JPH08259125A - Voice inputting device for elevator - Google Patents

Abstract

PURPOSE: To obtain a voice inputting device at a high reliability, which is easy to be used by users of elevator, by comparing recognition results of plural voice recognizing units so as to judge the suitableness and unsuitableness of voice input. CONSTITUTION: The voice, which is input from a microphone 4, is converted to the digital value by an A/D converter 10, and thereafter, band limitation is performed by a band pass filter unit 11. A spectrum computing unit 30 forms the voice recognizing input pattern on the basis of the voice detected by a voice zone detecting unit 12, and outputs it to a first and a second voice recognizing units 31, 32. The first voice recognizing unit 31 is formed as a multiple layered perceptron type neural net work, and the second voice recognizing unit 32 is formed as a LVQ net. The first and the second voice recognizing units 31, 32 recognize the voice of the voice data, and a result thereof is output to a voice recognition confirming unit 33. On the other hand, the voice recognition confirming unit 33 computes the reliability of the voice recognition, and outputs a result thereof to an elevator control unit 18 and an informing unit 34.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a voice input device for an elevator which recognizes voice of an elevator user and registers a call for a car or a hall.

[0002]

2. Description of the Related Art A conventional elevator apparatus having a voice recognition device will be described with reference to the drawings. This conventional example is of a type in which a destination floor is registered before getting on an elevator car. FIG. 4 is a schematic diagram in which a voice recognition device is used in an elevator hall, and a car position display device 2, a destination floor call registration button 3, a voice recognition microphone 4, and a voice recognition process are normally performed near the hall door 1. An OK lamp 5 indicating that the voice has been broken and a reject lamp 6 indicating that the voice recognition cannot be performed are installed.

When the operation by the voice recognition is not performed, the elevator customer uses the destination floor call registration button 3 as in the conventional case.
Select the floor you want to go to and press the button, and the button lamp for that floor will light up. When the elevator car arrives, the previously operated destination floor call is transferred to the call in the car to bring the user to the destination floor.

In the case of performing voice recognition processing, when an elevator user approaches the microphone 4 and speaks, the voice recognition device operates. Then, voice recognition processing is performed,
When it is determined that the voice is normal, the OK lamp 5 lights up and the lamp of the floor corresponding to the destination floor call registration button 3 blinks. When it is determined that the sound is abnormal, only the reject lamp 6 is turned on.

The above-mentioned normal voice / abnormal voice is judged by comparing the input voice with a standard pattern stored in advance,
When the degree of similarity exceeds a certain standard, the OK lamp 5 is lit, and when the degree of similarity is less than the standard, the reject lamp 6 is lit for a certain time (1 to 2 seconds). Therefore, if the voice recognition device determines that the third floor
The OK lamp 5 may be turned on, and the lamp on the third floor of the destination floor call registration button 3 may blink. In this case, while the destination floor call registration button 3 is blinking, it is only displayed, and the elevator is not controlled. Therefore, if the user sees this result and is normal, the above constant If you wait a while, it will be registered automatically, and if it is abnormal, you can correct it by speaking again.

Next, the configuration of the voice recognition processing section will be described with reference to FIG. The voice input from the microphone 4 is A /
After being converted into a digital value by the D conversion unit 10, the bandpass filter unit 11 limits the band of the voice, and obtains a 12-bit digital value with a sampling frequency of 12 kHz, for example. Further, the bandpass filter unit 11 extracts only the characteristics of the audio signal from the digital value, converts it into a spectrum sequence of 8 msec unit, and compresses the information.

The voice section detector 12 detects valid voices, collects voice data to be actually voice-recognized,
Output to the sampling unit 13. The sampling unit 13 normalizes the voice data so as to match the voice section length of the standard pattern stored in the dictionary storage unit 14. As a result, the voice data is converted into data of 256 points, and compared with the standard pattern of 256 points stored in the dictionary storage unit 14, the CPU 15 calculates the degree of similarity and recognizes the standard pattern with the highest degree of similarity. As a result, it outputs to the operation output unit 16. The program storage unit 17 is a program of the above procedure.

Data from the operation output unit 16 is input to the elevator control unit 18 to control the elevator. The recognition result notification unit 19 has means for blinking the temporary floor call registration button 3 displayed from the call output unit 20 of the elevator control unit 18 for provisional display. Also,
Lighting of the OK lamp 5 and the reject lamp 6 is controlled.

Next, this voice recognition processing will be described with reference to the flowcharts of FIGS. First, when there is a voice input from the microphone 4 (step S1), it is detected whether the input is the voice level of the user or the background noise that cannot be judged as voice (step S2). This step S2
Then, it is also detected by the voice section detection unit 12 that proper voice is input, and if it is proper, step S
The voice recognition process 3 is performed.

In step S3, correction is possible by re-inputting voice. As shown in FIG. 7, step S3
In A, the voice input after the recognition processing is completed is detected, and when there is the input, the previous recognition result is invalidated (step S3B), and the recognition result by the current voice is prioritized (step S3C). ).

In step S4, the voice data as a result of the voice recognition process is compared with the standard pattern stored in the dictionary storage unit 14. If a similarity exceeding a certain standard is recognized, it is OK in step S5. The lamp 5 is turned on to notify the elevator user that the sound is valid. If the similarity is equal to or lower than the reference, the reject lamp 6 is turned on in step S6 to request re-input.

In step S7, it is determined whether a valid voice command can be registered. For example, when the non-stop floor is instructed, the blinking cycle 2 is selected in step S9 to notify the user that it is different from step S8 of the normal processing. In step S10, the recognition result is blinked at the above blinking cycle. The blinking cycle in step S8 is 0.5 seconds, and the blinking cycle in step S9 is about 0.3 seconds.

In step S11, it is determined whether the voice command is finished or before the operation. If it is after the voice command, the lamp lighting process is finished in step S12. That is, the lamp is turned off about 1 second after the OK lamp 5 is turned on, the recognition result which has been blinking until then is sent to the elevator control unit 18, and it is processed as if a call was actually created.

According to the above conventional example, the elevator user stands in front of the voice recognition microphone 4 and speaks the floor he / she wants to go to, and confirms the result by checking the lighting state of the destination floor call registration button 3. If the desired result is obtained, the call is registered after waiting about 1 second. Even if the result is incorrect, if you say it again while the destination floor call registration button 3 is blinking, the previously entered content, that is, the currently blinking data, will be erased and the result entered this time will blink. indicate. This allows correction until a correct result is obtained.

[0015]

However, in the case of the above-mentioned conventional example, even if the similarity between the voice data and the standard pattern is extremely high, the destination floor call registration button 3 is always blinked for a certain period of time, so that the user is accustomed to it. If the user is unfamiliar, the user may accidentally register the next call while the destination floor call registration button 3 is blinking, and the previously entered content, that is, the current blinking. There is a possibility that the existing data will be erased.

Therefore, if the similarity between the voice data and the standard pattern exceeds a certain standard, it is immediately registered. If the similarity is less than the standard, the word with the highest similarity and its score (similar (The degree is displayed with 100 points)
A method of displaying is also considered. However, in the case of this conventional example, since there are only two judgments as to whether or not the voice recognition is correctly performed, all the cases where the degree of similarity is equal to or lower than the reference are unregistered, and as in the previous conventional example, during blinking. It cannot be corrected.

Further, in each of the above-mentioned conventional examples, whether or not the voice input is proper is determined by the similarity between the voice data and the standard pattern. That is, since the voice recognition processing is performed by only one voice recognition unit, there is a problem that the reliability of voice recognition is not high.

[0018]

According to the present invention, a plurality of types of voice recognition means having different voice recognition processing methods are used as means for recognizing whether or not voice input is proper, and the results of both recognition means are used. By comparing, the suitability of voice input is determined. Further, according to the present invention, if the results of both recognizing means match, it is immediately registered, and if they do not match, the confirmation is requested from the user or the non-registration (re-registration request) depending on the degree of similarity. This is a configuration for selecting whether or not.

[0019]

According to the present invention, the suitability of voice input is judged by comparing the results of a plurality of types of recognition means. Therefore, the suitability of voice input is judged only by the similarity between the voice data and the standard pattern. The reliability of voice recognition is higher than that of the conventional example, and the result of the voice recognition processing is in three stages: immediate registration, requesting confirmation from the user, and non-registration (re-registration request). It is easy for customers to use.

[0020]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to FIG. FIG. 1 is a view corresponding to FIG. 5 of the related art, and the same reference numerals as those in FIG. In the figure, reference numeral 30 is a spectrum calculation unit, which creates a voice recognition input pattern (voice data) from the voice detected by the voice section detection unit 12 and outputs it to the first and second voice recognition units 31 and 32.

The first and second speech recognition units 31 and 32 are both constructed of a neural network, and the first speech recognition unit 3
Reference numeral 1 is a multilayer perceptron type neural network (normal BP net), and the second speech recognition unit 32 is L
It is a VQ net. The first and second voice recognition units 31 and 3
In step 2, voice recognition of voice data is performed, and the result is output to the voice recognition confirmation unit 33. The voice recognition confirmation unit 33 calculates the certainty factor of voice recognition based on the voice recognition result, and outputs the result to the elevator control unit 18 and the notification unit 34.

FIG. 2 is a diagram showing a recognition process in the voice recognition confirmation section 33. Here, the first and second speech recognition units 31, 3
The features of the BP net and the LVQ net that compose 2 are outlined. The BP net has a high-performance classification function for the learned data, but it cannot be judged as “unknown”. Therefore, the recognition result that any voice data matches any one of the standard patterns stored in advance is output, and the similarity is not output. Also, L
Although the VQ net does not have the performance of the BP net, it has the feature that it can output the similarity. Also, basically, since the BP net and the LVQ net perform different processing, if the results of both nets match, it means that their reliability is very high.

The output x1 of the first voice recognition unit 31 is a standard pattern as a recognition result, and the output x2 of the second voice recognition unit 32 is a standard pattern as a recognition result, and d2.
Is the similarity between the audio data and x2, and is either "near" or "far". 33a is a comparison means for comparing x1 and x2, and when x1 = x2, the determination coefficient d
When 1 = 0 and x1 ≠ x2, d1 = 1 is output together with x1. 33b is a gate, and when d1 = 0, y
= X1 is output, and when d1 = 1, y = x1 is output if d2 = “close” and y = 0 is output if d2 = “far”.

Next, the voice recognition process will be described with reference to the flowchart of FIG. First, in step S20, it is determined whether or not the certainty factor is high. That is, if d1 = 0, it is determined that the certainty factor is high and step S21 is executed, and if d1 = 1, step S22 is executed. In step S21, it is determined whether the voice command can be registered. For example, when a non-stop floor is commanded, an error display is displayed (step S2
3) Request the customer to re-register, and if registration is possible, perform registration processing (step S24). At this time, a guidance display or the like indicating that the registration has been normally completed may be displayed.

In step S22, it is determined whether the certainty factor is medium. If y = x1 (that is, d2 = “close”), it is determined that the certainty factor is medium, step S25 is executed, and y = 0.
If (ie, d2 = "far"), step S26 is executed. In step S25, like step S21,
It is determined whether or not the voice command can be registered. If the voice command cannot be registered, step S26 is executed, and if the voice command can be registered, the confirmation request of step S27 is executed. This confirmation request asks the user to determine whether or not the result of voice recognition is correct, that is, whether the output y = x1 is acceptable. For example, voice confirmation by voice synthesis or display on a display device, etc. By.

If the customer confirms in step S28, the registration processing in step S24 is performed, and if the confirmation is not made or the user denies, step S26 is executed. Step S26 is a non-registration step, and requests the user to re-register. Note that the error display and re-registration request in step S23, the normally registered guide display in step S24, the re-registration request in step S26, and the confirmation request in step S27 are performed by the notification unit 34 in FIG.

As described above, the present embodiment uses two types of neural nets that perform different processes as means for recognizing voice data, so that the reliability of voice recognition is higher than that of the conventional device. Also, the result of the voice recognition processing is in three stages of immediately registering, requesting confirmation from the user, and non-registering (re-registration request), so that it is easy for the elevator user to use.

By the way, in the above embodiment, when it is judged in step S22 whether or not the certainty factor is medium, y is x.
It is judged whether it is 1 or 0, that is, whether d2 = "close" or d2 = "far". The certainty factor at this time indicates the similarity between the voice data and x1. However, d2
Indicates the degree of similarity between the voice data and x2, which is inconsistent. Therefore, in order to perform the process more accurately, when the comparison means 33a in FIG. 2 determines that x1 ≠ x2, x1 is input to the second voice recognition unit 32, and the similarity between the voice data and x1 is d2. And output again. However, the method of the above embodiment has no practical problem as long as it is determined whether the audio data is true audio data or noise.

In the above embodiment, the BP net and the LVQ net are used as the first and second voice recognition units.
Other neural nets may be used. For example, the first
LVQ, LVQ-2, MVQ, ART, RCE, RB that uses either BP net or CPN net as the voice recognition unit and can output the degree of similarity as the second voice recognition unit.
Either of the F nets may be used. Also, three or more kinds can be combined. Further, it is possible to combine the neural network and a voice recognition device having an HMM mechanism.

Further, in the above embodiment, the type in which the destination floor is registered at the hall has been described, but the same applies to the type having a call button for raising and lowering like an ordinary elevator. You can The same can be done by using the call button or door open / close button in the car.

[0031]

As described above, according to the present invention,
By comparing the recognition results of a plurality of voice recognition units, the suitability of voice input is determined, so the reliability of voice recognition is higher than in conventional devices, and the results of voice recognition processing are also registered immediately. Since there are three stages of requesting confirmation from the user and non-registration (re-registration request), there is an effect that it becomes easy for the elevator user to use.

[Brief description of drawings]

FIG. 1 is a block diagram of a voice recognition processing unit showing an embodiment of the present invention.

FIG. 2 is a diagram showing a recognition process in a voice recognition confirmation unit according to an embodiment of the present invention.

FIG. 3 is a flowchart showing a voice recognition process according to an embodiment of the present invention.

FIG. 4 is a schematic view of using a voice recognition device in a conventional elevator hall.

FIG. 5 is a block diagram showing a conventional voice recognition processing unit.

FIG. 6 is a flowchart showing a conventional voice recognition process.

FIG. 7 is a flowchart showing processing in the voice recognition unit in FIG.

[Explanation of symbols]

 3 Destination floor call registration button 4 Voice recognition microphone 10 A / D conversion unit 11 Band pass filter unit 12 Voice section detection unit 18 Elevator control unit 30 Spectrum calculation unit 31 First voice recognition unit 32 Second voice recognition unit 33 Voice recognition Confirmation unit 34 Notification unit

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI Technical display location G10L 9/10 301 G10L 9/10 301C

Claims (2)

[Claims] 1. A microphone provided in a hall or a car of an elevator, a voice recognition device for recognizing a voice signal input to the microphone, and an elevator control device for controlling the elevator according to a recognition result of the voice recognition device. In a voice input device for an elevator provided, a plurality of voice recognition units for processing input voice data and outputting a recognition result are provided, and the certainty factor of voice recognition is classified into three types by the output of the plurality of voice recognition units. A voice input device for an elevator, comprising: 2. One of the speech recognition units is a BP net or a CPN net, and the other is a neural net capable of outputting a similarity. The outputs of both nets are compared and the outputs are compared to determine whether they match or not. However, it is provided with means for outputting a high degree of certainty in the case of coincidence, and in the case of disagreement, for outputting whether or not the degree of certainty is medium according to the similarity of the other neural network. The elevator voice input device described.