DESC:FIELD OF THE INVENTION

The present invention relates to speech recognition security system and more particularly, to an offline speech recognition security system and a method for detecting, processing and reporting emergency situation at the installed location such as home, offices, shops, banks, ATMs and the like for providing security to the people.

BACKGROUND OF THE INVENTION

The use of security systems to protect people has seen an increasing demand. The basic idea of a security system is sensing a predetermined parameter and delivering a response upon changes in the value of that parameter. However, the security systems available today are mainly related to indicating existence of an unacceptable condition, such as an open door which should be closed, the presence of an intruder, or a condition of an emergency nature (such as fire or flood). Such security systems employ many types of sensors such as motion sensors for both the indoors and outdoors, glass-break detectors, magnetic contact sensors, flood detectors, temperature sensors, smoke detectors, sensitized door mats, and the like.

However, the security systems available currently do not help people in panic situation. The systems of the prior art provide alerts when something wrong happens to a property such as burglary and the like. But when it comes to securing life of individual by providing help during burglary, attack by thieves or during emergency situation such as heart attack or the like, the security systems of the prior art of no use.

Hence, there exists a need to provide a system and a method for providing security or help to people during emergency situation.

OBJECTS OF THE INVENTION

An object of the present invention is to provide help to everyone in emergency situations.

Another object of the present invention is to provide a simple, cost effective, offline speech recognition security system for securing people in an emergency situation.

SUMMARY OF THE INVENTION

Accordingly, in one aspect, the present invention provides a speech recognition security system. The speech recognition security system is an offline speech recognition system that comprises an input device, a speech recognition module, a controller unit, an alarm unit, an alarm termination unit, a communication unit, an audio playback unit, at least one external sensor, a power supply unit and a battery backup.

The input device is positioned in a premise for receiving speech/voice signals of persons/individuals. The input device is at least one microphone selected from any one of a condenser and a micro electrical-mechanical system (MEMS). The at least one microphone is selected from any one of a wired microphone and a wireless microphone.

The speech recognition module is disposed in operable communication with the input device for continuously monitoring and analyzing the speech signals received therefrom on a real time basis to recognize/detect predefined trigger words contained therein in a predefined language and generate a pulse in response thereto. The speech recognition module includes language selection functionality for selecting different languages, and trigger level selection functionality for controlling the accuracy/level of the detection of the trigger words.

The controller unit is disposed in operable communication with the speech recognition module to be activated by the pulse.

The alarm unit is activated by the controller unit in response to the pulse generation. The alarm unit includes an audio device and an imaging device. The audio device is adapted to alert neighbors or a security guard of the emergency situation. The audio device is selected from any one of a speaker, a siren, a wireless information system for example Wi-Fi modules, Bluetooth units and ZigBee units. The imaging device for example a camera is adapted to record the emergency situation until the termination of the alarm unit.

The alarm termination unit is interfaced with the controller unit for deactivating the alarm unit within a time period of ‘X’ time to avoid sending false alarms.

The communication unit is activated by the controller unit in response to the pulse generation and in absence of the termination of the alarm unit after end of time period of ‘X’ to inform the users on their registered mobile numbers about the emergency situation via messages/SMS and voice calls.

The audio playback unit is configured to record and store voice messages by the users for being played by the communication unit upon activation by the controller unit.

The at least one external sensor is monitored by the controller unit to detect an intrusion of the premises. The at least one external sensor is selected from any of a motion sensor, a switch, a magnetic door sensor, a door bell and a combinations thereof. The power supply unit is configured to distribute power and the battery backup is configured to distribute power in case of mains failure.

In another aspect, the present invention provides an offline speech recognition security method.

BRIEF DESCRIPTION OF THE DRAWING

The objectives and advantages of the present invention will become apparent from the following description read in accordance with the accompanying drawing wherein,
Figure 1 shows a block diagram of a speech recognition security system, in accordance with the present invention;

Figure 2 shows a flowchart illustrating the working of a speech recognition module of the speech recognition security system, in accordance with the present invention;

Figure 3 shows a flowchart illustrating the working of a controller unit of the speech recognition security system, in accordance with the present invention; and

Figure 4 shows a flow chart of a communication unit of the speech recognition security system, in accordance with the present invention.

DETAILED DESCRIPTION OF THE INVENTION

The foregoing objects of the present invention are accomplished and the problems and shortcomings associated with the prior art are overcome by the present invention as described below in the preferred embodiment.

The present invention provides an offline speech recognition security system and a method which is used for detecting, processing and reporting an emergency situation at the installed location. The method includes a standalone, offline continuous speech recognition system that determines whether present situation is an emergency situation or not based on recognized command; and controls a wireless communication unit to inform external world. The speech recognition security system of the present invention is simple, cost effective security system for securing people in a panic situation or in a medical emergency. For example, if a person is alone and intruder enters in the home or if an old individual gets a heart attack, then by just saying the predefined words such as “Bachao Bachao Madat Karo” the speech recognition security system will trigger an alarm system and a communication unit.

The present invention is illustrated with reference to the accompanying drawing, throughout which reference numbers indicate corresponding parts in the various figures.

Referring now to figure 1, in an aspect, a speech recognition security system (100) in accordance with the present invention is shown. Specifically, the speech recognition security system (100) is an offline speech recognition security system. The speech recognition security system (100) (hereinafter “the system (100)”) comprises an input device (10), a speech recognition module (20), a controller unit (30), an alarm unit (40), an alarm termination unit (50), a communication unit (60), an audio playback unit (70), at least one external sensor (80), a power supply unit (90) and a battery backup (95) as shown in figure 1.

The input device (10) is placed at a convenient location within a premise, for example, a house for receiving / capturing speech / voice signals of persons / individuals from external word who require help in an emergency situation. In an embodiment, the input device (10) is at least one microphone selected from any of a condenser and a micro electrical-mechanical system (MEMS). However, it is understood that other types of microphone known in the art can also be used as per various alternate embodiments of the present invention. The at least one microphone is selected from any one of a wired microphone and wireless microphone. Further, in order to detect the voice from multiple rooms in one apartment, an array of microphones can be used.

The speech recognition module (20) is disposed in operable communication with the input device (10) for continuously monitoring and analyzing the speech signals received as an input therefrom on a real time basis to recognize/detect predefined trigger words / phrases contained therein in a predefined language and generate a pulse in response thereto. Once the system (100) is ON, the speech recognition module (20) starts receiving and analyzing all the speech signals of the surrounding environment and generates a trigger pulse (herein after ‘the pulse’) immediately after finding the trigger word in the speech signals. In an embodiment, pulse is a logical high pulse at GPIO pin which is connected to an input of the controller unit (30). The speech recognition module (20) includes a speech recognition IC equipped with a trigger word detection algorithm, language selection functionality (12), and trigger level selection functionality (14) configured therein.

The speech recognition IC equipped with the trigger word detection algorithm is adapted to recognize the words by analyzing more than 100 samples of trigger word or words of different people of different age group, which are commonly uttered during panic situation. These samples are tuned to word and linked with corresponding vowel so that the speech recognition module (20) only recognizes the predefined word of any tone, and pitch and intensity. The speech recognition module (20) is configured with at least three levels of sensitivity depending on person’s voice and surrounding environment. The user can select the sensitivity. Further, the speech recognition module (20) is independent of the user such that any person who says the trigger word can alarm the system (100).

The language selection functionality (12) allows the user to select a desired language from different languages in which the phrase/ trigger word is recorded. In an embodiment, the speech recognition module (20) is configured to recognize the trigger words “bachao bachao, madat karo” in Hindi speaking region. However, it is evident to those skilled into art to configure the speech recognition module (20) to recognize trigger words in their preferred language, such as English, Marathi and the like.

The trigger level selection functionality (14) allows controlling the accuracy / level of the detection of the trigger words. In other words, the trigger level selection functionality (14) gives us features to control the accuracy of the detection level of the speech recognition module (20).

The controller unit (30) is disposed in operable communication with the speech recognition module (20) for being activated by the pulse. The controller unit (30) includes a trigger processing module uploaded to an integrated circuit configured within the system (100). Besides controlling language selection and trigger level, the controller unit (30) is also used for controlling all other units such as monitoring the at least one external sensor (80), turning ON the alarm unit (40), controlling the communication unit (60) and turning on the audio playback unit (70). Thus, the activated controller unit (30) processes the generated pulse to activate the alarm unit (40) and the communication unit (60).

The alarm unit (40) alternately referred as an emergency alert unit is activated by the controller unit (30) in response to the pulse generation. The alarm unit (40) includes an audio device (not shown) and an imaging device (not shown). The audio device is selected from any one of a speaker, a siren or wireless information system e.g. Wi-Fi modules or Bluetooth units or ZigBee units or any other alarming device. An example of the imaging device is a camera. Whenever there is detection of the trigger word at the speech recognition module (20), the controller unit (30) activates the alarm unit (40) such that the audio device for example the siren blares to alert the neighbour or a security guard of the panic/unhealthy/ emergency situation of the individual, and the camera starts to record the current emergency environment until the alarm unit (40) is terminated. The alarm unit (40) is terminated by any of the alarm termination unit (50) in case of false alarms or by manually in case of emergency situations.

The alarm termination unit (50) is interfaced with the controller unit (30) using wired or wireless channels to deactivate or terminate the alarm unit (40) to avoid sending false alarms.

The communication unit (60) such as a messaging system for example a Global System for Mobile Communications (GSM) unit is activated by the controller unit (30) in response to the pulse generation and in absence of the termination of the alarm unit (40) after end of time period of ‘X’ to inform the users of the emergency situation on their registered mobile numbers via messages/SMS and thereafter via voice calls so that immediate help can be provided to the person in need.

The audio playback unit (70) is configured as an add-on feature to allow the users to record and store their voice message which will be played in case of premise intrusion. In an embodiment, the user can store a voice message of 30 seconds which will be played over the voice call made by the communication unit (60) upon activation by the controller unit (30).

The at least one external sensor (80) (herein after ‘the external sensor (80) for example a motion sensor, a switch, a magnetic door sensor, a door bell and the like is configured for being monitored by the controller unit (30) to detect an intrusion of the premises/ room. If there is a change in the motion in an armed situation or when the door bell rings, the alarm will switch the recognition level to more acceptance mode.

The power supply unit (90) is configured to distribute power to run the entire unit of the system (100). The power supply unit (90) is used to regulate external AC supply to regulated DC voltage and the supply is distributed to all the units. The battery backup (95) is configured to distribute power in cases of power / mains failure to keep the system (100) running.

Referring now to figures 2 to 4, in another aspect, a speech recognition security method in accordance with the present invention is shown. Specifically, the speech recognition security method (hereinafter “the method”) is an offline method which is described herein below in conjunction with the system (100) of figure 1. The method starts when the system (100) is turned ON.

Figure 2 shows the detailed flow chart to illustrate the working of the speech recognition module (20) from steps (301) to (307). At step (301), the speech recognition module (20) checks the GPIO pin to select language for the speech recognition using the language selection functionality (12). Once the language is selected, the speech recognition module (20) selects the recognition level using the trigger level selection functionality (14) at step (302). The lower numbered level will have more rejecting and have the lowest false accept rates while the higher numbered level will be more accepting and have lower false reject rates. Whenever the system (100) is turned ON by default, it will select the predefined level by the user. At step (303), the method involves turning on the speech recognition process (303). As soon as the speech recognition process starts, first it checks if there is any interruption from the controller unit (30) at step (304) in response to detection of intrusion of premise by the external sensor (80). If yes, then detection level is switched to more acceptance mode at step (305). If interruption is not detected, then it continues to the execution and checks for detection of trigger word by the speech recognition module (20) at step (306). This loop between the steps (306) and (304) stays unless and until there is any trigger word detection. At step (307), the speech recognition module (20) informs the controller unit (30) of the detected trigger word by sending the pulse thereto so that the further situations can be handled. Thereafter, the speech recognition module (20) again goes to the speech recognition process of step (303).

Figure 3 shows the flow chart to illustrate the working of the controller unit (30) from steps (401) to (412). When the system (100) is ON/ boots up, the controller unit (30) turns on to set the speech recognition language at step (401). This language selection can be done by selecting GPIO pins between speech recognition module (20) and the controller unit (30). Based on this pins only the speech recognition functionality (12) selects its language at step (401). At step (402), the controller unit (30) sets/configures the detection level of the speech recognition module (20) using the trigger level selection functionality (14). By default it is set to higher level by the controller unit (30) on start. At step (403), the speech recognition process is turned ON. Once the speech recognition module (20) is configured and turned ON, the controller unit (30) starts its own execution in parallel to execute step (404) which checks if there are any interruptions from the external sensor (80) such as a door bell. If there is a ringing of door bell, the controller unit (30) sends the interrupt signal to the speech recognition module (20) at step (405) switching it to more acceptance mode for some time. In absence of interruptions, the controller unit (30) continues its execution and checks whether there is any interrupt from other external sensors (80) for e.g. motion sensor, magnetic door switch and so on which are used in intrusion control at step (406). In such case, the controller unit (30) turns ON the alarm unit (40) at step (407). After this condition is checked, the controller unit (30) checks for any interrupt generated by the speech recognition module (20) at step (408). The loop between steps (408) to (404) keeps on executing unless and until there is detection of the trigger word at the speech recognition module (20) or there is intrusion at the controller unit (30). When controller unit (30) receives the pulse due to detection of trigger word, it turns ON the alarm unit (40) at step (409). Thereafter, it waits for some time so that if it is a false alarm, the situation can be terminated at step (410) using the alarm termination unit (50). This waiting time is set by the value of “X” variable at step (411). The controller unit (30) waits till the count is reached; by the time if no one terminates the alarm unit (40) then eventually it turns on the communication unit (60) at step (412).

Figure 4 shows the flow chart of the communication unit (60) from steps (501) to (504). When there is no termination of the alarm unit (40), the controller unit (30) turns on the communication unit (60) and loads the list of the registered contact numbers for emergency situation into own memory. Once the contacts are loaded, the communication unit (60) sends SMS/ SOS/ Predefined message to the users/ all emergency contacts at step (501). All this communication between the communication unit (60) and the controller unit (30) is done using AT commands. Once the SMS is sent, the communication unit (60) starts to make the voice call for example GSM voice calls to the users at step (502). When the call gets established, the controller unit (30) turns on pre-recorded voice message over the cellular line at step (503). If there is no termination of the system (100), the controller unit (30) keeps the siren at step (505) till there is manual turning off or there is termination using the alarm termination unit (50) to end the method.

ADVANTAGES OF THE INVENTION

1. The system (100) and the method are simple and cost effective.
2. The system (100) and the method provide help to everyone in emergency situations.

The foregoing objects of the invention are accomplished and the problems and shortcomings associated with prior art techniques and approaches are overcome by the present invention described in the present embodiment. Detailed descriptions of the preferred embodiment are provided herein; however, it is to be understood that the present invention may be embodied in various forms. Therefore, specific details disclosed herein are not to be interpreted as limiting, but rather as a basis for the claims and as a representative basis for teaching one skilled in the art to employ the present invention in virtually any appropriately detailed system, structure, or matter. The embodiments of the invention as described above and the methods disclosed herein will suggest further modification and alterations to those skilled in the art. Such further modifications and alterations may be made without departing from the spirit and scope of the invention.
,CLAIMS:WE CLAIM:

1. A speech recognition security system (100) comprising:
• an input device (10) positioned in a premise for receiving speech/voice signals of persons/individuals;
• a speech recognition module (20) disposed in operable communication with the input device (10) for continuously monitoring and analyzing the speech signals received therefrom on a real time basis to recognize/detect predefined trigger words contained therein in a predefined language and generate a pulse in response thereto, the speech recognition module (20) having a language selection functionality (12) and a trigger level selection functionality (14) configured therein, the language selection functionality (12) adapted for selecting different languages and the trigger level selection functionality (14) adapted for controlling the accuracy/level of the detection of the trigger words;
• a controller unit (30) disposed in operable communication with the speech recognition module (20) to be activated by the pulse;
• an alarm unit (40) activated by the controller unit (30) in response to the pulse generation, the alarm unit (40) having an audio device to alert neighbours or a security guard of the emergency situation, and an imaging device to record the emergency situation until the termination of the alarm unit (40);
• an alarm termination unit (50) interfaced with the controller unit (30) for deactivating the alarm unit (40) within a time period of ‘X’ time to avoid sending false alarms;
• a communication unit (60) activated by the controller unit (30) in response to the pulse generation and in absence of the termination of the alarm unit (40) after end of time period of ‘X’ to inform the users on their registered mobile numbers about the emergency situation via messages/SMS and voice calls;
• an audio playback unit (70) configured to record and store voice messages by the users for being played by the communication unit (60) upon activation by the controller unit (30);
• at least one external sensor (80) monitored by the controller unit (30) to detect an intrusion of the premises;
• a power supply unit (90) configured to distribute power; and
• a battery backup (95) configured to distribute power in case of mains failure.

2. The speech recognition security system (100) as claimed in claim 1, wherein the input device (10) is at least one microphone selected from any of a condenser and a micro electrical-mechanical system, the at least one microphone being selected from any one of a wired microphone and wireless microphone.

3. The speech recognition security system (100) as claimed in claim 1, wherein the audio device of the alarm unit (40) is selected from any one of a speaker, a siren, a wireless information system for example Wi-Fi modules, Bluetooth units and ZigBee units, and the imaging device is a camera.

4. The speech recognition security system (100) as claimed in claim 1, wherein the at least one external sensor (80) is selected from any of a motion sensor, a switch, a magnetic door sensor, a door bell and a combinations thereof.

5. The speech recognition security system (100) as claimed in claim 1 is an offline speech recognition security system.

6. A speech recognition security method comprising:
• selecting a speech recognition language using a language selection functionality (12) of a speech recognition module (20), and selecting a speech recognition level by a trigger level selection functionality (14) of the speech recognition module (20);
• turning a speech recognition process, wherein the speech recognition module (20) continuously monitors and analyzes the speech signals captured by an input device (10) on a real time basis to check for any interruption from a controller unit (30) due to detection of premises intrusion by at least one external sensor (80);
• switching the speech recognition level to a more acceptance mode to recognize/detect predefined trigger words contained in the speech signal in a predefined language by the speech recognition module (20) depending on the states of the at least one external sensor (80);
• generating a pulse by the speech recognition module (20) in response to the detected trigger words to activate the controller unit (30);
• activating an alarm unit (40) by the controller unit (30) in response to the pulse generation, wherein the alarm unit (40) having an audio device to alert neighbours or a security guard of the emergency situation of the individual, and an imaging device to record the emergency situation until the termination of the alarm unit (40);
• waiting for a time period of ‘X’ to allow an alarm termination unit (50) to turn off the alarm unit (40) activated in case of false alarm situation;
• turning on the communication unit (60) by the controller unit (30) in absence of the termination of the alarm unit (40) after end of time period of ‘X’; and
• informing users on their registered mobile numbers about the emergency situation via messages/SMS and thereafter via voice calls by playing a pre-recorded voice message stored in an audio playback unit (70) by the communication unit (60),
wherein, the speech recognition security method runs on power distributed by any of a power supply unit (90) and a battery backup (95), wherein the battery backup (95) being configured to distribute power in case of mains failure.

7. The speech recognition security method as claimed in claim 5, wherein the input device (10) is at least one microphone selected from any of a condenser and a micro electrical-mechanical system, the at least one microphone being selected from any one of a wired microphone and wireless microphone.

8. The speech recognition security method as claimed in claim 5, wherein the audio device of the alarm unit (40 is selected from any one of a speaker, a siren, a wireless information system for example Wi-Fi modules, Bluetooth units and ZigBee units, and the imaging device is a camera.

9. The speech recognition security method as claimed in claim 5, wherein the at least one external sensor (80) is selected from any of a motion sensor, a switch, a magnetic door sensor, a door bell and a combinations thereof.

10. The speech recognition security method as claimed in claim 5 is an offline speech recognition security method.

Dated this 16th day of May 2017

Priya Pandey
(Agent for Applicant)
IN/PA-2018