JP4131586B2 - Voice recognition device - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a speech recognition device that converts an input speech signal into character information, and particularly suitable for speech input of a numeric string. In Related.
[0002]
[Prior art]
Conventionally, in a speech recognition device, when a digit string is input with a digit such as “thousand” or “ten thousand”, characters up to the end of recognition are displayed as Chinese numerals in order or input The method of displaying on the screen collectively after the end of was taken.
[0003]
[Problems to be solved by the invention]
However, in the above-described prior art, even in the case of input to a form or the like that should be originally input with Arabic numerals, the input is performed with Chinese numerals. For this reason, there is a sense of incongruity, and a troublesome operation such as correcting the Chinese numeral notation to Arabic numerals later is necessary.
[0004]
Also, the method of displaying the numbers collectively after the completion of the number input has a problem that the Arabic notation of the numbers cannot be confirmed in real time during the input.
It should be noted that the numbers are read out in order from the beginning, for example, “1”, “0”, “0”, “0”, etc. In this way, Arabic numerals can be obtained in real time, but if the number of digits in the character string is large, there is a possibility that it will be input incorrectly.
[0005]
For example, when a unit prefix (a prefix attached to a unit) such as “kilo” is added and a number is input by voice, a conventional speech recognition apparatus can only display a word such as “kilo” as a recognition result. It was not possible to confirm the recognition result with the numerical value.
[0006]
The present invention has been made in view of the above points, and when recognizing a numeric string input by voice such as “thousand” or “ten thousand”, the recognition result is displayed in real time in Arabic numerals. Voice recognition device The The purpose is to provide.
[0008]
A speech recognition apparatus according to the present invention includes a number input unit that inputs a number string consisting of a plurality of digits in order from the beginning, and converts a voice input by the input unit into a voice signal. Speech recognition means for recognizing a reading of a word indicating a word and a reading of a word indicating a position; a storage means storing a reading of a word indicating a numerical value and a reading of a word corresponding to the numerical value and the position; In the case of reading the word recognized by the voice recognition means is a word reading indicating a numerical value, the storage means is referred to and digitized so as to be right-justified. If there is a numerical value before that, numerical expression means for converting the numerical value of the reading of the word indicating the position with reference to the storage means and determining the numerical value by arithmetic processing with the previously numerical value And the numerical expression means Display means for sequentially displaying numerical values numerically expressed in Arabic numerals, and distinguishing and displaying a portion of the numerical value whose position is fixed and a non-fixed portion displayed right-justified It is characterized by comprising.
[0009]
According to such a configuration, the recognition result of a numeric string input by voice can be confirmed at all times, and by using Arabic numerals, it can be displayed without a sense of incongruity even in places where Chinese numerals such as those in the table are inappropriate. Is possible.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Embodiments of the present invention will be described below with reference to the drawings.
(First embodiment)
FIG. 1 is a block diagram showing the configuration of the speech recognition apparatus according to the first embodiment of the present invention. The apparatus is realized by a computer that reads a program recorded on a recording medium such as a magnetic disk and whose operation is controlled by the program.
[0013]
In FIG. 1, an input device 101 is a microphone or the like that converts an audio signal uttered by an operator into an electrical audio signal. A voice signal obtained by the input device 101 is converted into a character code by the voice recognition unit 102 as a corresponding word. At this time, the voice recognition unit 102 refers to the recognition dictionary 103 describing the correspondence between the voice signal and the character code of the word.
[0014]
The numerical expression control unit 104 extracts a numerical expression indicating the rank of “Man (ten)”, “Hyaku (hundred)”, etc. from the word string output from the speech recognition unit 102. At this time, the numerical value expression control unit 104 refers to the numerical value expression table 105 in which “100000000” is stored if the value, for example, “Put (Billion)”, is read together with the reading and notation of the numerical value / position.
[0015]
Next, the numerical expression processing unit 104 checks a series of numerical expression words in order from the oldest input time, and stores the numerical values in the pre-determination buffer A 107 if the word represents a number from 0 to 9. If the word represents the place of “thousand”, “hundred”, or “ten”, the product of the value stored in the pre-determination buffer A 107 and the digitized value of each place in the numerical operation unit 106 is calculated. The result is stored in the pre-determination buffer B108 and the contents of the pre-determination buffer A107 are initialized.
[0016]
Here, when the place immediately before is not a numerical value, that is, when the pre-determination buffer A 107 is empty, the above calculation is performed after substituting 1.
Further, if the word represents the place of “trillion”, “100 million”, and “ten thousand”, the product of the sum of the values of the pre-determination buffer A 107 and the pre-determination buffer B 108 and the numerical value of each position is taken, The result is stored in the confirmed buffer 109 and the contents of the pre-confirmation buffer A 107 and the pre-confirmation buffer B 108 are initialized.
[0017]
Each time one word is processed, the sum of the values of the pre-confirmation buffer A 107, the pre-confirmation buffer B 108, and the confirmed buffer 109 is stored in the display buffer 111. At this time, the display attribute addition unit 110 assigns a character attribute (for example, emphasized character) that has been confirmed to the number string obtained from the confirmed buffer 109, and the number string in the pre-confirmation buffer A107 and the pre-confirmation buffer B108. Is assigned a character attribute (for example, underlined) indicating that it is not confirmed.
[0018]
The display device 112 displays the confirmed part and the unconfirmed part of the numeric string obtained as a recognition result on the screen according to the information in the display buffer 111 based on the display attributes.
[0019]
Next, the operation of the first embodiment will be described.
2 and 3 are flowcharts showing the flow of processing when inputting a numeric string in the first embodiment of the present invention.
[0020]
The voice signal input by the operator through the input device 101 is stored in the recognition buffer 113 (steps 201 and 202). If the voice input is completed (Yes in Step 203), the process here ends. Otherwise (no in step 203), the speech signal stored in the recognition buffer 113 is given to the speech recognition unit 102 and converted to a word with reference to the recognition dictionary 103 (step 204).
[0021]
Here, when the word is a numerical expression (yes in step 205), the numerical expression control unit 104 checks the word immediately before the word, and when it is other than the numerical expression (yes in step 206). , Register K1, pre-confirmation buffer A107, pre-confirmation buffer B108, and confirmed buffer 109 are initialized (step 207). The numerical expression is a word described in the numerical expression table 105, and includes a word representing a place such as “ten”, “hundred”, “thousand”, “ten thousand” in addition to the numbers 0 to 9. It is.
[0022]
Next, it is checked whether the input numerical expression is a numeral or whether it is a digit such as “ten”, “hundred”, “thousand”, “ten thousand” (step 208). If it is a number (yes in step 208), it is checked whether it was a number immediately before (step 219). If it is not a number such as a digit (no in step 219), the number is stored in the buffer before confirmation. Store in A107 (step 220).
[0023]
On the other hand, if it is determined in step 208 above that the numerical expression is other than a number, that is, a digit (no in step 208), the value obtained by digitizing the word in the register K0, that is, for example, “100”. A numerical value such as “100” is stored (step 209).
[0024]
Next, the numerical display control unit 104 checks whether the value of the register K0 is any one of “10”, “100”, “1000”, or any other value (step 210).
[0025]
As a result, when the register K0 is any one of “10”, “100”, and “1000”, it is further checked whether a value stored in a register K1 described later is “10000” or less (step S1). 211). If the register K1 is not set or is greater than “10000” (no in step 211), the data stored in the buffer A107 before determination is examined (step 212), and the buffer A107 before determination is empty. (No in step 212), 1 is substituted into the pre-determination buffer A107 (step 213).
[0026]
The numerical operation unit 106 calculates the product of the pre-determination buffer A 107 and the register K0, and adds the result to the value stored in the pre-determination buffer B 108 (step 214). When the calculation results are stored, the pre-determination buffer A107 is initialized for the next calculation (step 215).
[0027]
On the other hand, if the determination in step 210 is false or the determination in step 211 is true, the value of the register K0 is substituted for the register K1 (step 216), and the pre-determination buffer A107 and the pre-determination buffer The product of the sum of B108 and the register K1 is added to the value stored in the confirmed buffer 109 (step 217). Further, the pre-confirmation buffer A 107 and the pre-confirmation buffer B 108 are initialized (step 218).
[0028]
After step 215, step 218, or step 220, the numerical expression control unit 104 calculates the sum of all the values stored in the pre-confirmation buffer A107, the pre-confirmation buffer B108, and the confirmed buffer 109 in the numerical operation unit 106. Then, it is stored in the display buffer 111 (step 227).
[0029]
At the same time, the display attribute adding unit 110 checks whether a value is stored in the register K1 (step 228). As a result, if it has been stored (no in step 228), a display attribute indicating that it has been determined is assigned to the character of the digit greater than or equal to the register K1 in the display buffer 111 (step 229). An unconfirmed display attribute is assigned to the character that is not given (step 230).
[0030]
The display device 112 displays the numeric string stored in the display buffer 111 on the screen according to the display attribute assigned to each character (step 234). In other words, a portion where the rank is determined based on the display attribute and a portion where the position is not fixed are displayed separately.
[0031]
If it is determined in step 205 that the numerical value input has been completed (step 231: yes), the numerical operation unit 106 calculates the sum of the values stored in the pre-confirmation buffer A 107, the pre-confirmation buffer B 108, and the confirmed buffer 109. And the result is stored in the display buffer 111 (step 232). At the same time, a confirmed display attribute is assigned to the character string in the display buffer 111 (step 233). Information in the display buffer 111 is output from the display device 112 (step 234).
[0032]
On the other hand, if it is determined in step 219 that the numbers are consecutive, for example, if a numerical value after the decimal point is input, the pre-determination buffer A 107 is checked (step 221). As a result, when a numerical value is stored in the pre-confirmation buffer A107 (yes in step 211), the confirmed buffer 109 is shifted one digit to the left and the value of the pre-confirmation buffer A107 is added (step 222). The pre-determination buffer A107 is initialized (step 223).
[0033]
After completion of initialization of the pre-determined buffer A107, and when the pre-determined buffer A107 is empty in step 221, as described above, the finalized buffer 109 is shifted to the left by one digit, and the input number is added (step 224). The value of the confirmed buffer 109 is stored in the display buffer 111 (step 225), the confirmed display attribute is given to the value (step 226), and the value is displayed on the display device 112 (step 234).
[0034]
FIG. 4 is a configuration example of the numerical expression table 105.
Here, readings and notations relating to numerically expressed words are stored, and it is possible to find a number when the recognition result output from the speech recognition unit 102 is either reading or notation. In this case, there are three types of notation relating to numerals: Arabic numerals, Chinese numerals, and large letters. For example, the reading “1” is associated with the Arabic numeral “1”, the Chinese numeral “1”, and the large letter “壱”. This notation is used for the display buffer 111.
[0035]
In the numerical expression control unit 104, the numerical values described in the numerical expression table 105 are used for each buffer (pre-confirmation buffer A107, pre-confirmation buffer B108, and confirmed buffer 109). In other words, this numerical value is used when performing the calculation.
[0036]
FIG. 5 shows an example of data storage in the pre-confirmation buffer A 107, the pre-confirmation buffer B 108, the pre-confirmation buffer 109, and the display buffer 111 when a numerical value is input by voice. The underline character of the display buffer 111 indicates an undetermined character string, and the character without the underline indicates that it has been confirmed, and the same display is performed on the display device 112.
[0037]
Now, the flow of data in each buffer will be described by taking as an example a case where a numerical example consisting of a plurality of digits such as “520304” is input with a rank.
(1) First, a word indicating a number such as “go” is input by voice. In this case, “go” alone may be the next place, such as “thousand” or “ten thousand”, and the numerical value cannot be determined. Therefore, when “go” is recognized, the numerical value “5” is stored in the pre-determined buffer A107, and the display buffer 111 stores the underlined notation “5” (here, Arabic numerals are the first notation). To do. Thereby, the notation “5” is displayed on the display device 112 in an unconfirmed state.
[0038]
(2) Next, the word “ju” indicating the place is input by voice. Even at this stage, I don't know if it is simply "50" or if it is followed by "ten thousand", "100 million", or "trillion". Therefore, the numerical value “50”, which is the product of “5” input last time and “10” input this time, is stored in the pre-determination buffer B108.
[0039]
Further, the notation “50” is stored in the display buffer 111 with an underline, and the notation is displayed on the display device 112.
(3) Next, even when a word indicating a number such as “ni” is inputted by voice, it cannot be confirmed yet. Therefore, when “ni” is recognized, the numerical value “2” is stored in the pre-determination buffer A107. Further, the notation “52” is stored in the display buffer 111 with an underline, and the notation “52” is displayed on the display device 112.
[0040]
(4) Next, the word “Man” indicating the rank is input by voice. Here, with the unit “10,000” as a boundary, the unit of the numerical value can be determined when more units, that is, “10,000”, “Billion”, “Trillion”,. On the other hand, for “ten”, “hundred”, and “thousand”, there is a possibility that “ten thousand”, “billion”, “trillion”, etc. may come after that, and therefore the place of the numerical value cannot be determined.
[0041]
Therefore, when “Man” is input, the numerical value can be confirmed to the place of “10,000”, and the numerical value “520000” is stored in the confirmed buffer 109. Further, the notation “520000” is stored in the display buffer 111 and the lower four digits are underlined and displayed on the display device 112.
[0042]
(5) Next, a word indicating a number such as “san” is input by voice. In this case, since there is a possibility that digits such as “thousand”, “hundred”, and “ten” may enter after that, the numerical value cannot be determined. Accordingly, the numerical value “3” corresponding to “san” is stored in the pre-determination buffer A 107, the notation “520003” is stored in the display buffer 111, and the lower four digits are underlined, and this is displayed on the display device. 112.
[0043]
(6) Next, the word “Baku” indicating the position is input by voice. In this case, since the place of “10,000” has already been decided, the place of “hundred” can be decided. Therefore, “3” input immediately before is calculated as “300”, and “520300” is stored in the confirmed buffer 109.
[0044]
Further, the notation “520300” is stored in the display buffer 111, and the lower two digits are underlined and displayed on the display device 112.
(7) Next, a word indicating a number such as “Yon” is input by voice. In this case, the number cannot be determined because there is a possibility that the place of “ten” will be entered after that. Therefore, the numerical value “4” corresponding to “Yon” is stored in the pre-determination buffer A 107, the notation “520304” is stored in the display buffer 111, and the lower one digit is underlined, and this is displayed on the display device. 112.
[0045]
In this way, it is possible to sequentially recognize a numerical example such as “520304” and display the recognition result in real time in Arabic numerals. Note that the last “4” may be displayed with an underline because there is a possibility that the place of “ten” will be entered.
[0046]
Similarly, even when a numerical example such as “7000080010500” is input with a rank, the recognition result can be displayed in real time in Arabic numerals using a numerical value obtained by internal calculation. In this case, the position of “trillion” can be determined when “Cho” is spoken after “Nana”.
[0047]
Further, it is possible to cope with a case where a numerical example including a decimal point such as “80503.64” is input with a rank. In this case, since the numbers after the decimal point are read out like “Roku” and “Yon”, they can be displayed as Arabic numerals as they are.
[0048]
Note that if the position is entered incorrectly, such as “1”, “put”, “cho”, etc., the recognition of the character string is delimited by the error portion. That is, in this example, it is divided into “Ichiku” and “Cho”, and “Ichiku” is displayed as “100000000”, but “Cho” is, for example, “Book” or “Ding”. , Will be recognized as another word.
[0049]
(Second Embodiment)
Next, a second embodiment of the present invention will be described.
FIG. 6 is a block diagram showing a configuration of a speech recognition apparatus according to the second embodiment of the present invention. The apparatus is realized by a computer that reads a program recorded on a recording medium such as a magnetic disk and whose operation is controlled by the program.
[0050]
Since the input device 501 to the recognition buffer 513 shown in FIG. 6 correspond to the input device 101 to the recognition buffer 113 of FIG. 1, the description thereof is omitted here.
[0051]
In the present embodiment, the numerical expression control unit 504 considers the words stored in the unit prefix table 514 as a part of the numerical expression in addition to numerical expressions such as numbers and digits. The unit prefix table 514 has the same structure as the numerical expression table 505, and stores “kilo”, “milli”, and the like here.
[0052]
When a word in the numerical expression table 505 is input, the pre-confirmation buffer A 507 and the pre-confirmation buffer B 508 are added to the value of the confirmed buffer 509 and stored in the display buffer 511. In addition, the unit prefix table 514 is referred to at the end of the display buffer 511, the notation of the input unit prefix is added, and the display attribute addition unit 510 has confirmed the display of all characters in the display buffer 511. Give attributes.
[0053]
Further, the product of the value of the confirmed buffer 509 and the value obtained by digitizing the unit prefix is stored in the confirmed buffer 509 and the rewrite buffer 515. When the operator issues a rewrite instruction, the value stored in the rewrite buffer 515 by the notation selection unit 516 (that is, the value obtained by expressing the unit prefix in numerical form) can be used as the input result.
[0054]
Next, the operation of the second embodiment will be described.
7 to 9 are flowcharts showing the flow of processing when inputting a numeric string including a unit prefix in the second embodiment of the present invention. Note that Steps 601 to 605 in the figure correspond to Steps 201 to 205 in the flowcharts (first embodiment) shown in FIGS. 2 and 3, and the input voice signal is recognized, and the character code Is output as a word consisting of
[0055]
Here, it is determined whether or not the word is a numerical expression. In addition to the numbers and positions described in the numerical expression table 505, unit prefixes such as “kilo” described in the unit prefix table 514 are also expressed in numerical expressions. Suppose that there is (step 605).
[0056]
If it is not a numerical expression, the process proceeds to step 631. Steps 631 to 633 correspond to steps 231 to 233 in the flowcharts shown in FIGS.
[0057]
If it is a numerical expression (Yes in Step 605), the word immediately before it is checked (Step 606). If the word is a unit prefix or other than a numerical expression (yes in step 606), the register K1, the pre-confirmation buffer A507, the pre-confirmation buffer B508, and the finalized buffer 509 are initialized (step). 607).
[0058]
Next, it is checked whether or not the input word is a number (step 608). The processing from step 619 to step 626 is the same as that from step 219 to step 226 in the flowcharts shown in FIGS.
[0059]
If it is not a number, it is checked whether the word is a unit prefix (step 635). If it is not a unit prefix (No in Step 635), Step 609 to Step 618 (the same processing as Step 209 to Step 218 in the flowchart shown in FIGS. 2 and 3), Step 627 to Step 630 (FIG. 2 and FIG. 3). The process proceeds from step 227 in the flowchart shown in FIG.
[0060]
If it is a unit prefix (yes in step 635), the magnification of the word obtained by referring to the unit prefix table 514 in the register K0, for example, “1000” in the case of “kilo”, and “mm” Is stored as “0.001” (step 636).
[0061]
Next, the sum of the pre-determined buffer A 507, the confirmed buffer B 508, and the confirmed buffer 509 is stored in the confirmed buffer 509 and the display buffer 511 (step 637). A unit prefix notation, for example, “k”, is added to the display buffer 511 (step 638), and a confirmed display attribute is given (step 639). Further, the product of the confirmed buffer 509 and the register K0 is stored in the rewrite buffer 515 (step 640).
[0062]
The contents of the display buffer 511 are displayed on the screen of the display device 512 (step 634), while the rewrite buffer 515 is replaced with the recognition result when an operator rewrite instruction is given.
[0063]
FIG. 10 is a configuration example of the unit prefix table 514.
Although the configuration is almost the same as the numerical expression table 105 shown in FIG. 4, “10 (ten)” has a numerical meaning in the word itself, whereas the unit prefix is in the word itself. Since it has no numerical meaning, it is expressed here in terms of magnification.
[0064]
That is, the unit prefix notation such as “nano” is “n”, but when expressing numerically including this, the magnification “0.000000001” is used. Similarly, in the numerical expression of “micro”-“μ”, the magnification “0.000001” is used, and in the numerical expression of “milli”-“m”, the magnification “0.001” is used.
[0065]
Also, in the numerical expression of “kilo”-“k”, the magnification “1000” is used, and in the numerical expression of “mega”-“M”, the magnification “1000000” is used, and “giga”-“G”. In the numerical expression, the magnification “1000000000” is used.
[0066]
FIG. 11 shows an example of data storage in the pre-confirmation buffer A507, the pre-confirmation buffer B508, the preconfirmation buffer 509, the display buffer 511, and the rewrite buffer 515 when a numerical value including a unit prefix is input. The underline character of the display buffer 511 indicates an undetermined character string, and the character without the underline indicates that it has been confirmed, and the same display is performed on the display device 512.
[0067]
Now, the flow of data in each buffer will be described by taking as an example a case where a voice is input with a unit prefix with an example of a number consisting of a plurality of digits such as “380000k”.
[0068]
(1) First, a word indicating a number such as “san” is input by voice. In this case, “san” alone may be the next place in the order of “thousand”, “ten thousand”, etc., and the numerical value cannot be determined. Therefore, the numerical value “3” corresponding to “san” is stored in the pre-determination buffer A 507 as the recognition result, and the display buffer 511 is underlined and read as “3” (here, the Arabic numeral 1). Thereby, the notation “3” is displayed on the display device 512 in an unconfirmed state.
[0069]
(2) Next, the word “ju” indicating the place is input by voice. Even at this stage, I don't know if it's just "30", or if it is followed by "ten thousand", "100 million" or "trillion". Therefore, the numerical value “30”, which is the product of “3” input last time and “10” input this time, is stored in the pre-determination buffer B508.
[0070]
(3) Next, even when a word indicating a number such as “Hachi” is inputted by voice, since it cannot be confirmed yet, the numerical value “8” corresponding to the “Hachi” is stored in the pre-determination buffer A507. Further, the notation “38” is stored in the display buffer 511 with an underline, and the notation “38” is displayed on the display device 512.
[0071]
(4) Next, the word “Man” indicating the rank is input by voice. Here, with the unit “10,000” as a boundary, the unit of the numerical value can be determined when more units, that is, “10,000”, “Billion”, “Trillion”,. On the other hand, for “ten”, “hundred”, and “thousand”, there is a possibility that “ten thousand”, “billion”, “trillion”, etc. may come after that, and therefore the place of the numerical value cannot be determined.
[0072]
Therefore, now, when “Man” is input, the numerical value can be fixed to the place of “10,000”, and the numerical value “380000” is stored in the fixed buffer 509. Further, the notation “3800000” is stored in the display buffer 511 and the lower four digits are underlined and displayed on the display device 512.
[0073]
(5) Next, when a unit prefix such as “Kiro” is input by voice, all the numerical values are fixed at that time, and “k” is added to the display buffer 511 so that “380000k” is represented.
[0074]
When rewriting is instructed here, the magnification “1000” corresponding to “k” is applied, the notation “3800000k” is replaced with “380000000”, and the rewriting buffer 515 is stored. As a result, the recognition result can be obtained with the notation “380000000”.
[0075]
Thus, the operator can always confirm the recognition result by sequentially displaying the numbers inputted by voice on the screen. In addition, by using Arabic numerals, it is possible to display without discomfort even in places where Chinese numerals are inappropriate in the table.
[0076]
Further, when a unit string is added and a numeric string is input by voice, not only a notation using the unit prefix but also a numerical notation can be selected as a rewriting candidate. Therefore, it is possible to input easily without being aware of the number of digits after the decimal point.
[0077]
When inputting numbers with a large number of digits, it was necessary to input many “0” s on the keyboard. However, using this method, it is easy to speak “10,000”, “Billion”, “Trillion”, etc. with voice. Therefore, it is possible to input numerical values in a short time, and in an application that frequently uses numerical inputs such as spreadsheet software, the work time can be shortened.
[0078]
The method described in each embodiment described above is a recording medium such as a magnetic disk (floppy disk, hard disk, etc.), optical disk (CD-ROM, DVD, etc.), semiconductor memory, etc., as a program that can be executed by a computer. Can be applied to various devices, or transmitted by a communication medium and applied to various devices. A computer that implements this apparatus reads the program recorded on the recording medium, and executes the above-described processing by controlling the operation by this program.
[0079]
【The invention's effect】
As described above, according to the present invention, it is possible to sequentially display a voice input numeric string on the screen, and the operator can always check the recognition result. In addition, by using Arabic numerals, it is possible to display a sense of incongruity even in places where Chinese numerals in the table are inappropriate.
[Brief description of the drawings]
FIG. 1 is a block diagram showing a configuration of a speech recognition apparatus according to a first embodiment of the present invention.
FIG. 2 is a flowchart showing a process flow when inputting a numeric string in the first embodiment.
FIG. 3 is a flowchart showing a flow of processing when inputting a numeric string in the first embodiment.
FIG. 4 is a configuration example of a numerical expression table in the first embodiment.
FIG. 5 shows an example of data storage in each buffer when a numerical value is inputted by voice in the first embodiment.
FIG. 6 is a block diagram showing a configuration of a speech recognition apparatus according to a second embodiment of the present invention.
FIG. 7 is a flowchart showing a processing flow when inputting a numeric string including a unit prefix in the second embodiment.
FIG. 8 is a flowchart showing a processing flow when inputting a numeric string including a unit prefix in the second embodiment.
FIG. 9 is a flowchart showing a processing flow when inputting a numeric string including a unit prefix in the second embodiment.
FIG. 10 shows a configuration example of a unit prefix table in the second embodiment.
FIG. 11 shows an example of data storage in each buffer when a numerical value including a unit prefix in the second embodiment is inputted by voice.
[Explanation of symbols]
101 ... Input device
102 ... voice recognition unit
103 ... Recognition dictionary
104: Numerical expression control unit
105 ... Numerical expression table
106: Numerical calculation unit
107: Buffer A before confirmation
108 ... Buffer B before confirmation
109 ... Final buffer
110 ... display attribute addition part
111 ... Display buffer
112 ... Display device

Claims (1)

An input means for inputting a voice string in order from the top of a digit string consisting of multiple digits,
The voice inputted by said input means into a speech signal, the reading of a word indicating the number from the speech signal, a speech recognition means for recognizing processes the readings of the word indicating the position,
Storage means for word reading is stored which indicates the values and positions of the read word corresponding to the reading of numerical,
Wherein when the reading of a word indicating the numerical value read word recognized processed by the speech recognition means, to quantify such that right-aligned with reference to said storage means, for reading the word indicating the position, If there is a numerical value before that, numerical expression means for digitizing the reading of the word indicating the position with reference to the storage means and determining the numerical value by performing arithmetic processing with the previously numerical value When,
The numerical values digitized by the numerical expression means are sequentially displayed as Arabic numerals, and the portion of the numerical value whose position is fixed and the portion of the numerical value that is not fixed are displayed separately. A speech recognition apparatus comprising: a display unit;

Images