SI22823A - Process and device for intelligent access control - Google Patents

Abstract

Subject of the invention is a process and device for intelligent physical access control for protected locations by a system for multi-sensor biometric and behaviouristic identity verification based on a device for the integration of sensor data. The process for physical access control according to the invention is performed via a multisensor data acquisition on the behaviour of users during the time directly, during and after registration at the access terminal, adaptable integration of any number of sensors in a single device, storing the acquired data in a database and a two-level intelligent processing of data with machine learning. The first level of data processing consists of several adaptable modules for learning, among them is also learning based on time differences between the individual events. The second level of data processing is meta learning, which is performed based on information from the first level of learning. Regarding the adjustable parameters which originate from the security requirements and the acquired knowledge intelligent deduction on potential deviant behaviour of the controlled persons is performed.

Description

System and procedure for voice-guided telephone communication

The invention belongs to the field of systems and procedures for establishing telephone communication by means of speech, more specifically to the field of accessories for persons with special needs who cannot use normal telephone sets or because of motor impediments. to the field of digital signal processing and the field of voice interfaces in telecommunications.

A technical problem

A technical problem solved by the voice-guided telephone communication system of the invention is the design of an electronic system that enables the user, and especially the user with special needs, to voice (voice) establish and interrupt telephone communications, allowing the user to communicate smoothly with their surroundings , without inadvertently triggering the system. The solution should allow a user with a disability to activate the voice-guided telephone at any time by means of speech, select the appropriate telephone number, make a call, deactivate the voice-guided telephone, and then continue voice communication with his or her surroundings, as usual telephone users do after the telephone conversation . A problem solved by the present invention is the design of an algorithm that will activate, by means of digital signal processing, individual parts of a voice-guided telephone communication system to establish a telephone line, dial a dialed number, make a conversation, and interrupt a telephone line with a dialed number. The object and object of the invention is to perfect the process of communication by telephone by eliminating the need for tactile communication, so that the entire process from the establishment to the disconnection of the telephone connection can be performed only by voice and by computer activation of certain functions of existing processor-controlled devices such as computer, mobile phone, game console, and similar devices.

Voice-over-telephone communication is an indispensable method of communication today. In addition to the basic idea of transmitting a message remotely, the content of the message itself is important. Generally speaking, using a phone improves people's communication skills. This crowd also includes people with disabilities who, due to a medical condition or other causes, cannot use a regular telephone in the same way as other people. The problem most often occurs at the stage of using the phone, when the user has to use the tactile communication, that is, using his hands, to select the appropriate phone number and make a phone call.

The speech recognizer recognizes the input speech signal. Generally speaking, there is a problem with speech recognition when, in a given situation, the user does not want to forward the message to the recognizer but to a third party. In such a case, the use of a switch, the use of tactile communication, should prevent unwanted triggering of the speech recognizer and the wrong response of the system.

The present invention is implemented in such a way that, while using the voice-guided communication of the telephone, the user can also conduct daily conversation without using a switch or switch. tactile communication that would allow interruption of speech signal acquisition.

The state of the art

There are quite a few solutions on the subject of voice-controlled phones. One of the first solutions is disclosed in US Patent 5,007,081. The aforementioned patent file discloses a voice-guided telephone embedded in the case of a conventional desktop telephone with a keyboard, display, and handset. The phone is intended to be connected to an analogue public Switched Telephone Network (PSTN). The display is used to control a recognized voice command and allows the user to prevent an incorrectly recognized voice command by pressing the appropriate key. The operation of said voice-guided telephone also requires tactile communication with the device. The automatic speech recognition process in this solution is based on the Dynamic Time VVarping (DTW) process and depends on the speaker. This means that each user must learn the system with its set of voice commands before just using the system.

The proposed solution differs from US Patent No. 5,007,081 described above in that it requires no tactile communication, requires no prior learning of voice commands, is independent of the speaker, and the automatic speech recognition system is based on covert Markov models. The main difference between the proposed solution and the patent of US 5,007,081 is that the proposed solution enables the voice activation of the telephone dialing or telephone dialing. called persons. The proposed solution is not only limited to the analogue telephone network, but can operate on any digital telephone network (ISDN, VolP).

Another known solution is disclosed in U.S. Patent No. 5,452,340. It is an improvement of the aforementioned solution, since it allows voice dialing of the called person and the corresponding number in the phone book during the process of making a phone call. Here again, both voice, voice command input and tactile communication are used to control the validity of voice commands.

The proposed solution differs from US Pat. No. 5,452,340 described above in that no tactile communication is required to confirm commands. The main difference between the proposed solution and the patent of US 5,452,340 is that the proposed solution enables the voice dialing of the telephone number or the called person to be activated.

U.S. Pat. No. 5,483,579 introduces a device that enables voice-controlled dialing and can be used in combination with several classic telephone sets on the same telephone line.

The proposed solution differs from the above-described US Pat. No. 5,483,579 in that it is not used as an upgrade of classic telephone sets but acts as a standalone voice-guided communication device. The proposed solution enables voice activation of the dialing start of the telephone number or the person called, while in the aforementioned patent, voice dialing is only activated by lifting the handset and is not completely hands free.

An example of a voice-controlled telephone without any use of tactile communication is presented in U.S. Patent No. 6,167,251. The patent describes a mobile cellular telephone system (GSM or CDMA) that does not contain any key or key. keyboards. Both voice commands and telephone conversations are transmitted wirelessly through the base station to the exchange. The PBX server enables voice command recognition, telephone connections, voice-controlled directories editing, etc.

The proposed solution differs from US Pat. No. 6,167,251 above in that it does not contain a remote central unit that would perform the automatic speech recognition process, but that the recognition process is integrated directly into the voice-guided communication device system. Likewise, the proposed solution is not limited to mobile cellular telephony systems from a specific carrier that supports voice-guided services. The main difference between the proposed solution and the patent of US 6,167,251 is that the proposed solution enables the voice dialing of the telephone number or the called person to be activated. Although the solution in this patent does not include a classic keyboard or dial, it does have a key to turn the phone on and off, giving the center a signal to start processing the voice signal. The user speaks the telephone number only after a message is sent to the server.

US 2002 / 0097843A1 describes a special built-in device that provides voice support for any telephone set. The recognition system depends on the speaker, who must first learn the system to detect the control words.

The proposed solution differs from the above patent US 2002 / 0097843A1 in that it does not require any prior learning of spoken commands and is independent of the speaker. Also, the proposed solution is not used as a built-in device for classic telephones, but acts as a standalone voice-guided communication device. The main difference between the proposed solution and the one described above is that the proposed solution enables the voice dialing of the telephone or the called person to be dialed.

U.S. Pat. No. 5,799,065 describes a device called "virtual telephone operator" for voice-guided calling without the use of tactile communication. The three-pass automatic speech recognition system is based on Hidden Markov Models (HMM) using PLPRAST (Perceptual Linear Predictive Coding - Relative Spectral) features. The system is capable of processing current speech and is integrated into the telephone exchange.

The proposed solution differs from the above-described U.S. Pat. No. 5,799,065 in that the voice signal processing system, or it is not included in the telephone exchange for automatic speech recognition, but is used as a standalone voice-guided communication device. The main difference between the proposed solution and the patent of US 5,799,065 is that the proposed solution enables the voice dialing of the telephone or the called person to be dialed.

Description of solution to a technical problem

The essence of a voice-guided telephone communication system according to the invention is that the device comprises a microphone, a headset, a speech processing device on which speech recognition is performed or a remote microphone, a speaker, a customized voice processing device on which, in addition to speech recognition, additional preprocessing of speech is performed, and a control processor device running the application of establishing a telephone connection and voice transmission. The essence of the invention is an algorithm whereby all user-specific voice data is entered into a speech recognition processor that controls a controlling processor device whose algorithm enables telephone connection and speech transmission. The algorithm enables voice-rendered calls without any other user intervention. Applications of voice processor algorithms, custom voice processor devices, and control processor devices may be implemented on a computer, digital communication terminal, game console, personal organizer, and the like.

The subject of the invention is also a method for voice-guided telephone communication, which is implemented in such a way as to enable the call to be made by voice without any other intervention of the user.

The voice-guided telephone communication system of the invention will be described in greater detail by means of pictures showing:

Figure 1 - Block diagram of the voice-guided telephone communication system Figure 2 - System and procedure for voice-guided telephone communications Figure 3 - System and procedure for voice-guided telephone communications after embodiment I

The subject of the invention is a method for voice-guided telephone communication, which is implemented in such a way that the system described for the communication described in the invention also enables a call to be made by voice without any other intervention of the user. Using such a system of 'voice conversations', the user activates the voice-guided telephone communication system at any time, dials the appropriate telephone number, makes a call, deactivates the system, and then resumes voice communication with his or her surroundings, as usual telephone users do after the telephone conversation has ended.

The present invention is designed to allow the user, while using the voice-guided communication of the telephone itself, to conduct daily conversation without the use of a switch, tactile communication that would allow interruption of speech signal acquisition.

In the present invention, the system described also makes it possible for people with disabilities to use the telephone in a manner equivalent to that used by other people through speech technology. Using a voice-guided telephone, the user can activate the voice-guided telephone at any time, select the appropriate telephone number, make a call, deactivate the voice-guided telephone, and then resume voice communication with his or her surroundings, as users of a regular telephone do after completing a telephone conversation.

The voice-guided telephone communication system consists of a voice processor 1.1, a control processor 1.4, a handset 1.3 and a microphone 1.2. Speech processor 1.1 implements a speech recognition algorithm that converts the input speech signal to text format. The result of the recognition in text format is transferred to the control processor 1.4. The control processor 1.4 processes the command received from the speech processor 1.1 and performs the appropriate action. An action can be to enable a phone connection, select a phone number or name associated with the selected phone number, make a phone connection, answer an incoming call, reject a phone call, or end the phone connection. The headset 1.3 and microphone 1.2 can be wired or wirelessly connected to a speech processor 1.1 for speech recognition. When establishing a telephone connection, the user communicates with another user via microphone 1.2 and handset 1.3.

In embodiment I of the voice-guided telephone communication system, the system consists of a custom voice processor 1.5, a control processor 1.4, a speaker 1.6, and a remote microphone 1.7. On the custom voice processor 1.5, in addition to the speech recognition algorithm, an algorithm for additional preprocessing of the speech signal is implemented, which reduces the echoes that result from hands-free calling. The result of the recognition in text format is transferred to the control processor 1.4 in the same way as in the first version. The processor processor 1.4 processes the command received from the personalized voice processor 1.5 and performs the appropriate action to enable a telephone connection, select a telephone number or name associated with the selected telephone number, establish a telephone connection, answer an incoming call, reject a telephone call, or disconnect a telephone connection. Loudspeaker 1.6 and Remote Microphone 1.7 can be wired or wirelessly connected, as in the first embodiment, to a customized speech processor 1.5 for preprocessing and speech recognition. When establishing a telephone connection, the user communicates with another user via a remote microphone

1.7 and the speaker 1.6.

Voice processor 1.1, customized voice processor 1.5 and control processor 1.4 may be represented by a digital computer, a laptop computer, a digital communication terminal (mobile communication device - for example a mobile phone), a game console, a digital personal organizer, without the use of the invention. limited to the aforementioned devices only.

The process of performing an enabling, selecting a telephone number, establishing, receiving and rejecting a telephone call by voice-guided telephone communication is given in Figure 2. The system user of the invention, via a predefined keyword, triggers a virtual START key, which is bound via modules S2.1, S2 .2, S2.3 to module S2.4, and represents the voice processor device 1.1. Module S2.1 checks whether speech has been detected. Module S2.2 defines the characteristics of the speech signal and module S2.3 defines the spoken word. Module S2.4 determines the reliability of a recognized word. If there is a question in module S2.1 whether a speech has been detected, the NO will return to the START key again. When the answer is at the output of module S2.1 YES, the process proceeds through modules S2.2, S2.3 and S2.4 to module S2.5, which verifies that the recognized word is sufficiently reliable. If NO, the process returns to the virtual START key. When the answer is YES, the process continues in module S2.6, which checks whether the command to enable telephone connection is recognized. If yes

The YES process continues on module S2.14. If NO, the process continues in module S2.7, which checks whether the command to establish a telephone connection is recognized. If YES, the process continues in module S2.13 and module S2.14. If the answer is in module S2.7 NO, the process is continued in module S2.8, which checks whether the command to re-enter the telephone number is recognized. If YES, the process proceeds to module S2.14, the output of which is connected to the START key. If the answer to No is S2.8, then the process continues in module S2.9, which checks whether the rejection or rejection command is recognized. interruption of telephone connection. When the answer is YES, the process resumes in module S2.15, whose output is bound to the START key. When NO is answered in Module S2.9, the process continues in Module S2.10, which checks if a command to receive a telephone conversation is recognized. If YES, the process is continued via the S2.11 module which establishes a telephone conversation. If NO is answered in module S2.10, the process is continued via module S2.12, which allows the selected telephone number to be displayed. The output of Module S2.12 is linked to Module S2.14 to select the appropriate grammar. Module S2.13 has the command - Make a call. Module S2.15 has the command - End telephone conversation. Modules S2.5 to S2.15 perform the functions of a control processor 1.4.

The process of executing an enabling, selecting a telephone number, establishing, receiving, and rejecting a telephone call of a voice-guided telephone communication according to embodiment I is shown in Figure 3. The system user according to the invention triggers via a predetermined keyword a virtual START key which is bound via modules

S3.1, S3.2, S3.3, S3.4 to module S3.5 and is a customized voice processor 1.5. Module S3.1 checks whether speech has been detected. Module S3.2 additionally preprocesses the captured voice signal in order to reduce echoes. Module S3.3 defines the characteristics of the speech signal and module S3.4 defines the spoken word. Module S3.5 determines the reliability of the recognized word. If there is a question in module S3.1 whether a speech has been detected, the NO will return to the START key again. When the answer is at the output of module S3.1 YES, the process proceeds through modules S3.2, S3.3, S3.4 and S3.5 to module S3.6, which verifies that the recognized word is sufficiently reliable. If NO, the process returns to the virtual START key. When YES is answered, the process continues in module S3.7, which checks whether the command to enable telephone connection is recognized. If YES, the process continues on module S3.15. If NO, the process continues in module S3.8, which checks whether the command to establish a telephone connection is recognized. If YES, the process continues in module S3.14 and module S3.15. If the answer is in Module S3.8 NO, the process is continued in Module S3.9, which checks whether the command to re-enter the telephone number is recognized. If YES, the process proceeds to the module

53.15, the output of which is connected to the START key. If Module S3.9 answers No, the process continues in Module S3.10, which verifies that the reject or reject command is recognized. interruption of telephone connection. When the answer is YES, the process continues in the module

53.16, the output of which is bound to the START key. When NO is answered in Module S3.10, the process continues in Module S3.11, which checks whether the command to receive a telephone conversation is recognized. If YES, the process is continued through module S3.12, which establishes a telephone conversation. If NO is answered in module S3.11, the process is continued via module S3.13, which allows the selected telephone number to be displayed. The output of Module S3.13 is linked to Module S3.15 to select the appropriate grammar. Module S3.14 has the command - Make a call. Module S3.16 has the command - End telephone conversation. Modules S3.6 to S3.16 perform the functions of a control processor 1.4.

The user has an activated microphone 1.2 and headset 1.3 or remote microphone 1.7 and speaker 1.6 and guides normal communication with their surroundings. As soon as he wishes to make a telephone conversation, he utters a keyword or an appropriate sequence of keywords to enable a telephone connection using the invention described herein. After a correctly recognized keyword, the user speaks the desired phone number or the name of the person he wants to call and to which the corresponding phone number is associated in the system. If the wrong phone number is identified, the user can repeat it, otherwise it will trigger a call by saying the keyword to connect. During an active telephone connection only, the user can terminate it at any time by uttering a keyword or a selected sequence of keywords to end the telephone connection. By saying keywords to receive a telephone conversation or to reject a telephone connection, the user has complete control over the telephone communication. It is up to the user to determine the keywords to establish and disconnect the telephone connection.

The solution described in this patent application differs from the existing solutions in the following points:

Enable voice activation to start dialing a phone number or a person called.

The use of modern approaches in the process of automatic processing and speech recognition enables the processing of current speech and the recognition of voice commands even during the telephone conversation itself, without prejudice to starting and stopping the recognition. Such an approach, however, results in the conduct of everyday conversation without the use of a switch (tactile communication) that would allow the interruption of speech signal acquisition for the proper functioning of the speech recognition system.

Operating in the modern voice technologies of Vol P (Voice over

Internet Protocol ") or any digital telephone connection that transmits voice packets.

No specific hardware is required to operate.

Recognition is independent of the speaker and no prior learning is required on the individual speaker.

- Call Logic Recognition is integrated locally, saving on phone line load (only required when actually making a call).

Claims (10)

1. Voice-guided telephone communication system, which may be applied by means of a digital computer, a laptop computer, a digital communication terminal such as a mobile phone, a game console, a digital personal organizer or other similar electronic devices, characterized in that the user is using speech without the use of tactile communication, carry out telephone communication with a system consisting of a speech processing device (1.1) connected to a control processing device (1.4), a microphone (1.2) connected to a speech processing device (1.1), and headphones (1.3), which are connected to the speech processing device (1.1). Voice-guided telephone communication system according to claim 1, characterized in that the system consists of a custom voice processor (1.5) connected to a control processor (1.4), a remote microphone (1.7) connected to a custom voice processor ( 1.5), and a loudspeaker (1.6), which is also connected to a custom voice processor (1.5). 3. Voice-guided telephone communication system according to claim 1, characterized in that it allows the user to activate voice dialing of the telephone number or the called person by voice triggering via a predetermined keyword a virtual START key connected via modules (S2.1 , S2.2, S2.3) to the module (S2.4) and represents a speech processing device (1.1) for the module (52.1) to check whether speech has been detected, for the module (S2.2) to determine the characteristics of the speech signal and module (S2.3) determines the spoken word that module (S2.4) determines the reliability of the recognized word, that if module (S2.1) answers the question whether speech has been detected, the answer NO returns to the key again START that in the case when the answer is at the output of module (S2.1) YES proceeds through the modules (S2.2, S2.3 and S2.4) to the module (S2.5), which checks that the word is recognized enough it is certain that in the case when the answer is NO, the process returns to the virtual START key; the YES response proceeds in the module (S2.6), which checks whether the command to enable telephone connection is recognized, that if the YES response process proceeds on the module (S2.14), that if the answer is NO process proceeds in module (S2.7), which checks whether the command to establish a telephone connection is recognized, so that in the event that the answer is YES, the process continues in module (S2.13) and module (S2.14), so that when the module NO (S2.7) has the NO process resumed in the module (S2.8), which checks whether the command to re-enter the telephone number is recognized, so that if the YES process is continued to the module (S2.14), the output of which is connected to the START key, so that, if the module (S2.8) has the No response, the process continues in the module (S2.9), which checks whether the reject or execute command is recognized. interruption of the telephone connection to continue in the module (S2.15) whose output is linked to the START key in the case when the answer is YES, that in the module (S2.9) the NO process will continue in a module (S2.10) that checks whether a command to receive a telephone conversation is recognized so that in the event that the YES response is continued through the module (S2.11) that initiates a telephone conversation; that if the module NO (S2.10) has NO response, the process is continued via the module (S2.12), which allows the selected telephone number to be displayed, that the output of the module (S2.12) is tied to the module (S2.14) for Selecting the appropriate grammar for the module (S2.13) to execute the command - Make a call; for the module (S2.15) to execute the command - End the telephone conversation for the modules (S2.5 to S2.15) to perform the functions of the control processor device (1.4). Voice-guided telephone communication system according to Claims 1 and 2, characterized in that it allows the user to activate voice dialing of the telephone number or the called person by voice triggering via a predefined keyword a virtual START key connected via modules (S3 .1, S3.2, S3.3, S3.4) per module (S3.5) and represents a custom voice processor (1.5) for the module (S3.1) to verify that speech has been detected that the module (S3 .2) the captured speech signal is further preprocessed in order to reduce echoes, so that the module (S3.3) determines the characteristics of the speech signal and the module (S3.4) determines the spoken word in order for the module (S3.5) to determine the reliability of the recognized word; that if the answer is NO in module (S3.1) to the question whether speech has been detected, then the procedure returns to the START key, so that if the answer is YES at the output of module (S3.1), the process is carried on via modules (S3.2, S3.3, S3.4 and S3.5) to a module (S3.6) that verifies that the recognized word is sufficiently reliable; that in the case of NO, the process returns to the virtual START key, so that in the case of a YES response, the process resumes in a module (S3.7) that checks if a command is provided to enable a telephone connection to continue in the case of a YES response to Module (S3.15) that in the case of NO, the process continues in Module (S3.8), which checks whether the command to establish a telephone connection is recognized, that in the case of YES, the process continues in Module (S3.14), and module (S3.15) that in the case of NO in module (S3.8), the process continues in module (S3.9), which checks whether the command to re-enter the telephone number is recognized, so that in the case of YES the process continues to the module (S3.15), the output of which is connected to the START key, so that, in the case of NO module response (S3.9), the process continues in the module (S3.10), which checks whether the rejection command or. interruption of telephone connection that in case of YES answer the process continues in module (S3.16), whose output is bound to START button, that in case of NO answer in module S3.10 the process continues in module (S3.11) which checks whether the command to receive a telephone conversation is recognized so that in the case of a YES response, the process continues via the module (S3.12), which establishes a telephone conversation, that in the case of a NO response in the module (S3.11), the process continues through the module (S3 .13) that allows the selected telephone number to be displayed that the module output (S3.13) is bound to the module (S3.15) for Selecting the appropriate grammar that the module (S3.14) has a command - Make a call that the module ( S3.16) command End the telephone conversation that the modules (S3.6 to S3.16) perform the functions of the control processor device (1.4). 5. Voice-guided telephone communication system according to claim 1, characterized in that the system comprises a speech processing device (1.1), with a headset (1.3) and a microphone (1.2), which includes an automatic speech recognition system. Voice-guided telephone communication system according to claim 2, characterized in that the system comprises a customized voice processing device (1.5), with a loudspeaker (1.6) and a remote microphone (1.7), which includes an automatic speech recognition system. 7. Voice-guided telephone communication system according to claim 2, characterized in that the system comprises a custom voice-processing device (1.5), with a loudspeaker (1.6) and a remote microphone (1.7), which includes an additional preprocessing system for speech reduction. 8. Voice-guided telephone communication system according to claims 1 and 2, characterized in that the system comprises a control processor device (1.4) which includes a system for establishing a telephone connection via modern VolP telephone technologies or any digital telephone connection which 10 carries voice packets. 9. Voice-guided telephone communication system according to claims 1 to 8, characterized in that the speech processing device (1.1) and the adapted voice processing device (1.5) recognize the keywords. 10. The method of voice-guided telephone communication according to claims 1 to 9, characterized in that the voice processing device (1.1) and the adapted voice processing device (1.5) evaluate the correctness of the recognized hypotheses. 20. A method of voice-guided telephone communication according to claims 1 to 10, characterized in that the speech processing device (1.1) and the adapted speech processing device (1.5) allow the selection of recognition grammars to be selected on an ongoing basis to provide different points of dialogue.