JPH03293274A - Voice register device for elevator - Google Patents

Abstract

PURPOSE:To enable to simply register the floor belonging to an target place and improve serviceability by constituting the device so that a voice command by a user is converted to a destination floor register operation of an elevator through a plurality of voice dictionaries. CONSTITUTION:An effective voice is extracted from signals from a band bus filter part 7 in a voice section detecting part 8, the voice is recognized to be input by giving it to a CPU 5 through a sampling part 9, voice recognition process is performed. and the degree of similarity against respective dictionary patterns stored in a dictionary memory part 10A is computed. The result that the degree of similarity is on the grade from the first rank to the third rank recognition word, is output. and the obtained word is extracted from the dictionary memory part 10A. The CPU 5 refers to the comparison table of a floor data converting part 15, converts the word to the designation of floor belonging to the extracted word to recognize the correct destination floor, the recognized result is displayed by lighting the destination floor lamp corresponding to the destination floor register lamp in a recognition result display part 14, and the result is sent to an elevator control part 13.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Field of Industrial Application) The present invention relates to a voice registration device that recognizes the voice of passengers using an elevator and controls the operation of the elevator.

(Prior art) Elevators are installed in various buildings, but the actions that elevator users directly operate are as follows: 1) Press the call button in the elevator hall.

j) Press the destination call button in the elevator car.

iN) Press the button to open and close the elevator door.

There are three types of items:

These are the sequences in which the user presses the corresponding button and the elevator responds as a result.

However, the buttons installed in elevators are not always easy to use, and often leave users dissatisfied.

For example, if you are carrying baggage in both hands, you will have to put down the baggage before pressing the button, and if you are a blind person, you will have to grope for the small button.

Furthermore, if a person is standing in front of the button, it may be difficult to operate it.

In order to solve these problems, voice registration devices that recognize and process the user's voice have been proposed.

In this elevator voice registration device, a microphone for voice input is installed in the place where the call button is conventionally installed, and the sound input to the microphone is 7'! As a result of analyzing 1, if it is determined that it is a command to operate an elevator, the corresponding operation is performed.

For example, if a user pronounces "gokai", 5
The call response lamp on the floor is turned on and the Nirebeta is operated to the 5th floor, allowing call registration to be made in exactly the same way as conventionally when pressing the 5th floor call button.

This speech recognition process stores pre-registered words as a dictionary, analyzes the frequency of the input voice, and compares the results with the data stored in the dictionary.
If the closest voice is similar to a certain percentage, the corresponding processing is performed. Generally, the speech recognition method used in elevators is called a speaker-independent word recognition method, which allows speech from any speaker, in other words, anyone, to be recognized.

In addition, the audio content is limited to about 16 words, such as ``Hirake'', ``Tojiro'', and ``Gokaij'', so in order to further improve the recognition rate,
For words that are difficult to recognize, several types of words are stored in the dictionary. For example, for registration on the first floor, "
By having multiple dictionaries such as ``Ikkai'' and ``Ichikai,'' the recognition rate is increased.

FIG. 8 is a block diagram of such a conventional elevator voice registration device that has been put into practical use, and FIG. 9 shows an external view of an elevator hall.

In the elevator voice registration device shown in FIG. 9, a hall indicator 2 and a pole call button 3 are installed near the elevator door 1, and a microphone 4 for voice registration is also installed near the hall call button 3. There is.

Therefore, in the elevator hall on the first floor, the user speaks into the microphone 4 in the direction he or she wants to go, for example, ``sea urchin''.
Alternatively, if the user utters "Gokai", this will be recognized and the upward direction hall call button 3 will light up, indicating that the call has been registered for the elevator, and at the same time, the response control for this hall call will be performed.

This voice recognition process will be explained based on FIG. 8. Since the voice input from the microphone 4 is an analog signal, the A/D converter 6 converts it into a digital value according to the amplitude of the voice for processing by the CPU 5. It is converted and output to the bandpass filter section 7. The bandpass filter unit 7 that divides the digital value into the frequency domain limits the audio band, and thereby, for example, the sampling frequency is 1.2.
k Obtain an array of H2112-bit digital values. However, since this value is a continuous numerical value with respect to the voice, only the voice features are extracted, converted into a spectral sequence of 8 m5ec units, the amount of information is compressed, and this is output.

The speech section detecting section 8 separates silent parts and effective speech parts in order to extract effective speech, and as a result, collects data of speech to be actually recognized.

When the sampling section 9 receives the actual speech data to be recognized outputted from the speech section detection section 8, the sampling section 9 normalizes the length of the input speech, which differs for each user, according to the speech section length of the dictionary pattern used for recognition. become As a result, the audio data is converted into 256-point data, and the CPU 5 compares this audio data with the 256-point dictionary pattern stored in the dictionary storage unit 10 to calculate the similarity, and Operation output unit 1 with high dictionary pattern as recognition result
]. The data from this operation output section 11 is input to the elevator control section 13 to perform various controls.

Incidentally, the operation output section 11 mentioned in "1" is often a simple digital output circuit. Further, 12 is a program storage unit that stores the program for the above-mentioned speech recognition process.

FIG. 10 illustrates the process of two series of speech recognition as speech waveforms. FIG. 4(a) shows the audio waveform input from the microphone 4, and this waveform includes various noises in addition to the noise that should be recognized. Therefore, as shown in FIG. 3(b), the speech section detecting section 8 extracts the necessary word sections. Then, in the sampling section 9,
), the frequency band is divided into 16 channels, as shown in
Speech recognition is performed comprehensively by determining the degree of similarity in each speech band.

(Problem to be Solved by the Invention) As described above, the conventional elevator voice registration device recognizes the voice and operates according to the voice command of the elevator user, so it is possible to perform the elevator registration process without contact. Users who have baggage at the same time do not have to operate the registration call button after unloading their baggage, and can register the call by speaking orally, which can improve service. , Furthermore, the other side of Zapis was desired.

In other words, the purpose of elevator users is not necessarily the name of the destination floor, but rather to go to the department or store (hereinafter collectively referred to as "department") to which that floor belongs. , the name of the floor to which the department you want to go is replaced with the name of the floor to which you belong, and normally if the user inputs the department you want to go to by voice, the floor to which you belong would be automatically registered. If this is done, the user will not have to memorize the comparison between the names of various departments and the floors to which they belong, and there will be less need to look at the information board, reducing the burden on the user. However, such an elevator voice registration device has not been known so far.

In addition, in conventional elevators, it is necessary to input the name of the floor you want to go to using voice words such as ``Ikkai'' and ``Gokai''. For users, when inputting by voice, they can clearly tell if it is registered by saying something like .
For example, if you want to go to the third floor, you may only say "san". In such cases, the voice recognition means is unable to process the voice recognition correctly and prompts the user to input the voice again, but inexperienced users simply cannot recognize the input voice because it is unclear. This could happen if you try to input "san" by voice again, thinking of something that didn't happen. In this case, voice recognition is still not possible, so the user may be confused as to what to do, and the voice registration device, which has been designed to improve service, becomes inconvenient instead. .

The present invention has been made in view of these conventional problems, and an object of the present invention is to provide an elevator voice registration device that can further improve service through voice registration.

[Structure of the Invention] (Means for Solving the Problem) [Claim] The invention of [Claim 1] Noveta Ondo 1 Registration Device 1r] is a microphone installed in an elevator car or an elevator hall, and a occupancy section name on each floor. , a floor layout information dictionary storage means for storing a voice dictionary of floor layout information such as tenant store name, and each floor layout information stored in this floor layout information dictionary storage means and each floor layout information to which it belongs. A floor information storage means for storing the correspondence relationship with floors, and a voice input from the microphone are compared with an audio dictionary of floor layout information stored in the floor layout information dictionary storage means to determine the input floor. A voice recognition processing means that vocally recognizes floor layout information, and a correspondence relationship between each floor layout information stored in the floor information storage means and the floor to which it belongs, based on the input floor layout information voice-recognized by the voice recognition processing means. The elevator system is equipped with a destination floor registration means for determining and registering a destination floor by referring to the destination floor, and an elevator control means for operating an elevator to the destination floor registered by the destination floor registration means.

Further, the elevator voice registration device according to claim 2 includes a microphone installed in an elevator car or an elevator hall;
Storing the voice dictionary of each destination floor] 1- A voice/dictionary storage means performs voice recognition on the voice input from the microphone by comparing it with the voice dictionary of the destination floor stored in the voice dictionary storage means. voice recognition processing means for registering a destination floor; and voice dictionary information display means for displaying in text the contents of the voice dictionary information stored in the voice dictionary storage means when the voice recognition processing means cannot voice recognize the input voice. and elevator control means for operating the elevator to the destination floor registered by the voice recognition processing means.

(Function) In the elevator voice registration device of claim 1, when a user speaks into the microphone the name of the department or store he or she wants to go to, the voice recognition processing means converts the input voice into floor layout information. The corresponding floor layout information is identified by comparing it with each floor layout information stored in the dictionary storage means. and,
Based on the specified floor layout information, the destination floor registration means determines the destination floor by referring to the correspondence between the floor layout information stored in the floor information storage means and the assigned floor. and register.

Then, the elevator control means registers the destination floor registration means and operates the elevator to the destination floor.

In this way, when an elevator user inputs the name of the department he or she wants to go to by voice, the elevator automatically determines the floor to which the user belongs, registers the call for that floor, and transports the user. be.

In the elevator voice registration device of claim 2, when an elevator user speaks into the microphone the floor he or she wants to go to, the voice recognition processing means generates the voice n2) the voice dictionary of the destination floor stored in the storage means. A voice recognition process is performed on the destination floor by comparing it with the input tone, and the destination floor whose voice is recognized is registered. Part 1. Then, the elevator control means recognizes the voice by the voice recognition processing means and operates the elevator to the registered destination floor.

Here, if the voice recognition processing means cannot recognize the voice inputted by the user because it is not registered in voice sound-1, the voice dictionary information display means displays the contents of the voice dictionary information. to notify the user and prompt the user to enter the correct voice input.

In this way, even users who are unfamiliar with elevator voice registration can use the elevator with peace of mind.

(Example) Hereinafter, embodiments of the present invention will be explained in detail based on the drawings.

FIG. 2 shows the appearance of the Nirebeta Hall according to an embodiment of the invention, in which the elevator user's destination floor is registered by the voice uttered by the elevator user. In FIG. 2, the same parts as in the conventional example shown in FIG. In this embodiment, a destination floor call button 3A is provided in place of the conventional hall call button 3 so that the user's destination floor can be directly registered from the second hall. The destination floor call button 3A functions similarly to a normal destination call registration button, and is provided with a lamp for each floor, and is configured to turn on the registered lamp of the destination.

In the voice registration device of this engineering 1novator, the call is automatically registered by the elevator user entering the voice input through the microphone 4 before getting on the elevator car. No need to register.

FIG. 1 is a block diagram showing an embodiment of the invention according to claim 1, and the same components as in the conventional example of FIG. 8 are designated by the same reference numerals.

The dictionary storage unit 10A, which serves as a floor layout information dictionary storage means, stores predetermined voice patterns.
For example, "Ikkai (1st floor)", "Nikai (2nd floor)",
The first store name such as "Sankai (3rd floor)" and the name of the department belonging to each floor such as "Uketsuke (reception)", "
"Lobby", "Soumka (General Affairs Department)", "Gijutsubu (Technical Department)", "Eigyobu (Sales Department)", and if the building where the elevator is installed is a multi-tenant building of multiple companies or stores, "00 Company". , ``△Store,'' company names, store names, etc. are stored as voice dictionary patterns. If the words to be stored are words that are clearly displayed on a guide map inside the building, etc., sufficient functionality can be achieved. Furthermore, if a function is added to change or add registered words, it will be possible to flexibly respond to changes in the number of residents.

Note that the dictionary storage unit 1. The larger the number of words in OA, the better the service, but this tends to lower the speech recognition rate, so it is desirable to set an appropriate number of words according to the speech recognition performance.

For example, if speech recognition of about 100 words is possible, 5
Assuming a multistory building, up to 20 words can be stored on each floor.

The floor data converter 15, which serves as a floor information storage means, stores a correspondence table between each department name and the floor name to which it belongs, in order to convert the department name recognized from the voice recognition processing result into the name of the floor to which it belongs. is the part that is memorized, for example, "
``Ikkai'', ``Uke]-6 Register ``1st floor'' for the words ``Soke'' and ``Lobby'', and register ``12th floor'' for the words ``Nikai'', ``Soum'', and ``Restrum''. This is possible.

Next, the operation of the elevator voice registration device configured as described above will be explained.

The sound section detecting section 8 extracts effective sound from the signal output from the band-pass filter section 7, and the extracted sound is sent to the CPU 5 via the sampling section 9, so that the CPU 9 uses the microphone. It is recognized that the elevator user has input the voice through step 3 (step 3).
1.01).

When the voice input is recognized in this way, voice recognition processing similar to that normally performed is performed (step S102),
Dictionary storage unit 1 - calculates the degree of similarity with each dictionary pattern stored in the OA. That is, 256 points of audio data given from the sampling section 9 and the dictionary storage section 1-
The degree of similarity is calculated by comparing it with the 256 patterns stored in the OA, and the recognized words with the first to third degree of similarity and the similarity results are output (step 3103).
).

In the following step 5104, if the first-ranked word among the candidate words performed in step S'+-03 shows a degree of similarity exceeding 80%, that word is determined to be the recognition result.

This determination method differs depending on the configuration, but the higher the similarity criterion, the lower the rate of normal recognition, so recognition will not be possible unless the pronunciation is accurate. Therefore, in general, the environment in which elevators are used has a constant low level of background noise, but unsteady noise such as the rustling of users and the sound of footsteps walking in the hallway occurs, so the degree of similarity is set at around 80%. is preferable.

In the following step 5105, the words obtained in step 5104 are stored in the dictionary storage unit 1. Extract from OA.

Then, the CPU 5 refers to the comparison table of the floor data converter]5, converts the extracted jli word into the name of the floor to which it belongs, and recognizes the regular destination (step 5106).
, Subsequently, the recognition result display section 4 displays the recognition result by lighting up the corresponding destination precursor 8 lamp of the destination destination registration lamp 3A, and sends the recognition result to the elevator control section 13 (step 31. 07).

The data sent to this elevator control unit]3 is the same data as the state in which the call button used in a normal elevator is pressed, and the elevator control unit]3 operates the Nirebeta in response to this call registration data. Implement controls.

In this way, the locations indicated on the guide map in the building are stored in a voice dictionary (7), the input user's voice is converted into data for each floor, and the voice recognition results are displayed. The elevators are operated so that even people who come to the building for the first time can use the elevator without getting confused.
You can go to the floor where the department you want to go to is located.Also, since the elevator will have a kind of guide map function, you can use the guide map to find the floor where the department you want to go to is located. There is no need to search for it and then input it by voice.
Particularly in high-rise buildings where the layout information for each floor is complicated, the system registers the floor where the department you want to go to by voice inputting the department name, reducing the burden on users. can be expected to significantly improve service. In addition, users are less likely to have to stop in a narrow elevator hall to find the floor of the department they want to go to, making it easier to get on and off the elevator.

In the above embodiment, the department name may also be the resident's name or room number in the case of a condominium or the like. Also, for example, when inputting the third floor by voice, by inputting r 156 J as a code, it is possible to recognize the third floor by voice using a floor data conversion table, which will improve the ability of users who know the code. It will now be possible to input voice input, increasing the crime prevention effect.

FIG. 4 shows an embodiment of the invention of claim 2, and 21
22 is an A/D converter, 23 is an acoustic analyzer, and 24 is a standard pattern register, and this standard pattern register 24 stores data of standard words. 25 is data acoustically analyzed by the acoustic analyzer 23;
Standard pattern 0 A similarity calculator that compares the standard pattern data registered in the register 24, 26 a CPU, 27 a proximity sensor that detects the distance between the user and the microphone, and 28 a device capable of voice recognition. This is a call registration device for registering a preceding call using a signal generated from the CPU 26 when a call is received.

Furthermore, 2 are recognition status indicators having an OK clamp voice recognition not possible lamp (? lamp), a voice instruction lamp (S P E A K lamp), etc., 30 is a guide character indicator, 3
1 is a guide character register for registering character data to be displayed.

FIG. 5 shows the configuration of an information input panel 32 for inputting audio in this embodiment, and this information input panel 32 is normally installed on a wall near the elevator door in an elevator hall. The information input panel 32 includes a microphone 21-, a proximity sensor 27, a 5PEAK lamp 29a constituting the recognition status display 29, an OK lamp 29b, and a 5PEAK lamp 29b. A lamp 29c and a guide character display 30 made of a liquid crystal panel are provided.

The operation of the elevator voice registration device of this embodiment will be described next.

Now, when a user who is unfamiliar with how to use an elevator's voice registration device tries to register a destination call by voice, 1.
- When you approach the information input panel 32 installed in the elevator hall on the floor, the proximity sensor 27 detects this and sends a signal to the CPU.
Send to 26th. The CPU 26 receives this signal, performs user detection processing, and displays the S provided in the two recognition status indicators.
Turn on the PEAK lamp 29a. At the same time, the voice registration device itself is activated.

Suppose that after the SPEALK lamp 29a is turned on, the user wants to go to the third floor, so he just utters "Zan" without knowing the correct utterance. Then, the microphone 21 receives this voice, inputs the voice input to the A/D converter 22 for A/D conversion, and gives the converted digital signal to the voice analyzer 23, where it is frequency-analyzed. Extract the features.

Then, the similarity calculation unit 25 compares this feature data with the standard pattern registered in the standard pattern registration unit 24 to determine the degree of similarity and makes a determination. For example, the degree of similarity with data such as ``Nikai'', ``Zankai'', ``Yongkai'', etc. is calculated, and if a pattern is found that gives a degree of similarity that exceeds a threshold, it is determined that speech recognition is possible.
The K-clamp 9b is turned on to register the destination floor.

However, in this case, the user is not familiar with elevator voice registration and simply pronounces it as "Zan", so the word registered as a standard pattern is not found and the threshold is exceeded. No standard pattern was found that gives a degree of similarity, and it was determined that it was unrecognizable? The lamp 29c is turned on and call registration is no longer executed.

however,? Even if the lamp 29c lights up, a user who is not familiar with the usage may think that it is because the pronunciation of the word he/she entered is not clear, and the user may repeat the same word again.
There is a high possibility that he will utter "Zan". On the other hand, in this embodiment, character information of a certain word is registered in advance as a standard pattern in the guide character register 21, and this character information is transferred to the CP.
After the information is taken into U26, it is sent to the guide character display 30, and characters are displayed on the liquid crystal panel as shown in FIG.

By doing this, even users who are not familiar with elevator voice call registration can input their destination floor using the correct words by looking at the guide text display, and are able to correctly input voice calls the second time. You will be able to do it.

In addition, the timing to display the characters is after the first unrecognizable determination is made. This is done after the lamp 29c lights up. This is because constantly displaying the guide may be bothersome to users who are accustomed to using this elevator.

In this embodiment, the standard pattern of the guide character display 30 that displays guide characters is written in both katakana and Roman letters so that it can be easily understood, especially by foreign users who are not accustomed to using elevators. This is what I did.

However, the display mode of the guide character display 30 is not particularly limited, and if there are many floors and all the floors cannot be displayed at once, it is possible to scroll the display. The embodiment is not limited.

FIG. 7 shows a flowchart of the operation of this embodiment. When the user approaches the microphone 21, the proximity sensor 27 detects this and lights up the 5PEAK lamp 29a (step S20), 5202). When the user looks at this and inputs his/her destination floor by voice (step 8203), the degree of similarity with the standard pattern is determined for the voice input (step 5204).

Here, if a standard pattern giving a degree of similarity exceeding the threshold value is found, the OK clamp 9b is turned on and the call for the destination floor indicated by the standard pattern is registered (step 52058206).

However, if no standard pattern exceeding a predetermined threshold is found in step 5204? The lamp 29c is turned on and guide characters are displayed (step 52).
07.8208).

In this way, in this embodiment, if a word different from the pre-registered standard pattern words is input and it is determined that it is unrecognizable, the information input panel 3
Since the content of the words pre-registered on the guide character display 30 (No. 2) is displayed in text to inform the user, even users who are not familiar with voice registration can easily select the correct words. Voice input is possible, which improves the recognition rate, improves the efficiency of elevator operation, and improves service to users.

[Effects of the Invention] According to the invention as claimed above, the voice commanded by the user is converted into the advance registration operation of Nirebeta using a plurality of voice dictionaries. Compared to button operations, it is easier for the user to register the floor to which the desired department belongs, and the service can be improved.

In addition, in the invention of claim 2, if a user who is unaccustomed to using an elevator attempts to register a voice with the wrong words and it is determined that the voice cannot be recognized, the correct word is displayed as a text guide. By looking at the display and inputting the correct words by voice, the service for inexperienced users can be improved.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing an embodiment of the invention of claim 1;
Figure 2 is an external view of the elevator hall in which the above embodiment is installed, Figure 3 is a flowchart 1 explaining the operation of the embodiment 1, and Figure 4 is an embodiment of the invention of claim 2. FIG. 5 is a front view of the information input panel used in the first embodiment, FIG. 6 is a front view showing an example of the guide character display used in the above embodiment, and FIG. FIG. 8 is a block diagram of the conventional example, FIG. 9 is an external view of the elevator hall of the conventional example, and FIG. 10 is an output waveform diagram of each part of the conventional example. 7 ] - Elevator door 2 Indicator 3
A... Advance registration button 4... Microphone 5... CPU6... A
/D converter 7... Bandhus filter 8... Voice section detection section 9... Zumbling section 1
0 A...Dictionary storage section]1...Operation output section]2.
...Program storage unit]-3...Elevator control unit 1-4...Recognition result display unit]5...Floor data conversion unit 21...Microphone 22...A/D converter 23. ...Acoustic analyzer 24...Standard pattern registration section

Claims (2)

[Claims] (1) A microphone installed in the elevator car or elevator hall, a floor layout information dictionary storage means for storing a voice dictionary of floor layout information such as the name of the tenant department and store name of each floor, and this floor layout information. a floor information storage means for storing a correspondence relationship between each floor layout information stored in the dictionary storage means and the floors to which each floor layout information belongs; A voice recognition processing means that vocally recognizes the input floor layout information by comparing it with a voice dictionary of the floor layout information stored in the layout information dictionary storage means, and from the input floor layout information voice recognized by the voice recognition processing means, a destination floor registration means for determining and registering a destination floor by referring to the correspondence between each floor arrangement information stored in the floor information storage means and the assigned floor; and a destination floor registered by the destination floor registration means. An elevator voice registration device comprising an elevator control means for operating an elevator. (2) a microphone installed in an elevator car or an elevator hall; a voice dictionary storage means for storing a voice dictionary for each destination floor; and a destination floor that the voice dictionary storage means stores in response to voice input from the microphone. voice recognition processing means for performing voice recognition and registering a destination floor by comparison with a voice dictionary; and when the voice recognition processing means cannot recognize the input voice, the voice dictionary information stored in the voice dictionary storage means A voice registration device for an elevator, comprising: voice dictionary information display means for displaying contents in text; and elevator control means for operating the elevator to a destination floor registered by the voice recognition processing means.