Browse Prior Art Database

Real-Time Detection of Emergencies via Cognitive Analysis of Audio Data Streams Disclosure Number: IPCOM000244763D
Publication Date: 2016-Jan-11
Document File: 3 page(s) / 75K

Publishing Venue

The Prior Art Database


Disclosed is a method and system for the real-time detection of emergencies via cognitive analysis of audio data streams. The novel system analyzes audio to detect stress-related vocal patterns and then sends alerts to the appropriate emergency response personnel.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 52% of the total text.

Page 01 of 3


-Time Detection of Emergencies via Cognitive Analysis of Audio Data Streams

Time Detection of Emergencies via Cognitive Analysis of Audio Data Streams

Delays in detecting emergencies lead to a slower response and consequently a higher risk of casualties. Emergency detection in crowded or isolated environments can be a difficult task for emergency responder staff .

An example of the problem is emergencies on board commuter trains . The driver has sole responsibly over hundreds of commuters. The driver's role is to not only drive the train but also quickly respond in the event of an emergency (e.g., a commuter collapses, a violent altercation, etc.) Currently, a driver is notified of an emergency

when a passenger engages the emergency button, often located close to the door. This system relies on other passengers on the train to activate the emergency response button, which can take time and there may arise situations when the emergency alarm cannot be manually activated.

Furthermore, no current systems use neural networks to assist in identifying an emergency/problem situation.

The proposed solution is to set up an array of microphones within a commuter train or like environment, and stream the recorded audio to a low-power computer device. The novel system analyzes the audio to detect stress-related vocal patterns (e.g., screaming, shouting, etc.). Other audio indications of an emergency might include shattering of windows, barking of dogs, (in the case of a burglary), loud crashes, banging, etc.

When the system detects an emergency , it notifies the driver or security office, activates the closed-circuit television (CCTV) system within the specific area (e.g., carriage on a train), and stores the recorded audio data. The system can also trigger a police/ambulance call to the next stop. This method off-loads the emergency detection responsibility from the driver and commuters while reducing the time to emergency response.

Because CCTV systems are widespread in ubiquitous in modern cities , this proposed solution could find applicability in many different locations at which security must be maintained.

The novel system comprises a combination of hardware and software components .

The mobile hardware platform includes:

• Microphones for capturing audio streams
• Processor: a combination of central processing unit (CPU), graphics processing unit (GPU) or other data processing device (e.g., Field Programmable Gate Array (FPGA), neuromorphic processor, etc.) for transforming audio signals into a dataset that can be analyzed via an automated system (e.g., neural network or other machine learning classifier)

• Communication network: a means of communicating an identified event , such as


Page 02 of 3

  a potential emergency, to the appropriate person or system. This communication may be wirel...