JP2009075263A - Voice recognition device and computer program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce the frequency of editing operations of a candidate word by a user when the user edits the candidate word displayed on a screen while displaying the candidate word from a voice recognition result of an input voice on the screen. <P>SOLUTION: The device includes a voice recognition part 13 for performing processing to recognize an input voice and generating a recognition result composed of a string of recognized words; a candidate word generation part 16 for generating a candidate word from the recognition result; a candidate word editing-display part 17 for displaying the candidate word on a screen and updating the recognition result according to an editing content by the user; and an editing operation part 18 for making the user edit the candidate word displayed on the screen. The device has a word connection rule for connecting a plurality of continuous words. The candidate word generation part 16 connects a plurality of continuous words contained in the recognition result according to the word connection rule, and takes the resulting connected word string as the candidate word. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a speech recognition apparatus and a computer program.

Conventionally, in speech recognition using a computer, it is difficult to achieve a recognition rate of 100% due to the influence of a speaker's utterance method, background noise at the time of speech input, and the like. Therefore, for example, the speech recognition apparatus described in Patent Literature 1 compares a plurality of words included in the input speech with a plurality of words stored in the dictionary in advance, and has the highest competition probability among the competition candidates. A word is used as a speech recognition result, and the speech recognition result is displayed on the screen as a word string of a plurality of words. Select and display on the screen adjacent to the word with the highest probability of competition, and select and select an appropriate correction word from one or more competing words displayed on the screen according to the manual operation by the user The competing word is replaced with the word having the highest competition probability in the speech recognition result.
JP 2006-146008 A

  However, in the conventional speech recognition apparatus described above, the candidate words displayed on the screen from the speech recognition result are in units of words (see FIGS. 2, 19, and 20 of Patent Document 1). So, for example, if the sentence “It is a meeting at 3:00 pm today” is entered by voice, the speech recognition result is a particle such as “no” or “ni”, or a prefix such as “go”. , Only candidate words will be displayed. Therefore, the user edits the text on the screen by selecting the correct answer from the candidate words displayed on the screen, deleting the candidate word, changing the candidate word to a new candidate word, etc. In this case, it is necessary to make correct words that are meaningless by itself, such as particles and prefixes, or to perform operations such as deletion and change, which takes time and effort. In particular, when speech recognition of a long sentence such as e-mail is performed, the number of operations for editing a candidate word increases, and user convenience is poor.

  The present invention has been made in consideration of such circumstances, and its purpose is to display candidate words on the screen from the speech recognition result of the input speech and when the user edits the candidate words displayed on the screen. Another object of the present invention is to provide a speech recognition apparatus and a computer program that can reduce the number of operations of a user editing a candidate word.

  In order to solve the above problems, a speech recognition apparatus according to the present invention performs processing for recognizing input speech and generates a recognition result including a recognized word sequence, and a candidate based on the recognition result. A candidate word generating means for generating a word, a candidate word displaying means for displaying the candidate word on the screen, an editing operation means for the user to edit the candidate word displayed on the screen, and the recognition result according to the editing content by the user. Updating means for updating, and providing a word linking rule when linking a plurality of consecutive words, wherein the candidate word generating means links a plurality of consecutive words included in the recognition result according to the word linking rule The connected word string is used as a candidate word.

  In the speech recognition apparatus according to the present invention, the candidate word generating means determines whether it is possible to connect a plurality of consecutive words included in the recognition result according to the word connection rule, And word linking means for linking a plurality of consecutive words included in the recognition result according to the determination result.

  In the speech recognition apparatus according to the present invention, the word connection rule defines a combination of a plurality of connectable words and an order of the combined words.

  In the speech recognition apparatus according to the present invention, the speech recognition apparatus includes language probability storage means for storing a language probability for a combination of a plurality of words connectable according to the word connection rule and an order of the combined words, The combination of a plurality of words connectable according to the word connection rule and the order of the combined words are determined to be connectable when the language probability is equal to or higher than a threshold value. And

  The speech recognition apparatus according to the present invention is characterized by comprising a threshold value specifying means for the user to specify the threshold value.

  In the speech recognition apparatus according to the present invention, the word linking rule defines a combination of part-of-speech types of a plurality of connectable words and an order of the combined part-of-speech types.

  The speech recognition apparatus according to the present invention includes a part-of-speech connection cost storage unit that stores a combination of a plurality of part-of-speech types that can be connected by the word connection rule and a part-of-speech connection cost about the order of the combined part-of-speech types. The word linking determination means includes a plurality of corresponding parts of speech when the part of speech linking cost is equal to or lower than a threshold for a combination of part of speech types of a plurality of words connectable according to the word linking rule and an order of the combined part of speech types. It is determined that the words can be connected.

  The speech recognition apparatus according to the present invention is characterized by comprising a threshold value specifying means for the user to specify the threshold value.

The computer program according to the present invention performs a process of recognizing input speech, generates a recognition result including a recognized word sequence, a candidate word generation function that generates a candidate word from the recognition result, Candidate word display function for displaying candidate words on the screen, editing operation function for the user to edit the candidate words displayed on the screen, and update function for updating the recognition result according to the editing contents by the user are realized on the computer A word linking rule for linking a plurality of consecutive words, wherein the candidate word generation function connects a plurality of consecutive words included in the recognition result according to the word linking rule It is characterized in that the word string is a candidate word.
As a result, the speech recognition apparatus described above can be realized using a computer.

  According to the present invention, not only a word unit but also a phrase unit, a phrase unit, and a sentence unit candidate word are created, so that the user can select not only a word unit but also a phrase unit, a phrase unit, and a sentence unit candidate word. Can be edited. Thereby, the effect that the frequency | count of operation which a user edits a candidate word can be reduced is acquired.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing the overall configuration of a speech recognition apparatus 1 according to an embodiment of the present invention. In FIG. 1, a speech recognition apparatus 1 includes a speech input unit 11, an acoustic feature amount extraction unit 12, a speech recognition unit 13, an acoustic model storage unit 14, a language model storage unit 15, a candidate word generation unit 16, a candidate word edit / display. A unit 17 and an editing operation unit 18 are provided.

  The voice input unit 11 includes a microphone, an amplifier, an analog-digital converter (AD converter), and the like. The voice input unit 11 inputs the voice uttered by the user through a microphone, amplifies the input analog voice signal to an appropriate level, and converts it into digital voice data. This audio data is sent to the acoustic feature quantity extraction unit 12.

  The voice input unit 11 includes a communication interface connected to a communication line such as a telephone line or an IP (Internet Protocol) network, and sends digital voice data received via the communication line to the acoustic feature quantity extraction unit 12. It may be. Further, when the audio data is encoded, the decoded audio data is sent to the acoustic feature quantity extraction unit 12.

  The acoustic feature quantity extraction unit 12 extracts the acoustic feature quantity used for the subsequent speech recognition processing from the voice data received from the voice input unit 11. The acoustic feature amount data is sent to the speech recognition unit 13.

  The speech recognition unit 13 performs speech recognition processing on the acoustic feature amount data received from the acoustic feature amount extraction unit 12. For this speech recognition processing, an acoustic model stored in the acoustic model storage unit 14 and a language model stored in the language model storage unit 15 are used. The acoustic model and the language model are constructed by learning using learning data in advance as a preparation stage, and stored in the storage units 14 and 15.

  The speech recognition unit 13 recognizes a word from the acoustic feature data by speech recognition processing using an acoustic model and a language model, and creates a recognition result including the recognized word sequence. At this time, not only the recognition result including the most probable word string but also other recognized word strings are generated as the recognition result. For each recognition result, the speech recognition unit 13 calculates the certainty (reliability) of the recognition result from an acoustic score (acoustic likelihood) and a linguistic probability (language probability). The language probability is the probability that a certain number (for example, three) of word sequences will appear. The speech recognition unit 13 creates a recognition result in the form of a network of words using a recognition result having reliability up to a predetermined rank from the created recognition results.

  FIG. 2 shows a configuration example of the recognition result of the word network format. The network format of words as shown in FIG. 2 is conventionally called a lattice format. The example of FIG. 2 is a configuration example when the user reads the sentence “It is a meeting at 3 pm today”. As shown in FIG. 2, a plurality of recognition results (recognition results up to a predetermined rank) are used, and temporally corresponding word breaks included in each recognition result are connected in a network form. . The contents of the recognition result in FIG. 2 are for convenience of explanation.

  The speech recognition unit 13 includes part-of-speech information indicating the type of part-of-speech of each word, and information on acoustic likelihood, language probability, and reliability of each word in the recognition result of the word network format. The recognition result of the word network format is sent to the candidate word generation unit 16.

  The candidate word generation unit 16 generates a candidate word from the recognition result of the word network format received from the speech recognition unit 13. The candidate word generation unit 16 creates candidate words in phrase units, phrase units, and sentence units by concatenating a plurality of consecutive words included in the recognition result of the word network format. The candidate word generation unit 16 connects a plurality of consecutive words included in the recognition result of the word network format according to the word connection rule. The word concatenation rule is a rule for concatenating a plurality of consecutive words. The word connection rule is referred to when determining whether a plurality of continuous words included in the recognition result of the word network format can be connected. The word linking rule is set in the speech recognition apparatus 1 in advance as a preparation stage.

  The candidate word generation unit 16 directly selects a word that does not satisfy the word connection rule from a plurality of continuous words included in the recognition result of the word network format as a candidate word. The candidate word generation unit 16 outputs candidate word data composed of a sequence of candidate words generated from the recognition result of the word network format to the candidate word editing / display unit 17. Candidate word data is a recognition result represented as a sequence of candidate words.

  The candidate word editing / display unit 17 displays the candidate word data received from the candidate word generation unit 16 on the screen. The editing operation unit 18 includes various editing operation keys. For example, an operation key for the user to select a correct candidate word from candidate words displayed on the screen, an operation key for the user to delete the candidate word, an operation key for the user to input a new candidate word, An operation key for instructing the user to end editing of the recognition result is provided. The editing operation unit 18 notifies the candidate word editing / display unit 17 of the editing content performed by the user using the operation keys. The candidate word editing / display unit 17 updates the recognition result according to the editing content notified from the editing operation unit 18. Then, the display content of the screen is updated with the candidate word data corresponding to the updated recognition result. Thereby, the recognition result reflecting the content edited by the user is displayed on the screen.

  FIG. 3 is a configuration example of the candidate word generation unit 16 shown in FIG. In FIG. 3, the candidate word generation unit 16 includes a word connection unit 21, a word connection determination unit 22, a candidate word grouping unit 31, an unification unit 32 of the same candidate words, a candidate word addition unit 33, and a candidate word group storage unit 34. Have

  The word connection unit 21 requests the word connection determination unit 22 to determine whether a plurality of continuous words included in the recognition result of the word network format can be connected. The word connection determination unit 22 determines whether the plurality of words satisfy the word connection rule, and returns to the word connection unit 21 whether the determination is acceptable (connectable) or unacceptable (not connectable).

  The word connecting unit 21 connects a plurality of consecutive words included in the recognition result according to the determination result. The concatenated word string is a candidate word. On the other hand, a word that is not connected becomes a candidate word as it is. Thereby, the recognition result of the word network format shown in FIG. 2 becomes a candidate word network format as shown in FIG. FIG. 4 is a configuration example of a recognition result after word connection processing. In FIG. 2, the words are connected in a network, but in FIG. 4, the words are connected in a network in candidate word units in which a plurality of words are connected.

  The candidate word grouping unit 31 performs grouping processing of candidate words (words, phrases, phrases) on the recognition result after the word concatenation processing based on reading proximity, time information, and the like. The candidate word grouping unit 31 aligns the start time and end time of the candidate words in the same group with the start time and end time of the candidate word having the maximum reliability. Thereby, the recognition result after the word concatenation process shown in FIG. 4 is simplified as shown in FIG. FIG. 5 is a configuration example of a recognition result after grouping processing of candidate words.

  The unifying unit 32 of the same candidate words collects the same candidate words of the notation included in the same group into one candidate word and recalculates the reliability of the candidate words for the recognition result after the grouping process of candidate words To do. The reliability of the candidate words after the unification process is obtained by the average, addition, maximum value, etc. of the reliability of the candidate words before the unification process. Thereby, the recognition result after the candidate word grouping process shown in FIG. 5 is simplified as shown in FIG. FIG. 6 is a configuration example of a recognition result after the unification processing of the same candidate word. In the example of FIG. 6, the candidate words are arranged from the top in descending order of reliability in each time interval.

  The candidate word adding unit 33 adds the candidate word to the recognition result after the unification processing of the same candidate word based on the history of the group of past candidate words. The candidate word group storage unit 34 stores a history of past candidate word groups. The candidate word adding unit 33 reads, from the candidate word group storage unit 34, the group history of the candidate word having the maximum reliability in the recognition result after the unification processing of the same candidate word. When there is a candidate word that does not exist in the group in the recognition result after the unification processing of the same candidate word in the read history of the group, the candidate word adding unit 33 selects the candidate word as the same candidate word. Is added to the group in the recognition result after the unification processing. Conversely, if a candidate word that exists in the group in the recognition result after the unification processing of the same candidate word does not exist in the group history read from the candidate word group storage unit 34, the candidate word is selected as a candidate word. It adds to the history of the group in the group memory | storage part 34. FIG.

  The candidate word generation unit 16 outputs candidate word data corresponding to the recognition result after the candidate word addition processing to the candidate word editing / display unit 17.

  FIG. 7 is a configuration example of the candidate word editing / display unit 17 shown in FIG. In FIG. 7, the candidate word editing / display unit 17 includes a candidate word data analysis / update unit 41, a candidate word group / candidate word selection history storage unit 42, and a candidate word display unit 43.

  The candidate word data analysis / update unit 41 analyzes the candidate word data and creates and holds a provisional recognition result by connecting candidate words having the maximum reliability in each time interval. The provisional recognition result and the data of candidate words in the same group as each candidate word are sent to the candidate word display unit 43. At this time, only the amount that can be displayed on the screen is sent to the candidate word display unit 43.

  The candidate word display unit 43 displays the recognition result received from the candidate word data analysis / update unit 41 on the screen of the display device. At this time, the boundary of the candidate word is clearly indicated by a blank or the like. Further, if there are other candidate words grouped for each candidate word, this is indicated by an underline or the like. Further, the candidate words of the same group are displayed on a screen different from the screen displaying the recognition result, and the candidate words are displayed in the order of high reliability in the screen.

  When the candidate word data analysis / update unit 41 receives the editing content of the user from the editing operation unit 18, the candidate word data analysis / update unit 41 updates the recognition result according to the editing content. For example, the recognition result is changed according to editing contents such as selection of correct candidate words, deletion of candidate words, change of arrangement of candidate words, and input of new candidate words. When the correct candidate word is selected, the edited portion is replaced with the correct candidate word, and the other candidate words are deleted. If the candidate word is deleted, all the candidate words in the edited portion are deleted. When a new candidate word is input, the input candidate word is inserted at the edit location. The candidate word data analysis / update unit 41 sends the recognition result after editing and data of candidate words in the same group as each candidate word to the candidate word display unit 43.

  When the candidate word data analyzing / updating unit 41 receives an instruction to move the edited portion from the editing operation unit 18, the candidate word data is displayed for the recognition result corresponding to the destination and candidate word data in the same group as each candidate word. Send to part 43.

  The candidate word group / candidate word selection history storage unit 42 holds a group of candidate words and a probability that the user has selected a candidate word (user selection probability). The candidate word data analyzing / updating unit 41 refers to the candidate word group / candidate word selection history storage unit 42 and displays the candidate word of the group of candidate words corresponding to the edited portion, as a candidate word group / candidate word selection history storage unit 42. It is possible to perform processing for rearranging the items in descending order of user selection probability. It should be noted that whether or not to execute the display order changing process based on the user selection probability can be selected.

  Next, the word linking rules according to the present embodiment will be described in detail with some examples.

  FIG. 8 is Example 1 of the word connection rule according to the present invention. In FIG. 8, the word connection rule 101 is configured as a list describing combinations that can be connected to two consecutive words. Furthermore, the order of the two words of the connectable combination is defined as a forward word and a subsequent word. The word connection rule 101 is prepared in advance and stored in a memory in the speech recognition apparatus 1.

  When two consecutive words included in the recognition result match the connectable combination in the word connection rule 101 and the order of the two combined words, the word connection determination unit 22 connects the two words. Judge that it is possible. In the example of FIG. 8, two consecutive words are targeted, but three or more consecutive words may be included.

  FIG. 9 is a second embodiment of the word linking rule according to the present invention. In FIG. 9, the word connection rule 102 is configured as a list describing combinations that can be connected to three consecutive words. Further, the order of the three words of the connectable combinations is defined as the first word, the second word, and the third word. Further, the language probabilities about the combinations of three words that can be connected and the order of the combined words are stored in the list. The word connection rule 102 is prepared in advance and stored in a memory in the speech recognition apparatus 1. As the word connection rule 102, for example, a language model such as N-gram can be used.

  The word connection determination unit 22 determines that the language probability is equal to or greater than a threshold when three consecutive words included in the recognition result match the connectable combination in the word connection rule 102 and the order of the combined three words. If it is, it is determined that the three words can be connected. In the example of FIG. 9, three consecutive words are targeted, but two or four or more consecutive words may be included.

  Note that the language probabilities stored in the list of the word connection rules 102 may be updated according to the recognition result of the word network format created by the speech recognition unit 13.

  The threshold for determining the language probability may be specified by the user. In this case, an operation key is provided for the user to specify a threshold value. If the user decreases the threshold, the probability of generating a long candidate word can be increased. However, even if the language probability is low, they are connected as candidate words, so the probability that the number of incorrect answer candidate words increases. On the other hand, if the user increases the threshold, the probability of generating a long candidate word can be lowered. However, if the language probabilities are low, they are not connected as candidate words, so the probability that the number of correct candidate words will increase. In this way, by allowing the user to specify the threshold value for determining the language probability, the user can adjust the length of the candidate word and the accuracy rate.

  FIG. 10 is a third embodiment of the word linking rule according to the present invention. In FIG. 10, the word connection rule 103 is configured as a list that describes combinations of types of parts of speech that can be connected for two consecutive words. Furthermore, the order of two types of parts of speech that can be connected is defined as a front part of speech and a subsequent part of speech. The word connection rule 103 is prepared in advance and stored in a memory in the speech recognition apparatus 1.

  Based on the part-of-speech information about two consecutive words included in the recognition result, the word connection determination unit 22 combines the two parts of speech in which the parts of speech of the two words are connectable in the word connection rule 103 and the two parts of speech combined. When the order matches, the two words are determined to be connectable. In the example of FIG. 10, the part of speech of two consecutive words is targeted. However, the part of speech of three or more consecutive words may be included.

  FIG. 11 is Example 4 of the word connection rule based on this invention. In FIG. 11, the word connection rule 104 is configured as a list in which combinations of types of parts of speech that can be connected are described for two consecutive words. Furthermore, the order of two types of parts of speech that can be connected is defined as a front part of speech and a subsequent part of speech. Furthermore, the part-of-speech concatenation cost for the combination of two connectable parts of speech and the order of the combined parts of speech is stored in the list. The word connection rule 104 is prepared in advance and stored in a memory in the speech recognition apparatus 1.

  Based on the part-of-speech information about two consecutive words included in the recognition result, the word connection determination unit 22 combines the two parts of speech in which the parts of speech of the two words can be connected in the word connection rule 104. If the part-of-speech concatenation cost is equal to or less than a threshold value, the two words are determined to be concatenable. In the example of FIG. 11, the part of speech of two consecutive words is targeted, but the part of speech of three or more consecutive words may be included.

  The threshold for determining the part-of-speech connection cost may be specified by the user. In this case, an operation key is provided for the user to specify a threshold value. If the user increases the threshold, the probability of generating a long candidate word can be increased. However, even if the part-of-speech connection cost is high, the words are connected as candidate words, so the probability that the number of incorrect answer candidate words increases. On the other hand, the user can reduce the probability of generating a long candidate word by reducing the threshold. However, when the part-of-speech concatenation cost is high, since it is not concatenated as a candidate word, the probability that the number of correct candidate words increases. As described above, by allowing the user to specify the threshold value for determining the part-of-speech connection cost, the user can adjust the length of the candidate word and the correct answer rate.

  As described above, according to the present embodiment, a plurality of consecutive words included in the recognition result are connected according to the word connection rule, and the connected word string is set as a candidate word. Thereby, not only a word unit but a phrase unit, a phrase unit, and a sentence unit candidate word can be created. As a result, the user can edit not only the word unit but also the phrase unit, the phrase unit, and the sentence unit candidate word, so that the user can reduce the number of operations for editing the candidate word. Is obtained.

  Note that the speech recognition apparatus 1 according to the present embodiment may be realized by dedicated hardware, or may be configured by a computer system such as a personal computer to realize each function of the apparatus shown in FIG. The function may be realized by executing a program to do so.

In addition, an input device, a display device, and the like (none of which are shown) are connected to the voice recognition device 1 as peripheral devices. Here, the input device refers to an input device such as a keyboard, a mouse, or a key of a mobile phone terminal. The display device refers to a CRT (Cathode Ray Tube), a liquid crystal display device or the like.
The peripheral device may be connected directly to the speech recognition apparatus 1 or may be connected via a communication line.

Further, by recording a program for realizing the function of the voice recognition device 1 shown in FIG. 1 on a computer-readable recording medium, and causing the computer system to read and execute the program recorded on the recording medium, Processing related to speech recognition may be performed. Here, the “computer system” may include an OS and hardware such as peripheral devices.
Further, the “computer system” includes a homepage providing environment (or display environment) if a WWW system is used.
“Computer-readable recording medium” refers to a flexible disk, a magneto-optical disk, a ROM, a writable nonvolatile memory such as a flash memory, a portable medium such as a DVD (Digital Versatile Disk), and a built-in computer system. A storage device such as a hard disk.

Further, the “computer-readable recording medium” means a volatile memory (for example, DRAM (Dynamic Random Access Memory)), etc., which hold programs for a certain period of time.
The program may be transmitted from a computer system storing the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the “transmission medium” for transmitting the program refers to a medium having a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line.
The program may be for realizing a part of the functions described above. Furthermore, what can implement | achieve the function mentioned above in combination with the program already recorded on the computer system, and what is called a difference file (difference program) may be sufficient.

As mentioned above, although embodiment of this invention was explained in full detail with reference to drawings, the specific structure is not restricted to this embodiment, The design change etc. of the range which does not deviate from the summary of this invention are included.
For example, the voice recognition device 1 described above may be configured in combination with various devices that create documents such as a word processor device and an electronic mail device.

1 is a block diagram showing an overall configuration of a speech recognition apparatus 1 according to an embodiment of the present invention. It is an example of a structure of the recognition result of the network format of a word. It is a structural example of the candidate word production | generation part 16 shown in FIG. It is an example of composition of a recognition result after word connection processing concerning one embodiment of the present invention. It is an example of composition of a recognition result after grouping processing of a candidate word concerning the embodiment. It is an example of composition of a recognition result after unification processing of the same candidate word concerning the embodiment. It is an example of a structure of the candidate word edit / display part 17 shown in FIG. It is Example 1 of the word connection rule which concerns on this invention. It is Example 2 of the word connection rule which concerns on this invention. It is Example 3 of the word connection rule which concerns on this invention. It is Example 4 of the word connection rule which concerns on this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 ... Voice recognition apparatus, 11 ... Voice input part, 12 ... Acoustic feature-value extraction part, 13 ... Voice recognition part, 14 ... Acoustic model memory | storage part, 15 ... Language model memory | storage part, 16 ... Candidate word generation part, 17 ... Candidate Word editing / display unit, 18 ... editing operation unit, 21 ... word connection unit, 22 ... word connection determination unit, 101, 102, 103, 104 ... word connection rule

Claims (9)

Speech recognition means for performing processing for recognizing input speech and generating a recognition result including a recognized word string;
Candidate word generating means for generating candidate words from the recognition result;
Candidate word display means for displaying candidate words on the screen;
Editing operation means for the user to edit the candidate words displayed on the screen;
Updating means for updating the recognition result in accordance with the contents edited by the user,
Establish a word linking rule when linking consecutive words,
The said candidate word production | generation means connects the several continuous word contained in the said recognition result according to the said word connection rule, The speech recognition apparatus characterized by using the connected word string as a candidate word. The candidate word generation means includes:
Word connection determination means for determining whether it is possible to connect a plurality of consecutive words included in the recognition result according to the word connection rule;
Word linking means for linking a plurality of consecutive words included in the recognition result according to the determination result;
The speech recognition apparatus according to claim 1, comprising:   The speech recognition apparatus according to claim 2, wherein the word connection rule defines a combination of a plurality of connectable words and an order of the combined words. Language probability storage means for storing a combination of a plurality of words connectable by the word connection rule and a language probability for the order of the combined words;
The word connection determination means can connect a plurality of corresponding words when the language probability is equal to or greater than a threshold for a combination of a plurality of words connectable according to the word connection rule and an order of the combined words. To determine,
The speech recognition apparatus according to claim 3.   The voice recognition apparatus according to claim 4, further comprising a threshold value specifying means for a user to specify the threshold value.   The speech recognition apparatus according to claim 2, wherein the word connection rule defines a combination of part of speech types of a plurality of connectable words and an order of the combined part of speech types. Part-of-speech connection cost storage means for storing a part-of-speech connection cost for a combination of a plurality of part-of-speech types that can be connected by the word connection rule and an order of the combined part-of-speech types
The word linking determination means includes a plurality of corresponding parts of speech when the part of speech linking cost is equal to or lower than a threshold for a combination of part of speech types of a plurality of words connectable according to the word linking rule and an order of the combined part of speech types. Determines that the words can be connected,
The speech recognition apparatus according to claim 6.   The voice recognition apparatus according to claim 7, further comprising a threshold value specifying unit for a user to specify the threshold value. A speech recognition function that recognizes input speech and generates a recognition result consisting of a sequence of recognized words;
A candidate word generation function for generating candidate words from recognition results;
A candidate word display function for displaying candidate words on the screen;
An editing operation function for the user to edit the candidate words displayed on the screen;
A computer program for causing a computer to implement an update function for updating a recognition result in accordance with editing contents by a user;
Establish a word linking rule when linking consecutive words,
The computer program characterized in that the candidate word generation function connects a plurality of consecutive words included in the recognition result according to the word connection rule, and uses the connected word string as a candidate word.

Images