JP4042589B2 - Voice input device for vehicles - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a vehicle voice input device, and in particular, a character is input by voice recognition, and the recognized character is a specific character such as a character having the same vowel, a character having a high voice recognition score, or a character corresponding to a word. The present invention relates to an audio input device for a vehicle that allows a driver to drive safely when inputting characters by reducing the time for the driver to gaze at the screen.
[0002]
[Prior art]
As a character input without using a voice, a method has been put into practical use in which a Japanese syllabary input palette is displayed on a screen provided with a touch panel, and a character is input by bringing a finger into contact with a character portion to be input on the touch panel. However, this method is used when a screen with a touch panel is placed at a distance that is out of reach (for example, when it is placed on the center panel of an automobile and an operation is input while driving). There is a problem that it is not possible.
[0003]
Also, a method has been proposed in which a character input is performed by moving the cursor in all directions with the joystick while selecting and displaying the character selected with the cursor using the syllabary input palette and joystick displayed on the screen. . However, this method has a problem that the joystick can move in all directions of up, down, left, and right diagonal directions, so that the cursor can easily move in an unintended direction and an operation error easily occurs.
[0004]
Furthermore, a Japanese syllabary input method using a jog dial has also been proposed (for example, Patent Document 1). However, in order to input a single character, it is necessary to perform selection and confirmation operations twice with the jog dial, and the input efficiency is improved. There is a problem of being bad.
[0005]
On the other hand, various methods have been proposed for a character input device by voice recognition that is effective as a vehicle character input device. However, since speech recognition errors exist and speech recognition with a recognition rate of 100% is difficult, correction means for erroneous recognition is indispensable for putting the character input device into practical use. Conventionally, a method has been proposed in which auxiliary input such as a keyboard is used to delete and re-input previous input (for example, Patent Document 2).
[0006]
This method requires voice input again every time it is erroneously recognized, so that it is not convenient, and if the same erroneous recognition is repeated again, input may become impossible.
[0007]
Also, in speech recognition of words, a method has been proposed in which a syllable is corrected by pointing a misrecognized syllable with a cursor and sequentially presenting characters one step at a time in sequence with the corresponding syllable (for example, Patent Document 3). ). Furthermore, a single syllable word is registered in a series of addresses for each identical vowel group by a single syllable speech recognition device, and a single syllable word of a recognition candidate is displayed and corrected one by one in the group by adding and subtracting the address. (For example, Patent Document 4).
[0008]
[Patent Document 1]
JP 2000-193475 A [Patent Document 2]
JP-A-1-189699 [Patent Document 3]
JP 61-41199 A [Patent Document 4]
Japanese Patent Laid-Open No. 56-1116096
[Problems to be solved by the invention]
However, the method of Patent Document 3 has a problem that it takes a long time to gaze at the screen because it is necessary to visually check the screen by pressing the correction key and to correct each character one by one. In addition, since only one character is displayed on the display unit in the speech recognition device of Patent Document 4, it is difficult to confirm the correction direction of a single syllable word at a glance, and it takes a long time to gaze at the screen as described above. There is a problem. In general, it has been known that the longer the gazing time of the screen, the more the vehicle's wobbling increases. AGGale et al (Eds), pp.335-344, North Holland Press, Amsterdam (1998)), and a method with a long screen gaze time as an audio input device used by a driver is inappropriate.
[0010]
The present invention has been made to solve the above problems, and converts input speech into characters by speech recognition. The converted characters are characters of the same vowel on the character input palette and characters having a high speech recognition score. Or a voice input device for a vehicle that allows a driver to input characters safely by reducing the time for the driver to gaze at the screen by making corrections to only one specific character such as a character corresponding to a word The purpose is to provide.
[0011]
[Means for Solving the Problems]
In order to achieve the above object, the present invention provides a voice input means for inputting voice, and voice recognition for recognizing the voice inputted from the voice input means and converting the recognition result into a character or a character string corresponding to the recognition result. And a character or a character string converted by the voice recognition means are displayed by arranging 50-characters or a part of 50-characters so that the same vowels are arranged on the same line. Display means for displaying one character selected from the characters of the Japanese syllabary or a part of the character of the Japanese syllabary along with a Japanese syllabary input palette , and conversion by the speech recognition unit while characters or selects a character on the alphabetical input palette corresponding to the character string display, change only the selected display on the same line in response to the modification request, the selection display in response to confirmation request And control means for determining the character, which is constituted contains.
[0012]
According to the present invention, the voice input from the voice input unit is recognized by the voice recognition unit, and the recognition result is converted into a character or a character string corresponding to the recognition result. Converted character or character string by voice recognition means, the Japanese syllabary same vowel is allowed to display by arranging syllabary character or a portion of the Japanese syllabary character, so as to line up on the same line Or a character selected from a part of the 50- character character is displayed on the display means together with a 50-character input palette for inputting one character at a time.
[0013]
The control means selectively displays one character on the character input palette corresponding to the character or character string converted by the voice recognition means. Thereby, the result of voice recognition can be visually confirmed. If the recognition result by the voice recognition means is notified by voice, the result of voice recognition can be confirmed by hearing.
[0014]
When the correction is requested when the result of the speech recognition is not correct, the control unit changes the selection display only on the same line in response to the correction request.
[0015]
In the speech recognition, the recognition rate of the same vowel character is very high. Therefore, by changing the selection display only to the characters of the same vowel in response to the correction request, it is possible to correct only the characters of the same vowel on the Japanese syllabary input palette , so watch the driver's screen. Time is reduced and you can enter characters safely. In addition, since a plurality of characters are displayed on the Japanese syllabary input palette , the direction of changing the selection display can be confirmed at a glance.
[0016]
Further, the control means is one of a character having a high speech recognition score, a character corresponding to a word, a character having a high speech recognition score, and a character corresponding to a word on the same line in response to a correction request. It is possible to change only to a character that satisfies a condition combining conditions for defining one character.
As another example of recognition candidates arranged on the same line in the Japanese syllabary input palette, there is a character having a high speech recognition score. In speech recognition, a plurality of recognition candidates having a high recognition score is generally obtained for input speech. Among them, a candidate having the highest score is taken as one recognition result. Even when misrecognition occurs, correct answers are often included in candidates with high recognition scores. Therefore, by changing the selection display only to candidate characters having a high recognition score in response to a correction request at the time of input, it is possible to input characters safely as described above.
[0017]
Another example of recognition candidates arranged on the same line in the Japanese syllabary input palette is a character corresponding to a word, that is, a character that becomes a candidate constituting a word to be input. For example, the case where this apparatus is used for inputting a place name for each character will be described as an example. Assuming that "to" and "u" are entered, the next character to be entered is a candidate for a word to be entered, such as "ki" (such as Tokyo) or "ka" (such as Tokai). It is limited to letters and “n” does not follow. Therefore, according to the correction request at the time of input, by changing the selection display only to the character corresponding to the word (in the above example, “ki”, “ka”, etc.), the character can be safely input as described above. be able to.
[0019]
Then, the control means confirms the character selected and displayed in response to the confirmation request. The correction request and the confirmation request may be input by voice or may be input by a mechanical operation means such as a switch.
[0020]
The same line of the Japanese syllabary input palette may be a horizontal line or a vertical line, but a horizontal line is more preferable for the driver.
[0022]
DETAILED DESCRIPTION OF THE INVENTION
In the following, referring to the drawings, a character string is input by voice recognition by speaking characters one by one by voice , and input candidate characters having a high voice recognition score are displayed on the same line in the left-right direction. An embodiment in which the present invention is applied to a voice input device (character string input device) for car navigation that inputs a character that can be corrected only to other characters having a high voice recognition score will be described in detail.
[0023]
As shown in FIG. 1, the vehicular audio input device according to the present embodiment stores in advance an audio input device 11 composed of a microphone or the like for inputting audio from a driver and the like, and audio input from the audio input device 11. And a speech recognition device 12 composed of a microcomputer for converting the recognition result into characters or character strings corresponding to the recognition result. Connected to the speech recognition device 12 is a dictionary database 17 that records various dictionaries including a dictionary of single syllables. The speech recognition device 12 selects any one dictionary recorded in the dictionary database 17 and executes speech recognition processing.
[0024]
The voice recognition device 12 is connected to a voice output device 13 configured by a speaker or the like that outputs voice recognized by the voice recognition processing, and the voice output device 13 is connected to a determination device 14.
[0025]
Further, an auxiliary display device 16 composed of a liquid crystal display device or the like is connected to the voice recognition device 12. As shown in FIG. 2, the auxiliary display device 16 has a window 16A for displaying characters or character strings converted by the speech recognition device 12 for each single syllable, commands “different”, “confirm”, and “stop”. A command display unit 16B displaying names, a character input palette unit 16C for displaying a character input palette configured by arranging character strings of the same vowel in the left-right direction for inputting one-fifth characters one by one, and A switching command section that displays "left" and "right" command names for moving the character input palette selection display to the left and right, and "next" command names for switching the character input palette to the next screen. 16D is provided.
[0026]
The command unit 16B, the character input palette unit 16C, and the switching command unit 16D are provided with a touch panel. By touching the finger, a command corresponding to the displayed character can be input, or input candidate selection display. It is possible to change. The selection display of commands and input candidates can be changed not only by the touch panel, but also by commands input by voice as described later or signals input from the auxiliary input device 15.
[0027]
The character input palette according to the present embodiment includes a one-stage palette in which a plurality of characters of the same vowel are displayed side by side in the left-right direction (horizontal direction). A plurality of characters of the same vowel in this character input palette are composed of characters having a high speech recognition score. According to this character input palette, by inputting the “next” command, the displayed character input palette can be switched to the next character input palette and displayed. The next-stage character input palette is also composed of a one-stage palette in which characters of the same vowel are displayed in a row in the horizontal direction. Further, by inputting a “left” or “right” command, a correction request is input, and a character selection display (FIG. 2 shows a state where a single syllable character “Nya” is selected and displayed). Can move left or right. Therefore, in the present embodiment, the correction request can be input by moving the voice input from the voice input device 11 and the operation of the auxiliary input device 15, and the character selection display can be moved to the left or right.
[0028]
When the “Left” command is entered while the leftmost character of the character input palette is selected and displayed, the character input palette is switched to the previous character input palette (or the next character input palette) and the right end of the character input palette. When the “right” command is input in a state where the character is selected and displayed, the character input palette may be switched to the next character input palette (or the previous character input palette).
[0029]
The voice recognition device 12 is connected to an auxiliary input device 15 for moving the character selection display on the character input palette in the character arrangement direction of the character input palette by hand operation or switching the character input palette. . Note that “different”, “confirm”, and “cancel” commands may be further input from the auxiliary input device 15.
[0030]
Hereinafter, a processing routine for recognizing speech by the speech recognition device 12, changing the selection display of the input candidate character in response to the correction request, and confirming the input candidate character in response to the confirmation request and inputting the character string This will be described with reference to FIGS.
[0031]
When the car navigation switch mounted on the automobile is turned on, the processing routine shown in FIG. 3 is started, and the initial mode is set in step 100. As a result, the voice recognition device 12 selects the initial mode dictionary in the dictionary database and executes the voice recognition process.
[0032]
In the next step 102, it is determined whether or not there is a voice input from the voice input device 11. If there is a voice input, the voice input using the initial mode dictionary is recognized in step 104; A speech recognition process is performed for converting the recognition result into a character string corresponding to the recognition result. In step 106, it is determined whether or not a keyword such as “facility” has been input. Only when it is determined that a keyword has been input, in step 108, the recognition result and operation work information are transferred to the audio output device 13. As a result, the voice output device 13 informs the input request in a single syllable by voice and transfers the recognition result information to the determination device 14.
[0033]
Upon receiving the recognition result information, the determination device 14 switches the dictionary database to a single syllable mode dictionary and outputs a mode switching instruction to the speech recognition device 12 so as to execute speech recognition processing in an input mode suitable for the keyword content. .
[0034]
For example, if the recognized keyword is “facility”, the voice output device 13 outputs a voice message “Please speak the facility name one character at a time.” The determination device 14 enters the facility name input mode and outputs an instruction to switch the recognition dictionary database to the single syllable input mode to the speech recognition device.
[0035]
When a voice other than the keyword is input or when the keyword is not recognized due to misrecognition, the initial mode is continued as it is, and the keyword standby state is continued.
[0036]
If the recognition result is notified by voice from the voice output device 13, the driver can confirm the recognition result by hearing, so that the driver speaks the same voice as the recognized voice. This can be determined without viewing the auxiliary display device.
[0037]
In the next step 110, it is determined whether or not a mode switching instruction is input from the determination device 14. When the mode switching instruction is input, the recognition dictionary database is converted into a dictionary corresponding to the input keyword in step 112. Switch to the input mode corresponding to the keyword.
[0038]
In step 114, a single syllable input mode process is performed in which a speech recognition process is performed using a dictionary corresponding to the keyword, and the recognition result is converted into a character or a character string corresponding to the recognition result.
[0039]
Next, the details of step 114 will be described with reference to FIG. In step 200, it is determined whether or not there is a voice input from the voice input device 11. If there is a voice input, a voice recognition process in a single syllable mode is performed in step 201, and based on the voice recognition result in step 202. It is determined whether or not a command such as “different”, “confirm”, “left”, “right”, or the like has been input by voice.
[0040]
If it is determined in step 202 that a command other than a command, that is, a single syllable speech is input, it is determined in step 204 whether or not it is the first syllable of the input speech. In the case of the first syllable, the recognition result is output to the voice output device 13 to notify the recognition result by voice, and the character converted by the voice recognition result is displayed on the window 16A of the auxiliary display device 16. FIG. 2 illustrates the result of the recognition and display of the first syllable “ya” of “office”.
[0041]
If it is determined in step 204 that the syllable is not the first syllable, that is, if it is determined that the next syllable has been input, a process of confirming the syllable recognized last time in step 208 is performed. The recognition result is output so that a sound in which the previously recognized syllable and the currently recognized syllable are continuously output is output, and the previously recognized syllable and the currently recognized syllable are continuously displayed in the window 16A. Display the character string.
[0042]
As a result, the driver can confirm the result of the voice recognition by hearing, so that the result of the voice recognition can be confirmed without viewing the window 16A of the auxiliary display device 16.
[0043]
If it is determined in step 200 that there is no voice input, it is determined in step 212 whether or not a predetermined time has passed. If the predetermined time has not passed, the voice input is waited for. When there is no input, it is determined in step 214 whether or not there is an unrecognized voice that has been recognized by the voice recognition process. If there is an unrecognized voice that has been recognized, step 216 is determined. The voice that has been recognized but not yet confirmed is confirmed and converted into characters.
[0044]
Thereby, when there is no voice input for a predetermined time, the voice recognized last time can be automatically determined.
[0045]
When it is determined in step 202 that a command has been input, processing corresponding to the command described below is executed. First, if it is determined in step 218 that a command representing “different” has been input, the voice notified from the voice output device 13 based on the result of voice recognition is different from the voice intended for input. Then, without recognizing the recognition result, the recognition result is canceled in step 220, and the process returns to step 200 to be in a voice input standby state.
[0046]
If it is determined in step 222 that a command indicating “confirm” has been input, the voice notified from the voice output device 13 based on the result of the voice recognition is a voice intended for input. A process for confirming the voice recognized in step S3 is performed, and the process returns to step 200 to enter a voice input standby state. Therefore, in this embodiment, the input candidate can also be confirmed by inputting a command representing “confirm” by voice. That is, in the present embodiment, the confirmation request is made either by inputting the “confirm” command, inputting the next voice within a predetermined time after inputting the voice, or no voice input state for a predetermined time or more after inputting the voice. Can be performed.
[0047]
When it is determined in the next step 226 that a command representing movement such as “right”, “left”, “next”, etc. is input, a command representing “right”, a command representing “left” in step 228. 2 or a command indicating “next” is input, and if a command indicating “right” or a command indicating “left” is input, the character input palette shown in FIG. A process of moving one character “right” or “left” along the steps is performed, and the process returns to step 200. As a result, the character selected and displayed on the character input palette as an input candidate as a result of speech recognition can be corrected by moving one character only in the character arrangement direction of the character input palette, that is, only to the character of the same vowel.
[0048]
FIG. 2 shows a case in which “Nya” is recognized as a voice and “Nya” is selected and displayed on the character input palette even though “ya” is voiced. In this case, by inputting the command representing “right” twice, the character selection display moves to the right by two characters, and the correct input candidate “ya” is selected and displayed and confirmed by the confirmation request. Thus, correct characters can be displayed in the window 16A.
[0049]
If it is determined in step 228 that a command representing “next” has been input, a process of switching the character input palette to the next character input hallet so as to display the next candidate in step 232 is performed. Return to 200.
[0050]
If it is determined in step 226 that a command indicating “stop” has been input, this routine ends.
[0051]
In the above description, an example of inputting a command for moving the selection display by voice input has been described. However, when the driver operates the auxiliary input device 15, the character selection display is moved along a stage in which characters representing the same vowel are arranged. You may let them.
[0052]
In the above description, it is possible to output a plurality of recognition candidates in speech recognition. Considering that the recognition candidates are mostly characters of the same vowel, only a plurality of characters having a high speech recognition score are used. The example of switching and using the arranged one-stage character input palette has been described. However, as shown in FIG. 5, a 50-character input palette in which characters of the same vowel are arranged horizontally, or It is also possible to use a Japanese syllabary character input palette using a part of the Japanese syllabary in which characters are arranged in the horizontal direction. Further, as the character input palette, a palette in which characters corresponding to words, that is, characters that are candidates for configuring the word to be input, are arranged may be used. For example, when entering a place name, “to”, “u”, “ki”, “ka”, “ho” (Tokyo, Tokai, Tohoku, etc.) are entered in the character input palette until “to” and “u” are entered. In the case of words) is displayed.
[0053]
Furthermore, in the above description, an example in which the selection display is moved in the horizontal direction has been described. However, the selection display may be moved in the vertical direction.
[0054]
As described above, in this embodiment, since a character or a character string can be input by uttering characters one by one by voice and the input candidate character can be corrected only to another specific character. The effect of reducing the time for gazing at the screen and enabling safe character input is obtained.
[0055]
【The invention's effect】
As described above, according to the present invention, since characters are input by voice recognition and the recognized characters can be corrected to only one specific character on the input palette, the time for the driver to watch the screen is reduced. Thus, it is possible to input characters safely.
[Brief description of the drawings]
FIG. 1 is a block diagram of an embodiment of the present invention.
FIG. 2 is a diagram showing a display example of a screen of the auxiliary display device according to the embodiment of the present invention.
FIG. 3 is a flowchart showing a processing routine according to the embodiment of the present invention.
4 is a flowchart showing a routine of single syllable input mode processing of FIG. 3;
FIG. 5 is a plan view showing a Japanese syllabary input palette displaying Japanese syllabary characters.
[Explanation of symbols]
11 voice input device 12 voice recognition device 13 voice output device 16 auxiliary display device 17 dictionary database

Claims (2)

Voice input means for inputting voice;
Voice recognition means for recognizing the voice input from the voice input means and converting the recognition result into a character or a character string corresponding to the recognition result;
Characters or character strings converted by the speech recognition means are displayed by arranging 50- characters or parts of 50- characters so that the same vowels are arranged on the same line. Display means for displaying a sound character, or a character selected from a part of the character of the Japanese syllabary, together with a Japanese syllabary input palette for inputting one character at a time;
Select and display one character on the Japanese syllabary input palette corresponding to the character or character string converted by the speech recognition means, change the selection display only on the same line in response to a correction request, A control means for confirming the character selected and displayed in response,
A voice input device for vehicles including the above. In response to the correction request, the control means is any one of a character on the same line and having a high speech recognition score, a character corresponding to a word, a character having a high speech recognition score, and a character corresponding to a word. 2. The vehicle voice input device according to claim 1, wherein only the character that satisfies the condition that combines the conditions for defining one character is changed.

Images