TWM598502U - Voiceprint monitoring device - Google Patents

Abstract

This creation is about a voiceprint monitoring device, which includes a voiceprint input module to detect and receive audio data; an audio database electrically connected to the voiceprint input module, which stores multiple voiceprint audio data , Such as coughing and restless distress voiceprint data; a processing unit that electrically connects the voiceprint input module and the audio database to determine whether the audio data obtained by the voiceprint input module is the same as the voiceprint data in the audio database If it meets, it will generate an output information; a result database to receive and store the output information of the processing unit; and a remote server to perform online monitoring and obtain the results of various event records; thereby, this creation can be real-time Monitor and compare quickly, determine whether it is coughing or panic calling for help and other restless sounds, and carry out appropriate notification and rescue.

Description

Voiceprint monitoring device

This creation is about a voiceprint monitoring device, especially a device that can monitor cough voiceprints and restless voiceprints for help in real time, so as to achieve the purpose of timely and appropriate notification, recording and rescue.

In non-open spaces, such as public transportation vehicles or elevators, there are few devices that can instantly recognize and interpret coughs or restless calls for help in the room environment.

For example, the "Voice Recognition Method, Electronic Device and Voice Recognition System" disclosed in the Republic of China Patent Publication No. I574255 has a voice recognition function, but the recognition and comparison need to be connected to a remote server for calculation. , Because the information transmission period takes time, so the opportunity for real-time monitoring and rescue will be missed.

In addition, Chinese Patent Publication No. CN109360584A discloses a "deep learning-based cough monitoring method and device", which samples a sound signal and obtains the characteristic parameter matrix of Mel frequency cepstral coefficients of the sound signal, and extracts the signal from it. Feature to confirm whether the signal feature matches the pre-acquired cough signal feature model based on the support vector data description algorithm. If it matches, the sound signal is confirmed as a cough sound; although the previous case can be cough monitoring, and the cough sound Sampling and other calculations are performed in the local arithmetic processing unit of the identification device, or uploaded to a cloud server, smart terminal or other server through the network for processing, but its structure is complex and can only monitor the cough state, and cannot monitor restless states such as panic and seeking help .

The reason is that, in view of the many deficiencies in the actual implementation of the existing voiceprint monitoring system, the creator improved and created a voiceprint monitoring device to quickly compare and interpret the restless sounds such as coughing or panic calling for help in various spaces. Appropriate notification, record and rescue purpose.

In order to achieve the above implementation objectives, the author proposes a voiceprint monitoring device, which includes: a voiceprint input module to detect and receive an audio data; an audio database, which is electrically connected to the voiceprint input module, It also stores multiple voiceprint audio data, including cough voiceprint data and restless distress voiceprint data; a processing unit electrically connects the voiceprint input module and the audio database to determine the audio obtained by the voiceprint input module Whether the data matches the cough voiceprint data and the restless distress voiceprint data in the audio database, if they match, an output information is generated; a result database is electrically connected to the processing unit to receive and store the output information; a remote The server is wirelessly connected to the audio database and processing unit; and a first communication unit is electrically connected to the processing unit and wirelessly connected to the remote server to transmit the output information stored in the result database to the remote server Device to report a cough event or/and a restless event to perform online monitoring and obtain various event record results; in addition, a second communication unit that is electrically connected to the remote server can be further provided to notify a cough event or / And a restless event to a plurality of information receivers.

In one embodiment of this creation, the processing unit is provided with a deep learning operation unit, which can use Convolutional Neural Networks, Recurrent Neural Networks or Reinforcement Learning.

In an embodiment of this creation, the voiceprint input module samples and transmits it to the processing unit, and the operation time of the processing unit is 1 s.

In an embodiment of this creation, the plurality of voiceprint audio data in the audio database is obtained in advance from a large remote database, and preferably, the audio database is obtained by the remote server through the first communication unit Update multiple voiceprint audio data by transmission.

In an embodiment of the present creation, a lens unit is further provided, which is electrically connected to the processing unit to perform photography and photography, and the photos or videos recorded in the event are transmitted from the processing unit to the remote server through the first communication unit.

In this way, this creation can instantly monitor coughs in closed spaces for public health safety notifications or the restless voice for help activates emergency notifications or records. Because it is directly judged or identified by the voiceprint monitoring device, it can be avoided Learn to use the network to explain the time-consuming and easy to miss rescue opportunities.

The purpose of this creation and its structural and functional advantages will be explained based on the structure shown in the following figure with specific embodiments, so that the review committee can have a deeper and specific understanding of this creation.

First of all, please refer to the first and second pictures. This creation proposes a voiceprint monitoring device, which can be used in non-open spaces such as MRT cars or elevators. It includes: a voiceprint input module (1 ) To detect and receive an audio data; an audio database (2), which is electrically connected to the voiceprint input module (1), and stores multiple voiceprint audio data, including cough voiceprint data and agitation for help Voiceprint data; a processing unit (3), which is electrically connected to the voiceprint input module (1) and the audio database (2), and is equipped with convolutional neural networks (Convolutional Neural Networks) and recurrent neural networks (Recurrent Neural Networks) or Reinforcement Learning (Reinforcement Learning) and other deep learning algorithm deep learning operation unit (31) to train and determine whether the audio data obtained by the voiceprint input module (1) matches the audio database (2) The cough voiceprint data in the data is consistent with the restless distress voiceprint data; a result database (4) is electrically connected to the processing unit (3) to receive and store the output information; and a remote server (5) is The first communication unit (6) is wirelessly connected to perform online monitoring and obtain various event record results.

Furthermore, the processing unit (3) is further electrically connected to a first communication unit (6) to report a cough event or/and a restless event. A plurality of voiceprint audio data in the audio database (2) is obtained in advance from a large remote database, and can be updated by the remote server (5) through wireless transmission through the first communication unit (6).

In addition, as shown in the first figure, the voiceprint monitoring device can be further provided with a lens unit (7) electrically connected to the processing unit (3). When the event is detected, it will synchronize to take pictures and photos and record the event. The photos or videos of the processing unit (3) are sent to the remote server (5) through the first communication unit (6).

As shown in the second figure, the event record sent by the voiceprint monitoring device to the remote server can be transmitted to a plurality of information receiving ends by electrically connecting the second communication unit (8) of the remote server (5). Such as mobile devices, image storage devices, real-time image devices, and notification hosts, where the notification host can be further equipped with a buzzer for real-time alarm display function.

When the voiceprint monitoring device according to the above-mentioned preferred embodiment is actually implemented and used, taking the MRT train as an example, the voiceprint monitoring device is first installed in a suitable position in the cabin. The voiceprint monitoring device includes a voiceprint input module (1 ), an audio database (2), a processing unit (3) with a deep learning arithmetic unit (31), a result database (4), a first communication unit (6) and a lens unit (7).

The voiceprint input module (1) can continuously detect and receive a piece of audio data and send it to the audio database (2) for comparison with the processing unit (3). The audio database (2) has been pre-installed by the remote large database Acquire and store multiple voiceprint audio data, including cough voiceprint data and restless distress voiceprint data; after the processing unit (3) receives the audio data, it will contact the cough voiceprint data and restless distress in the audio database (2) The voiceprint data is compared with the voiceprint to determine whether the audio data obtained by the voiceprint input module (1) matches the audio database (2), and if so, an output information is generated to the result database (4), The result database (4) can receive and store output information, and send the output information stored in the result database (4) to the remote server (5) through the first communication unit (6), perform online monitoring and obtain Record the results of various events.

The lens unit (7) electrically connected to the processing unit (3) can also be used for photography and photography, to record the cough and restlessness of the current photos or videos for reference and confirmation, and the processing unit (3) and the first The communication unit (6) is sent to the remote server (5), and is notified by the second communication unit (8), which can, for example, send an alarm message to the captain's room, guard room or ambulance unit. When using this device to sample audio data, the voiceprint input module (1) samples and transmits it to the processing unit (3). The processing unit (3) takes about 1 s to work, so it can quickly identify and judge whether there is a cough Or restless calls for help to provide appropriate rescue in a short time.

In this way, this creation is used in non-open spaces such as MRT cars or elevators, and directly calculates and recognizes different voiceprints in a short time, without uploading voiceprint information to the Internet for calculations. Each space compares and interprets agitated sounds such as coughing or panic calling for help, so appropriate notifications, records and rescues can be made quickly.

In summary, the voiceprint monitoring device of this creation can indeed achieve the expected use effect through the embodiments disclosed above, and this creation has not been disclosed before the application, since it is in full compliance with the provisions of the patent law and Claim. If you file an application for a new type patent in accordance with the law, you are kindly requested to review it and grant a quasi-patent.

However, the above-mentioned illustrations and descriptions are only the preferred embodiment of this creation, and are not intended to limit the scope of protection of this creation. For those who are familiar with the art, the characteristics of the creation based on the scope of this creation, and others Equivalent changes or modifications should be regarded as not departing from the design scope of this creation.

1: Voiceprint input module

2: Audio database

3: Processing unit

31: Deep learning operation unit

4: Results database

5: Remote server

6: The first communication unit

7: lens unit

8: The second communication unit

Figure 1: The system block diagram of a preferred embodiment of this creation.

Figure 2: The system block diagram of another preferred embodiment of this creation.

1: Voiceprint input module

2: Audio database

3: Processing unit

31: Deep learning operation unit

4: Results database

5: Remote server

6: The first communication unit

7: lens unit

Claims (8)

A voiceprint monitoring device includes: A voiceprint input module to detect and receive audio data; An audio database, which is electrically connected to the voiceprint input module, and stores multiple voiceprint audio data, including cough voiceprint data and restless distress voiceprint data; A processing unit is electrically connected to the voiceprint input module and the audio database to determine whether the audio data obtained by the voiceprint input module is the same as the cough voiceprint data and the agitation in the audio database The voiceprint data matches, and if they match, an output message is generated; A result database, which is electrically connected to the processing unit to receive and store the output information; and A remote server wirelessly connects the audio database and the processing unit to perform online monitoring and obtain various event record results. The voiceprint monitoring device according to claim 1, further comprising a first communication unit electrically connected to the processing unit and a second communication unit electrically connected to the remote server to report a cough event or/and A restless event. The voiceprint monitoring device according to claim 1, wherein the processing unit is provided with a deep learning operation unit. The voiceprint monitoring device according to claim 3, wherein the deep learning operation unit uses Convolutional Neural Networks, Recurrent Neural Networks or Reinforcement Learning. The voiceprint monitoring device according to claim 1, wherein the operation time of the processing unit is 1 s. The voiceprint monitoring device according to claim 1, wherein the plurality of voiceprint audio data of the audio database is obtained in advance from a remote large database. The voiceprint monitoring device according to claim 6, wherein the audio database updates the plurality of voiceprint audio data through a remote server. The voiceprint monitoring device according to claim 1, further comprising a lens unit electrically connected to the processing unit for photography and photography.

Images