first information processing device 10 is applied to a translation device, machine translation can be performed with high accuracy while suppressing an increase in the amount of calculation related to the translation process.
[0034] The first information processing device 10 may be applied to an image recognition device. Specifically, the first information processing device may be configured as a device that recognizes characters included in image data to be recognized and outputs the characters as text data. In this case, the related character string may be acquired as a character string related to the image data to be recognized. When the first information processing device 10 is applied to an image recognition device, image recognition can be performed with high accuracy while suppressing an increase in the amount of calculation related to image processing.
[0035] The above-described application examples are merely examples, and the application of the first information processing device 10 is not limited to the above-described devices. The first information processing device 10 can be appropriately applied to any device that executes recognition processing on data to be recognized and outputs some character string as the recognition result.
[0036] Second Embodiment A second information processing device 10 will be described with reference to Figures 4 and 5. The second information processing device 10 differs in some configurations and operations from the first information processing device 10 described above, but other parts may be similar to the first information processing device 10. Therefore, the following will describe in detail the parts that differ from the first embodiment, and will omit explanations of other overlapping parts as appropriate.
[0037] (Device Configuration) First, the configuration of the second information processing device 10 will be described with reference to Fig. 4. Fig. 4 is a block diagram showing the configuration of the second information processing device. Note that in Fig. 4, the same elements as those described in Fig. 2 are denoted by the same reference numerals.
[0038] In Fig. 4, the second information processing device 10 is configured as a voice recognition device. That is, the second information processing device 10 is configured as a device that converts voice data to be recognized into text and outputs the text. The second information processing device 10 is configured to sequentially recognize voice data and output the recognition result character by character. Note that, for the sake of convenience of explanation, Fig. 4 illustrates only the decoder side configuration of the voice recognition device, but the second information processing device 10 may also include the encoder side configuration.
[0039] In the second information processing device 10, a related character string "for speech recognition..." is input to the conversion unit 110. The conversion unit 110 converts this related character string into a vector sequence of a fixed token length. Hereinafter, the vector sequence converted by the conversion unit 110 will be referred to as an "related embedding sequence" as appropriate.
[0040] The related strings here are acquired as strings indicating the content of the speech data to be recognized (i.e., the spoken content). For example, when attempting to recognize the spoken content of a meeting, the related strings may be strings summarizing the content of the agenda to be spoken. Alternatively, the related strings may be text data extracted from part or all of the materials to be used in the meeting. By acquiring related strings related to the content to be recognized in this way, the accuracy of the recognition process can be improved using the related embedding sequence. For example, by using the related embedding sequence as context information, it becomes possible to predict the spoken content to some extent, thereby achieving more accurate speech recognition.
[0041] The related string may indicate information other than the content of the recognition target. For example, the related string may indicate the type of language of the recognition target (e.g., Japanese, English, etc.). Alternatively, the related string may indicate information about the sender of the recognition target. For example, the related string may be information indicating the gender, age, etc. of the speaker. Alternatively, the related string may indicate information about the environment in which the authentication target is recorded. For example, the related string may be a string indicating information about the location where the speech is spoken (e.g., a conference room, the inside of a car, etc.). Even when obtaining related strings related to something other than the content of the recognition target in this way, the accuracy of the recognition process can be improved by using related embedding sequences. The related string may be a string including multiple types of information.
[0042] While the associated embedding sequence is generated, the embedding layer 115 receives the most recent recognition result (i.e., the character string already output as the recognition result) as input and outputs an embedding sequence. Hereinafter, this embedding sequence will be referred to as the "recognition result embedding sequence" as appropriate. The recognition result embedding sequence is combined with the associated embedding sequence and input to the recognition result output unit 120.
[0043] The recognition result output unit 120 takes as input the combined embedding sequence obtained by combining the related embedding sequence and the recognition result embedding sequence, and outputs the next character which is the recognition result of the recognition target. For example, in the example shown in FIG. 4, a character string "recognize voice" is obtained as the previous recognition result. In this case, the recognition result output unit 120 outputs 1 character (for example, "识") following "recognize voice" as the recognition result.
[0044] (Operation flow) Next, the operation flow in the second information processing apparatus 10 will be described while referring to FIG. 5. FIG. 5 is a flowchart showing the operation flow of the second information processing apparatus. In FIG. 5, the same reference numerals are assigned to the same processes as those described in FIG. 3.
[0045] As shown in FIG. 5, when the operation in the second information processing apparatus 10 is started, first, the conversion unit 110 acquires the related character string (step S201). Then, the conversion unit 110 converts the acquired related character string into an embedding sequence with a fixed token length (that is, the related embedding sequence) (step S202).
[0046] Subsequently, the embedding layer 115 extracts an embedding sequence (that is, the recognition result embedding sequence) from the recognition results up to the previous time (step S203). Then, the related embedding sequence and the recognition result embedding sequence are combined (step S204).
[0047] Subsequently, the combined embedding sequence obtained by combining the related embedding sequence and the recognition result embedding sequence is input to the recognition result output unit 120 (step S205). Thereafter, the recognition result output unit 120 outputs the next character which is the recognition result (step S206).
[0048] Next, the second information processing device 10 determines whether to end the recognition (step S207). For example, the second information processing device 10 may determine to end the recognition when recognition of all acquired recognition targets is completed. Alternatively, the second information processing device 10 may determine to end the recognition when it detects that the speaker has finished speaking. Alternatively, the second information processing device 10 may determine to end the recognition when it detects various operations by the user (for example, operation of a recognition end button).
[0049] If it is determined that recognition should be terminated (step S207: YES), the series of operations ends. On the other hand, if it is determined that recognition should not be terminated (step S207: NO), the process resumes from step S203. By repeating the process in this manner, characters that are the recognition results are output one by one.
[0050] (Technical Effects) Next, technical effects obtained by the second information processing device 10 will be described.
[0051] As explained in Figures 4 and 5, the second information processing apparatus 10 inputs a combined embedding sequence that combines a fixed-token-length related embedding sequence and a recognition result embedding sequence, and outputs the characters that are the recognition result one by one. In this way, it is possible to sequentially recognize the recognition target with high accuracy while taking into account the information indicated by the related character string. Furthermore, because the related embedding sequence has a fixed token length, it is also possible to suppress an increase in the amount of calculation.
[0052] <Third Embodiment> A third information processing device 10 will be described with reference to Figures 6 and 7. The third information processing device 10 differs in some configurations and operations from the first and second information processing devices 10 described above, but other parts may be similar to the first and second information processing devices 10. Therefore, the following will describe in detail the parts that differ from the embodiments already described, and will omit explanations of other overlapping parts as appropriate.
[0053] (Device Configuration) First, the configuration of the third information processing device 10 will be described with reference to Fig. 6. Fig. 6 is a block diagram showing the configuration of the third information processing device. Note that in Fig. 6, the same elements as those described in Fig. 4 are denoted by the same reference numerals.
[0054] 6, the third information processing device 10 is configured as a speech recognition device, similar to the above-described second information processing device 10. The third information processing device 10 further includes a sentence vector extraction unit 200 in addition to the configuration of the above-described second information processing device (see FIG. 4).
[0055] The sentence vector extraction unit 200 is configured to extract related sentence vectors from sentence data, which are related character strings. The related sentence vectors extracted by the sentence vector extraction unit 200 are input to the conversion unit 100, where they are converted into embedding sequences of fixed token lengths (i.e., related embedding sequences).
[0056] The related sentence vector may appropriately include information for improving the recognition accuracy of the recognition target. For example, the sentence vector extraction unit 200 may be configured to extract elements that can be used to improve the recognition accuracy from relatively long sentence data.
[0057] The related sentence vector may be a vector in a state suitable for conversion to a related embedding sequence. For example, the sentence vector extraction unit 200 may be configured to extract a fixed-length vector suitable for conversion in the conversion unit 110. However, the length of the related sentence vector may be different from the length of the related embedding sequence after conversion.
[0058] (Operation Flow) Next, the operation flow in the third information processing device 10 will be described with reference to Fig. 7. Fig. 7 is a flowchart showing the operation flow of the third information processing device. Note that in Fig. 7, the same processes as those described in Fig. 5 are denoted by the same reference numerals.
[0059] 7, when the operation of the third information processing apparatus 10 is started, the sentence vector extraction unit 200 first acquires sentence data (step S301), and then extracts related sentence vectors from the sentence data (step S302).
[0060] Next, the conversion unit 110 converts the related sentence vector extracted by the sentence vector extraction unit 200 into an embedding sequence of a fixed token length (i.e., a related embedding sequence) (step S303).
[0061] Next, the embedding layer 115 extracts an embedding sequence (i.e., a recognition result embedding sequence) from the immediately preceding recognition result (step S203), and then combines the associated embedding sequence with the recognition result embedding sequence (step S204).
[0062] Next, the combined embedding sequence obtained by combining the associated embedding sequence and the recognition result embedding sequence is input to the recognition result output unit 120 (step S205). After that, the recognition result output unit 120 outputs the next character as the recognition result (step S206).
[0063] Next, the second information processing apparatus 10 determines whether or not to end the recognition (step S207). If it is determined that the recognition should be ended (step S207: YES), the series of operations ends. On the other hand, if it is determined that the recognition should not be ended (step S207: NO), the process resumes from step S203.
[0064] (Technical Effects) Next, technical effects obtained by the third information processing apparatus 10 will be described.
[0065] As described with reference to Figures 6 and 7, in the third information processing device 10, a related sentence vector is extracted from sentence data that is a related string, and the related sentence vector is converted into a related embedding sequence of a fixed token length. In this way, the conversion unit 110 can more appropriately perform conversion into a related embedding sequence. As a result, for example, the processing in the conversion unit 110 can be simplified. It is also possible to prevent information indicated by a related string from being lost during conversion into a related embedding sequence.
[0066] The scope of each embodiment also includes a processing method in which a program that operates the configuration of each embodiment to realize the functions of the above-described embodiments is recorded on a recording medium, the program recorded on the recording medium is read as code, and the program is executed on a computer. In other words, a computer-readable recording medium is also included in the scope of each embodiment. Furthermore, each embodiment includes not only a recording medium on which the above-described program is recorded, but also the program itself.
[0067] Examples of recording media that can be used include floppy disks, hard disks, optical disks, magneto-optical disks, CD-ROMs, magnetic tapes, non-volatile memory cards, and ROMs. Furthermore, the scope of each embodiment is not limited to programs that execute processes by themselves, but also includes programs that execute processes by operating on an OS in conjunction with other software or expansion board functions. Furthermore, the program itself may be stored on a server, and part or all of the program may be downloadable from the server to a user terminal. The program may be provided to the user in, for example, a SaaS (Software as a Service) format.
[0068] <Supplementary Notes> The above-described embodiment may be further described as in the following supplementary notes, but is not limited to the following.
[0069] (Supplementary Note 1) The information processing device described in Supplementary Note 1 is an information processing device that includes a conversion means that converts a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, and an output means that recognizes the recognition target using the vector sequence of the fixed token length and outputs a string that is the recognition result.
[0070] (Supplementary Note 2) The information processing device described in Supplementary Note 2 is the information processing device described in Supplementary Note 1, wherein the output means recognizes and outputs the recognition target character by character, and receives as input a combined vector sequence obtained by combining the vector sequence of the fixed token length and a vector sequence extracted from a character string that is the immediately preceding recognition result, and outputs a character that is the next recognition result.
[0071] (Supplementary Note 3) The information processing device according to Supplementary Note 3 is the information processing device according to Supplementary Note 1 or 2, further comprising an extraction unit that extracts the related sentence vector from the related character string.
[0072] (Supplementary Note 4) The information processing device according to Supplementary Note 4 is the information processing device according to any one of Supplementary Notes 1 to 3, in which the related character string is a character string including a sentence related to the content of the recognition target.
[0073] (Appendix 5) The information processing device described in Appendix 5 is the information processing device described in any one of Appendices 1 to 4, wherein the related character string is a character string indicating at least one of the type of language of the object to be recognized, the sender of the object to be recognized, and the recording environment of the object to be recognized.
[0074] (Supplementary Note 6) The information processing device described in Supplementary Note 6 is the information processing device described in any one of Supplementary Notes 1 to 5, wherein the recognition target is voice data, and the output means outputs data that has been converted from the voice data into text as the recognition result.
[0075] (Appendix 7) The information processing device described in Appendix 7 is the information processing device described in any one of Appendices 1 to 5, wherein the recognition target is text data, and the output means outputs data obtained by translating the text data into another language as the recognition result.
[0076] (Appendix 8) The information processing device described in Appendix 8 is the information processing device described in any one of Appendices 1 to 5, wherein the recognition target is image data, and the output means outputs data in which characters contained in the image data are converted into text as the recognition result.
[0077] (Supplementary Note 9) The information processing method described in Supplementary Note 9 is an information processing method that converts, by at least one computer, a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, recognizes the recognition target using the vector sequence of the fixed token length, and outputs a string that is the recognition result.
[0078] (Supplementary Note 10) The recording medium described in Supplementary Note 10 is a recording medium having recorded thereon a computer program for causing at least one computer to execute an information processing method, which converts related strings, which are strings related to a recognition target, or related sentence vectors, which are sentence vectors extracted from the related strings, into vector sequences of a fixed token length, recognizes the recognition target using the vector sequences of the fixed token length, and outputs a string that is the recognition result.
[0079] (Supplementary Note 11) The computer program described in Supplementary Note 11 is a computer program that causes at least one computer to execute an information processing method that converts a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, recognizes the recognition target using the vector sequence of the fixed token length, and outputs a string that is the recognition result.
[0080] This disclosure may be modified as appropriate within the scope that does not contradict the gist or idea of the invention that can be read from the claims and the entire specification, and information processing devices, information processing methods, and recording media that involve such modifications are also included in the technical idea of this disclosure.
[0081] REFERENCE SIGNS LIST 10 Information processing device 11 Processor 12 RAM 13 ROM 14 Storage device 15 Input device 16 Output device 110 Conversion unit 115 Embedding layer 120 Recognition result output unit 200 Sentence vector extraction unit
Claims
1. An information processing device comprising: a conversion means for converting a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length; and an output means for recognizing the recognition target using the vector sequence of the fixed token length, and outputting a string that is the recognition result.
2. The information processing device according to claim 1, wherein the output means recognizes and outputs the recognition target character by character, and receives as input a combined vector sequence that combines the vector sequence of the fixed token length and a vector sequence extracted from the character string that is the previous recognition result, and outputs the next character that is the recognition result.
3. The information processing apparatus according to claim 1 or 2, further comprising an extraction means for extracting the related sentence vector from the related character string.
4. The information processing device according to claim 1 or 2, wherein the related character string is a character string including a sentence related to the content of the recognition target.
5. The information processing device according to claim 1 or 2, wherein the related character string is a character string indicating at least one of the type of language of the recognition target, the sender of the recognition target, and the recording environment of the recognition target.
6. The information processing device according to claim 1 or 2, wherein the recognition target is voice data, and the output means outputs data obtained by converting the voice data into text as the recognition result.
7. The information processing device according to claim 1 or 2, wherein the recognition target is text data, and the output means outputs data obtained by translating the text data into another language as the recognition result.
8. The information processing device according to claim 1 or 2, wherein the recognition target is image data, and the output means outputs data in which characters included in the image data are converted into text as the recognition result.
9. An information processing method comprising: converting, by at least one computer, a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length; recognizing the recognition target using the vector sequence of the fixed token length; and outputting a string that is the recognition result.
10. A recording medium having recorded thereon a computer program for causing at least one computer to execute an information processing method, which comprises converting a related string, which is a string related to a recognition target, or a related sentence vector, which is a sentence vector extracted from the related string, into a vector sequence of a fixed token length, recognizing the recognition target using the vector sequence of the fixed token length, and outputting a string that is the recognition result.