JP2009230068A - Voice recognition device and navigation system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide speech recognition technology for suppressing degradation in recognition performance by reducing comparison object patterns which are actually used, in a voice recognition by a large amount of comparison object patterns. <P>SOLUTION: In a dictionary section 312, there are a plurality of kinds of dictionaries classified based on a plurality of utterance patterns. By using a similarity degree between user's utterance pattern based on a recognition result, and the plurality of utterance patterns, a dictionary priority determination section 325 determines a priority of the plurality of kinds of dictionaries so that the priority may become higher, as the similarity degree is higher. If voice recognition is performed by preferentially using the dictionary of a high priority, in a voice recognition section 31, a final recognition result is obtained with high possibility, for example, only by using a dictionary of the priority 1. Thereby, appropriate voice recognition is attained with high possibility, only by collating with a fewer comparison object patterns than in a case when collating with all dictionaries. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a speech recognition technique effective for use in, for example, speech input of a telephone number or speech input of a destination in a navigation system.

  Conventionally, speech recognition apparatuses that compare input speech with a plurality of comparison target patterns (recognition dictionaries) stored in advance and recognize a result with a high degree of coincidence have already been put to practical use. For example, a navigation system Is used for a user to input a place name or facility name by voice, or for inputting a telephone number by voice in a hands-free system (see Patent Document 1). . In particular, when the driver himself uses the in-vehicle system, since it is not accompanied by button operation or screen gaze if the voice is input, it is effective because the safety is high even when the vehicle is running.

  In recent years, speech recognition technology that enables natural dialogue between humans and machines has increased, but in order for speech recognition devices to accept natural utterances, a large number of patterns to be compared are stored in the speech recognition device. It is necessary to keep. Therefore, in order to make it possible to recognize natural utterances, the technique disclosed in Patent Document 1 recognizes words at predetermined intervals, analyzes each word candidate as a keyword using a syntax analysis means, and is meaningless. It enables speech recognition of natural utterances including fluctuations of simple words and utterances. Further, the technique disclosed in Patent Document 2 tries to cover the low accuracy of spontaneous speech recognition with the speed of processing using the certainty of intermediate results.

However, the above method is performed after speech recognition of a huge comparison target pattern of natural utterances. After all, if the accuracy of the speech recognition result is low, it is difficult to cover by post-processing, and accurate natural It is considered that utterance recognition cannot be fully achieved.
JP-A-5-197389 JP 2005-283972 A

  As described above, in the conventional speech recognition apparatus, in order to recognize a natural utterance, an attempt is made to cover the low accuracy of the natural utterance speech recognition by performing the processing after performing speech recognition with a huge comparison target pattern. There were many techniques to do.

  However, in speech recognition, the recognition performance may be degraded as the number of comparison target patterns increases. Therefore, even if some processing is performed based on the results of speech recognition using a huge number of comparison target patterns, the lower the accuracy of the speech recognition results, the more difficult it is to cover with post-processing. The utterance may not be recognized.

  Accordingly, the present invention provides a speech recognition technology that can solve such problems and reduce the number of comparison target patterns that are actually used in speech recognition using a large number of comparison target patterns and suppress degradation in recognition performance. Objective.

  According to the voice recognition device of the first aspect, the recognition unit compares the voice input through the voice input unit with a plurality of comparison target patterns stored in advance in the dictionary unit and has a high degree of matching. The result is the recognition result. Here, the dictionary means has a plurality of types of dictionaries classified based on a plurality of predetermined utterance patterns, and sets priorities as follows.

  First, the dictionary priority order determination means uses a degree of similarity between the user's utterance pattern based on the recognition result by the recognition means and a plurality of predetermined utterance patterns, so that the higher the degree of similarity, the higher the rank. Determine dictionary priority. Then, the dictionary control means sets priorities of a plurality of types of dictionaries based on the determination result by the dictionary priority order determination means.

Even when substantially the same content is uttered, the utterance pattern may differ depending on the user's habit. For example, assuming that a restaurant is set as a destination in a navigation device, there are users who speak in the order of “object, action word” such as “I want to go to a restaurant” and “I want to make a restaurant a destination”. “Ah, restaurant” “Eh, restaurant”
As described above, there are users who omit operation words and add unnecessary words to the head and speak in the order of “unnecessary words, object words”.

  Therefore, in order to determine which of a plurality of types of dictionaries previously classified based on a plurality of predetermined utterance patterns can be used appropriately for the utterance pattern reflecting the user's habit, the user's utterance pattern The priorities of a plurality of types of dictionaries are determined such that the higher the degree of similarity between a plurality of predetermined utterance patterns, the higher the rank.

  If the recognition means obtains a recognition result by using a dictionary with high priority, even if it is a speech recognition apparatus having a huge comparison target pattern when viewed as a whole dictionary means, It is possible to reduce the number of comparison target patterns to be used and to suppress the degradation of recognition performance.

  Although priority is given according to the priority order, the priority degree may be fixedly set in advance, or the priority degree may be designated by the user as shown in claim 2. . That is, a receiving unit that can receive an instruction from the user is provided, and the dictionary control unit sets the priority level of the dictionary set with the priority order based on the instruction received through the receiving unit. The recognizing unit determines the degree of coincidence based on the priority set by the dictionary control unit.

  Even if the priority order of the dictionary itself is determined, there may be a case where it is better to make the priority higher or a lower priority. If it makes it as shown in Claim 2, such a user's intention can be reflected.

  When determining the priority order, in order to appropriately deal with the utterance pattern reflecting the user's habit, not only the recognition result corresponding to the immediately preceding utterance but also the past as shown in claim 3 It is also preferable to be based on a recognition result corresponding to a predetermined number of utterances. In that case, the recognition result by the recognition means is stored in the utterance history storage means for the past predetermined number of utterances, and the dictionary priority order determination means stores the predetermined number of utterances in the past stored in the utterance history storage means. The priority order of the dictionary is determined using the degree of similarity between the user's utterance pattern and a plurality of predetermined utterance patterns based on the recognition result corresponding to.

  In the speech recognition apparatus described above, the priority order of the dictionary is automatically determined and set by the apparatus side. However, as shown in claim 4, the priority order of the dictionary is set based on a user instruction. May be.

  According to the voice recognition device of the fourth aspect, the recognition means compares the voice input via the voice input means with a plurality of comparison target patterns stored in advance in the dictionary means and has a high degree of matching. The result is the recognition result. Here, the dictionary means has a plurality of types of dictionaries classified based on a plurality of predetermined utterance patterns, and sets priorities as follows.

  First, priorities of a plurality of types of dictionaries are set based on an instruction received via an accepting unit that can accept an instruction from a user. As described above, if the priority order is set based on the instruction from the user, the utterance pattern reflecting the user's habit is appropriately selected from a plurality of types of dictionaries previously classified based on a plurality of predetermined utterance patterns. Speech recognition can be performed using a dictionary with corresponding priorities.

  If the recognition means obtains a recognition result by using a dictionary with high priority, even if it is a speech recognition apparatus having a huge comparison target pattern when viewed as a whole dictionary means, It is possible to reduce the number of comparison target patterns to be used and to suppress the degradation of recognition performance.

  In this case as well, the priority level may be fixedly set in advance, or the priority level may be designated by the user as shown in claim 5. That is, based on an instruction from the user received through the receiving unit, the dictionary control unit sets the priority level of the dictionary set with the priority order.

  In addition, various application destinations of such a speech recognition device are conceivable. As an example, as shown in claim 6, it is necessary to designate a speech input means when the navigation device performs a navigation process. There is a navigation system in which an instruction of certain predetermined place name related data is used for a user to input by voice. As this navigation processing, for example, map display, route guidance, etc. can be considered.

  Embodiments to which the present invention is applied will be described below with reference to the drawings. The embodiment of the present invention is not limited to the following embodiment, and can take various forms as long as they belong to the technical scope of the present invention.

[Description of configuration]
(Description of the entire navigation system)
FIG. 1 is a block diagram showing a schematic configuration of a navigation system 2 having a voice recognition function. The navigation system 2 is a so-called car navigation system used by being mounted on a vehicle. The navigation system 2 is connected to a position detector 4, a data input device 6, an operation switch group 8, a control circuit 10 connected thereto, and a control circuit 10. And an external memory 12, a display device 14, a remote control sensor 15, a communication device 16, and a voice recognition device 30. The control circuit 10 is configured as a normal computer, and includes a well-known CPU, ROM, RAM, I / O, and a bus line for connecting these configurations.

  The position detector 4 has a known gyroscope 18, a distance sensor 20, and a GPS receiver 22 for detecting the position of the vehicle based on radio waves from a satellite. Each of these sensors 18, 20, and 22 has an error of a different nature, and is configured to be used while being interpolated by a plurality of sensors. Depending on the accuracy, a part of the above may be used, and further, a steering rotation sensor, a wheel sensor of each rolling wheel, or the like may be used.

  In addition to so-called map matching data for improving the accuracy of position detection, various data for navigation including map data and landmark data, the data input device 6 uses dictionary data used when the speech recognition device 30 performs recognition processing. Is a device for inputting. As the storage medium, another medium such as a hard disk, a DVD, or a CD-ROM may be used. When a DVD is used as the data storage medium, the data input device 6 is a DVD player.

  The display device 14 is a color display device. On the screen of the display device 14, the vehicle current position mark input from the position detector 4, the map data input from the map data input device 6, and further displayed on the map. Additional data such as guidance routes to be set and landmarks for setting points can be displayed in an overlapping manner. In addition, when a menu screen that displays a plurality of options, or when an option is selected, a command input screen that displays a plurality of options can be displayed.

The communication device 16 is for communicating with a contact specified by the set contact communication information, and is configured by a mobile communication device such as a mobile phone.
In addition, when the navigation system 2 inputs the position of the destination from the remote control sensor 15 or the operation switch group 8 via a remote control terminal (hereinafter referred to as a remote controller) 15a, the navigation system 2 can move from the current position to the destination. A so-called route guidance function is also provided which automatically selects an optimum route to form and display a guidance route. As a method for automatically setting an optimal route, a method such as the Dijkstra method is known. For example, a touch switch or a mechanical switch integrated with the display device 14 is used as the operation switch group 8 and is used for inputting various commands.

  The voice recognition device 30 can input various commands in the same manner when the user inputs voices, while the operation switch group 8 or the remote controller 15a is used for inputting various commands by manual operation. It is an apparatus for doing so.

(Description of voice recognition device 30)
The voice recognition device 30 includes a voice recognition unit 31, a dialogue control unit 32, a voice synthesis unit 33, a voice extraction unit 34, a microphone 35, a switch 36, a speaker 37, and a control unit 38. Yes.

  The voice recognition unit 31 performs input voice recognition processing on the voice data input from the voice extraction unit 34 according to an instruction from the dialogue control unit 32, and returns the recognition result to the dialogue control unit 32. That is, the speech data acquired from the speech extraction unit 34 is collated using the stored dictionary data, and the upper comparison target pattern having a higher degree of coincidence than a plurality of comparison target pattern candidates is displayed in the dialog control unit 32. Output to.

  In recognition of a word sequence in the input voice, the acoustic data (for example, cepstrum) is extracted by sequentially analyzing the voice data inputted from the voice extraction unit 34 with the acoustic model. The acoustic feature time series data obtained by this acoustic analysis is obtained. Then, the time series data is divided into several sections by a known HMM (Hidden Markov Model), DP matching method, or neural network, and it is determined which word stored as dictionary data corresponds to each section. .

  Based on the recognition result in the voice recognition unit 31 and the instruction from the control unit 38, the dialogue control unit 32 instructs the control circuit 10 to execute a response voice output instruction to the voice synthesis unit 33 or the processing of the navigation system itself. Thus, for example, a destination or command necessary for the navigation process is notified, and processing for instructing to execute the destination setting or command is performed. As a result of such processing, if the voice recognition device 30 is used, a destination can be instructed to the navigation system by voice input without manually operating the operation switch group 8 or the remote controller 15a. .

  The voice synthesizer 33 synthesizes a voice based on a response voice output instruction from the dialogue control unit 32 using a voice waveform stored in the waveform database. This synthesized voice is output from the speaker 37.

  The voice extraction unit 34 converts the surrounding voice captured by the microphone 35 into digital data and outputs the digital data to the voice recognition unit 31. Specifically, in order to analyze the feature amount of the input voice, for example, a frame signal of a section of about several tens of milliseconds is cut out at a constant interval, and whether the input signal is a voice section including the voice is included. Determine if there is no noise interval. Since the signal input from the microphone 35 includes not only the speech to be recognized but also noise, the speech section and the noise section are determined. Many methods have been proposed as this determination method. For example, the short-time power of the input signal is extracted at regular intervals, and depending on whether or not the short-time power equal to or greater than a predetermined threshold continues for a certain period. A method of determining whether a speech section or a noise section is often used. Then, when it is determined that it is a voice section, the input signal is output to the voice recognition unit 31.

  In the present embodiment, the user inputs voice through the microphone 35 while pressing the switch 36. Specifically, the control unit 38 monitors the timing at which the switch 36 is pressed, the timing at which the switch 36 is returned, and the time during which the pressed state continues, and if the switch 36 is pressed, The voice recognition unit 31 is instructed to execute processing. On the other hand, when the switch 36 is not pressed, the processing is not executed. Accordingly, voice data input via the microphone 35 while the switch 36 is being pressed is output to the voice recognition unit 31.

  By having such a configuration, in the navigation system 2 of the present embodiment, various processes such as route setting, route guidance, facility search, and facility display can be executed by the user inputting a command.

(Description of the voice recognition unit 31 and the dialogue control unit 32)
Here, the voice recognition unit 31 and the dialogue control unit 32 will be further described.
As shown in FIG. 2, the speech recognition unit 31 includes a collation unit 311, a dictionary unit 312, and an extraction result storage unit 313, and the dialogue control unit 32 includes a processing unit 321, an input unit 322, and a dictionary control unit. 323, an utterance history storage unit 324, and a dictionary priority order determination unit 325.

  In the voice recognition unit 31, the extraction result storage unit 313 stores the extraction result output from the voice extraction unit 34, and the collation unit 311 stores the extracted result in the dictionary unit 312. Collation is performed using existing dictionary data (hereinafter simply referred to as a dictionary). Then, the higher-level recognition result that has been compared with the dictionary by the matching unit 311 and determined to have a high degree of coincidence (likelihood) is output to the processing unit 321 of the dialogue control unit 32, and the processing unit 321 of the dialogue control unit 32 The recognition result is output to the control circuit 10.

  On the other hand, the control circuit 10 instructs the dialogue control unit 32 to weight the dictionary (priority level). The control circuit 10 receives an operation from the user via the operation switch group 8 (see FIG. 1), and outputs an instruction based on the operation to the dialogue control unit 32. The input unit 322 of the dialogue control unit 32 inputs the instruction and outputs it to the dictionary control unit 323.

  Further, the utterance history storage unit 324 outputs the utterance history to the dictionary priority order determination unit 325, and the dictionary priority order determination unit 325 outputs a dictionary priority order instruction to the dictionary control unit 323 based on the utterance history.

  The dictionary control unit 323 sets dictionary priorities and weights for the dictionary unit 312 of the speech recognition unit 31 based on the instruction input from the dictionary priority order determination unit 325 and the instruction input from the input unit 322. Do.

(Description of dictionary unit 312)
Here, the dictionary unit 312 will be described. The dictionary unit 312 includes a first dictionary 312a, a second dictionary 312b, a third dictionary 312c, and a fourth dictionary 312d. These four dictionaries 312a, 312b, 312c, and 312d are classified in advance based on four utterance patterns.

One example thereof will be described with reference to FIG. For example, assume that a restaurant is set as a destination. In that case, as an example of the user's utterance,
Uh, I want to go to a restaurant, where the restaurant is a destination. Um, I want to go to a restaurant, a restaurant.

This utterance can be classified, for example, by a combination of unnecessary words, object words, and action words.
In order to set a destination, an object is always required, so in the above example,
(1) Unnecessary word, object word, action word (2) Unnecessary word, object word (3) Object word, action word (4) It can be classified into a set of four utterance patterns. In the utterance example above,
(1) Uh, I want to go to a restaurant Oh, make a restaurant a destination (2) Uh, a restaurant Oh, a restaurant (3) I want to go to a restaurant Destination (4) A set of four utterance patterns Can be classified. FIG. 3A shows an example of pattern classification when the dictionary has a grammatical structure, and FIG. 3B shows an example of pattern classification when the dictionary has a linear structure.

  For these four dictionaries 312a, 312b, 312c, and 312d, priority and weight can be set respectively. The dictionary control unit 323 of the dialogue control unit 32 executes the setting of the priority order and the weight value. An example of setting priorities and weights for the dictionaries 312a, 312b, 312c, and 312d will be described later.

  Although the schematic configuration of the navigation system 2 has been described above, the correspondence between the configuration of the navigation system 2 in the present embodiment and the configuration described in the claims is as follows.

  In the present embodiment, the microphone 35 corresponds to a voice input unit, and the dictionary unit 312 in the voice recognition unit 31 corresponds to a dictionary unit. The collation unit 311 corresponds to a recognition unit, and the operation switch group 8 and the microphone 35 correspond to a reception unit. Further, the dictionary priority order determination unit 325 in the dialogue control unit 32 corresponds to a dictionary priority order determination unit, and the dictionary control unit 323 corresponds to a dictionary control unit. Further, the utterance history storage unit 324 corresponds to the utterance history storage means.

[Description of voice recognition processing]
The speech recognition process executed in the navigation system 2 of the present embodiment will be described with reference to the flowcharts of FIGS. These flowcharts show processing executed by the voice recognition unit 31 and the dialogue control unit 32.

  When the variable i is set to 1 in the first step S10 and a voice is input in the subsequent S20, the collation unit 311 collates the input voice with the dictionary of the priority order (i) in S30 and performs a recognition process. I do. In S40, the recognition result is sent to the processing unit 321.

  In S50, it is determined whether the likelihood of the recognition result is equal to or greater than a threshold value. If the likelihood is equal to or greater than the threshold value (S50: YES), the recognition result is confirmed (S80). Then, the recognition result is stored in the utterance history storage unit 324 (S140).

  On the other hand, if the likelihood is not greater than or equal to the threshold (S50: NO), it is determined whether or not the variable i is less than n-1. This n is the maximum priority. If i <n−1 (S60: YES), the variable i is incremented (i = i + 1) (S70). Then, the process returns to S30, where the incremented priority (i) dictionary and the input speech are collated to perform recognition processing.

  If i ≧ n−1 (S60: NO), the collation unit 311 collates the input speech with the dictionary of priority (n) to perform recognition processing (S90). In S100, the recognition result is sent to the processing unit 321.

  In S110, it is determined whether or not the likelihood in the priority (n) dictionary is equal to or higher than the likelihood in the priorities (1) to (n-1). If the likelihood in the priority (n) dictionary is equal to or higher than the likelihood in the priority (1) to (n-1) dictionary (S110: YES), the priority (n) dictionary is used. The recognition result is determined (S120). On the other hand, if the likelihood in the priority order (n) dictionary is less than the likelihood in the priority order (1) to (n-1) dictionary (S110: NO), the priority order (1) to (n- The recognition result with the highest likelihood in the dictionary of 1) is determined (S130).

After the recognition result is confirmed in S120 or S130, the recognition result is stored in the utterance history storage unit 324 (S140).
In subsequent S150 (see FIG. 5), the dictionary priority determination unit 325 determines whether there is a change in the dictionary priority. This determination is made based on the utterance history stored in the utterance history storage unit 324.

  If there is a change in the dictionary priority (S150: YES), the dictionary priority determination unit 325 instructs the dictionary control unit 323 to change the priority dictionary, and the dictionary control unit 323 determines the dictionary unit based on the instruction. Dictionary priority order is set for 312 (S160).

  For example, as shown in FIG. 2, the second dictionary has priority 1, the first dictionary has priority 2, the third dictionary has priority 3, and the fourth dictionary has priority 4. There may be a plurality of dictionaries with the same priority, such as one with priority 1 and three with the other three. For example, the second dictionary is given priority 1, and the first dictionary, the third dictionary, and the fourth dictionary are all given priority 2. Of course, depending on circumstances, there may be a plurality of prioritized dictionaries.

  Also, when determining the priority order, in order to be able to appropriately respond to the utterance pattern reflecting the user's habit, not only the recognition result corresponding to the immediately preceding utterance but also the recognition corresponding to the past predetermined number of utterances. You may make it based on a result. In that case, it may be possible to store a predetermined number of utterances (for example, 10 times) in the utterance history storage unit 324 and determine the priority order based on the recognition result corresponding to the predetermined number of utterances in the past. It is done.

The dictionary priority order is determined as follows, for example.
The number of appearances of the utterance pattern is obtained from the utterance history, and priorities are assigned in the order of the number of utterances. For example, when the past 10 utterance patterns are the first dictionary 3 times, the second dictionary 6 times, the third dictionary 1 time, and the fourth dictionary 0 times, the second dictionary has priority 1 and the first dictionary has priority 2. The priority dictionary 3 is determined as the third dictionary and the priority dictionary 4 is determined as the fourth dictionary.

  When the first dictionary is 3 times, the second dictionary is 3 times, the third dictionary is 3 times, and the fourth dictionary is 1 time, the first dictionary, the second dictionary, and the third dictionary are set to the priority order 1, and the fourth dictionary The dictionary is determined as priority 2. In this way, a plurality of dictionaries can be set with the same priority.

  There are several other methods for determining the dictionary priority order. For example, it is conceivable to set a threshold for the utterance history ratio. For example, it is set to 0.5 or more, 0.3 or more, 0.1 or more, and less than 0.1, and the past utterance pattern 10 times is the first dictionary 3 times, the second dictionary 4 times, the third dictionary 2 times. When the fourth dictionary is one time, the first dictionary and the second dictionary may be determined as the priority order 1, and the third dictionary and the fourth dictionary may be determined as the priority dictionary 2.

  The threshold of the past predetermined number of times and the ratio of the utterance history may be fixedly set to 10 times, 0.5, 0.3, 0.1, or configured to be instructable by the user. You may do it. In this case, for example, an instruction from the user is accepted via the operation switch group 8, and based on the accepted instruction, the dictionary control unit 323 sets a past predetermined number of times and a history ratio threshold.

In subsequent S170, it is determined whether or not there is a change in the weighting of the dictionary. This determination is made based on whether there is an instruction for weighting (priority) of the dictionary from the control circuit 10.
If there is a change in the dictionary weight (S170: YES), the dictionary control unit 323 sets the dictionary weight to the dictionary unit 312 based on the instruction (S180).

  Regarding the weighting of the dictionary, for example, the weighting is fixedly set to 1.0, 0.8, 0.7, 0.6 for the dictionaries of the priority order 1, 2, 3, 4 for example. Alternatively, the user may be able to give an instruction. In that case, for example, an instruction from the user is received via the operation switch group 8, and the dictionary control unit 323 sets weighting based on the received instruction.

As a case where the user instructs the setting of the weighting, for example, when it is felt that the recognition performance is poor with the current weighting, it may be instructed to change the priority level of the dictionary.
In addition, as a method of instructing weighting, a final weighting value itself (for example, a value such as 1.0, 0.8, 0.7, or 0.6) may be instructed, or instructed by a ratio or the like. May be. For example, if there are only two types of priorities, that is, a priority 1 dictionary and a priority 2 dictionary, if the ratio 6: 4 is indicated, the weight of the priority 1 dictionary is set to 1.0, and the priority 2 The dictionary weight is set to 0.67. The same applies even when there are three or more types of priority.

In S190, it is determined whether or not there is a continuation of the utterance. If there is a continuation of the utterance (S190: YES), the process proceeds to S10 in FIG.
If there is no continuation of the utterance (S190: NO), the speech recognition process is once ended with this, so the current dictionary priority and weight are stored (S200). In subsequent S210, the recognition result is notified. This notification may be notified by voice from the processing unit 321 via the voice synthesis unit 33 and the speaker 37, or the control circuit 10 displays the recognition result on the display device 14 based on an instruction from the processing unit 321. You may alert by.

  Then, it is determined whether there is a confirmation instruction (S220). The presence / absence of this confirmation instruction is determined based on, for example, voice input from the microphone 35 by the user. For example, if there is a voice input indicating content that may be interpreted as a confirmation instruction such as “Yes” or “Confirmation”, it can be determined that there is a confirmation instruction, and it is interpreted as not a confirmation instruction such as “No” or “No”. If there is a voice input indicating a good content, it can be determined that there is no confirmation instruction. Of course, the confirmation instruction by the user is not limited to the case where it is performed by voice input, and may be performed by, for example, a switch operation. In this case, the presence / absence of a confirmation instruction is determined based on whether or not an operation for instructing confirmation of the recognition result is performed via the operation switch group 8.

If there is no confirmation instruction (S220: NO), the process returns to S10 and the voice recognition process based on the voice input again is executed.
On the other hand, if there is a confirmation instruction (S220: YES), predetermined post-confirmation processing is executed (S230). The post-determination process in this case is that the processing unit 321 outputs a recognition result to the control circuit 10 and notifies that the recognition result is confirmed. In accordance with the post-confirmation processing, the control circuit 10 determines the destination or facility when the destination or facility is specified and input in the destination setting or facility search when using the navigation function, for example. Based on this, a search is performed.

[effect]
For example, when setting a restaurant as a destination, voice recognition processing is performed using the following four types of dictionaries shown in FIG.
(1) Utterance patterns of unwanted words, object words and action words Uh, I want to go to a restaurant, make the restaurant the destination (2) Utterance patterns of unwanted words, object words Uh, restaurants Oh, restaurant (3) Objectives , Utterance pattern of motion words I want to go to a restaurant I want a restaurant as a destination (4) utterance pattern of an object word Restaurant Here, a speaker speaks an unnecessary word, purpose, such as “Ah, restaurant”, “Uh, restaurant” If the utterance is often made in the form of a word, the priority of the dictionary (2), that is, the dictionary corresponding to the utterance pattern “unnecessary word, object” is set to 1. Even when a facility other than a restaurant is set as the destination, it is highly likely that the user speaks with the same utterance pattern, so by learning the utterance habits of the speaker in this way, the utterance pattern frequently used by the speaker And the priority of the corresponding dictionary is increased.

  If speech recognition is performed using a dictionary with higher priority, for example, there is a high possibility that a final recognition result can be obtained only by using a dictionary with priority 1, for example. Compared with the case where it does, the possibility that appropriate voice recognition will be realizable only by collating with few comparison object patterns becomes high.

  In addition, even if the recognition result cannot be determined simply by using the dictionary with priority 1, the final recognition is performed without collating with all dictionaries by using the dictionaries in order from the dictionary with the highest priority. The result may be finalized.

  As a result, when the dictionary means is viewed as a whole, even a speech recognition apparatus having a large number of comparison target patterns can reduce the number of comparison target patterns that are actually used and suppress a decrease in recognition performance.

  A description will be given taking a dictionary in the case of setting the restaurant shown in FIG. 3 as a destination as an example. The number of comparison target patterns in the entire dictionary of (1) to (4) is 7, but the number of comparison target patterns is 2 if only the dictionary of priority order (2) is used.

Note that voice recognition using other dictionaries is possible only with a low priority, so even if the utterance pattern of the speaker changes, it is possible to cope.
[Others]
(1) In the above embodiment, the case where there are four dictionaries 312a, 312b, 312c, and 312d in the dictionary unit 312 has been described. In the case of a dictionary, the process shown in the flowchart of FIG. 6 is executed instead of the process shown in the flowchart of FIG.

The process shown in FIG. 6 will be described.
When a voice is input in the first step S510, in S520, the collation unit 311 collates the input voice with the priority order dictionary and performs recognition processing. In S530, the recognition result is sent to the processing unit 321.

  In S540, it is determined whether the likelihood of the recognition result is equal to or greater than a threshold value. If the likelihood is equal to or greater than the threshold value (S540: YES), the recognition result is confirmed (S550). Then, the recognition result is stored in the utterance history storage unit 324 (S610).

  On the other hand, if the likelihood is not equal to or greater than the threshold value (S540: NO), the input speech is compared with the dictionary of priority 2 to perform recognition processing (S560). In S570, the recognition result is sent to the processing unit 321.

  In S580, it is determined whether the likelihood in the priority 2 dictionary is greater than or equal to the likelihood in the priority 1 dictionary. If the likelihood in the priority 2 dictionary is equal to or higher than the likelihood in the priority 1 dictionary (S580: YES), the recognition result in the priority 2 dictionary is determined (S590). On the other hand, if the likelihood in the priority 2 dictionary is less than the likelihood in the priority 1 dictionary (S580: NO), the recognition result with the highest likelihood in the priority 1 dictionary is determined (S600). ).

After the recognition result is confirmed in S590 or S600, the recognition result is stored in the utterance history storage unit 324 (S610). Thereafter, the process proceeds to S150 of FIG.
In this case as well, as in the case of the above-described embodiment, even if the speech recognition device has a large number of comparison target patterns when viewed as the entire dictionary unit, the comparison target patterns that are actually used are reduced, and deterioration of recognition performance is suppressed. be able to.

  (2) In the above embodiment, the case of destination setting has been exemplified. However, for example, when considering an execution command (eg, enlarge map) in a navigation device, for example, “Uh, enlarge the map”, “Oh, An utterance pattern such as “to 50 m scale” or “execute details” can be considered. And they can be classified as follows.

Unnecessary words: Uh, uh, Object: Large map, 50m scale, Detailed Action word: Execute, then, Therefore, set a dictionary for each utterance pattern based on these, and prioritize and weight each You only have to set it.

Various examples of these utterance patterns can be considered, but another example will be given.
When searching for facility names, a pattern of “unnecessary words, prefecture names, genre names, names, action words” can be considered. Further, when searching for a song name, a pattern of “unnecessary word, genre name, track number, singer name, song name, action word” can be considered.

  In addition, the utterance pattern includes “change of position of unnecessary words” such as “inverted” “operation word, object word”, “object word, unnecessary word, operation word”, “object word, operation word, unnecessary word”, etc. Such variations are also conceivable.

  (3) In the above embodiment, the priority of the four dictionaries 312a, 312b, 312c, and 312d in the dictionary unit 312 is automatically determined and set by the voice recognition device 30 side. You may make it set based on a user's instruction | indication.

  In that case, for example, an instruction from the user is accepted via the operation switch group 8, and the dictionary control unit 323 sets the priority order based on the accepted instruction. In the case of setting the priority order for each dictionary, for example, an utterance pattern is exemplified on the display device 14, and the user can set a desired order for each utterance pattern.

  (4) When a plurality of users use the dictionary, the priority order and the weight of the dictionary are stored in correspondence with the identification information (user ID) for each user, and when the navigation system 2 starts to be used (or the voice recognition device) This is possible if voice recognition using the setting information for each user can be executed by inputting the user ID at the start of use of 30).

It is a block diagram which shows schematic structure of the navigation system 2 provided with the speech recognition function. 3 is a block diagram showing the configuration of a voice recognition unit 31 and a dialogue control unit 32 in the voice recognition device 30. FIG. It is explanatory drawing which shows an example of the dictionary data of the dictionary part. It is a flowchart which shows the first half of a speech recognition process. It is a flowchart which shows the second half of a speech recognition process. It is a flowchart which shows another example of a speech recognition process.

Explanation of symbols

  DESCRIPTION OF SYMBOLS 2 ... Navigation system, 4 ... Position detector, 6 ... Data input device, 6 ... Map data input device, 8 ... Operation switch group, 10 ... Control circuit, 12 ... External memory, 14 ... Display apparatus, 15 ... Remote control sensor, 15a ... remote control, 16 ... communication device, 18 ... gyroscope, 22 ... GPS receiver, 30 ... speech recognition device, 31 ... speech recognition unit, 32 ... dialogue control unit, 33 ... speech synthesis unit, 34 ... speech extraction unit, 35 ... microphone, 36 ... switch, 37 ... speaker, 38 ... control unit, 311 ... collation unit, 312 ... dictionary unit, 313 ... extraction result storage unit, 321 ... processing unit, 322 ... input unit, 323 ... dictionary control unit, 324 ... utterance history storage unit, 325 ... dictionary priority order determination unit.

Claims (6)

Voice input means that can input voice in a row,
A speech recognition apparatus comprising: recognition means for comparing speech inputted through the speech input means with a plurality of comparison target patterns stored in advance in the dictionary means and having a high degree of matching as a recognition result. There,
The dictionary means has a plurality of types of dictionaries classified based on a plurality of predetermined utterance patterns,
Using the degree of similarity between the user's utterance pattern based on the recognition result by the recognition means and the predetermined plurality of utterance patterns, the priority order of the plurality of types of dictionaries is determined so that the higher the degree of similarity, the higher the ranking. Dictionary priority order judging means,
Dictionary control means for setting priorities of the plurality of types of dictionaries based on the determination result by the dictionary priority order determination means;
With
The speech recognition apparatus characterized in that the recognition means preferentially uses a dictionary with a high priority set by the dictionary control means to obtain a recognition result. The speech recognition apparatus according to claim 1,
It has a reception means that can receive instructions from users,
The dictionary control means sets a priority level of the dictionary in which the priority is set based on an instruction received through the receiving means,
The speech recognition apparatus characterized in that the recognition means determines the degree of matching based on the priority degree set by the dictionary control means. The speech recognition apparatus according to claim 1 or 2,
Utterance history storage means for storing the recognition result by the recognition means for a predetermined number of past utterances,
The dictionary priority determination means uses the degree of similarity between the user's utterance pattern and the predetermined plurality of utterance patterns based on a recognition result corresponding to the past predetermined number of utterances stored in the utterance history storage means, A speech recognition apparatus characterized by determining a priority of a dictionary. Voice input means that can input voice in a row,
A speech recognition apparatus comprising: recognition means for comparing speech inputted through the speech input means with a plurality of comparison target patterns stored in advance in the dictionary means and having a high degree of matching as a recognition result. There,
The dictionary means has a plurality of types of dictionaries classified based on a plurality of predetermined utterance patterns,
An accepting means capable of accepting an instruction from the user;
Dictionary control means for setting priorities of the plurality of types of dictionaries based on instructions received via the receiving means;
With
The speech recognition apparatus characterized in that the recognition means preferentially uses a dictionary with a high priority set by the dictionary control means to obtain a recognition result. The speech recognition apparatus according to claim 4,
The dictionary control means sets a priority level of the dictionary in which the priority is set based on an instruction received through the receiving means,
The speech recognition apparatus, wherein the recognition unit determines the degree of matching based on a weight set by the dictionary control unit. A voice recognition device according to any one of claims 1 to 5, and a navigation device that executes predetermined processing based on a result recognized by the voice recognition device,
The voice input means is used for a user to input an instruction of predetermined place name related data that needs to be specified at least when the navigation device performs a navigation process. system.