Voice-based data collection for crash reports



The Invention:

Our speech recognition framework for collection of crash reports enables officers to collect data as they speak to a voice recording system instead of the manual method. The framework integrates state-of-the-art AI techniques for Speech Recognition and Natural Language Processing proven to perform similarly or even better than humans on certain operations with texts (e.g., reading, writing, understanding, and answering questions). The semi-automated data collection platform that uses a voice-recording method will allow officers to naturally describe the scene by speaking to the voice recording system. This will then support and enhance accurate and quick operation of police officers and thus allow more comprehensive and error-free data sets for traffic engineering practices, and smooth and safe road experience for the public. In addition, this will be a critical step forward in improving the overall traffic safety management system and enhancing the welfare of crash data collectors. 


  • Enhances state-of-the-art practices for collection of crash reports by minimizing the time and effort required for this task
  • Performed with on-going traffic that presents hazards while managing the crash scene
  • Resolves quality of collected data, erroneous data including inaccurate identification of crash factors, misclassified severity levels, and incomplete data. 

Market Opportunity: 

The law enforcement software market, which refers to technologies meant to assist law enforcement in their operations, is currently valued at 14.9B in 2022 and projected to grow at a CAG of 8.4% between 2023 to 2032 (Acumen Research and Consulting).

Crash reports are an essential source of data for identification and prioritization of traffic safety countermeasures. The data set in a crash report comprises several pages of information about the scene, vehicles, parties involved in the crash, and other necessary data. Collecting all this data in a crash site requires substantial effort and time in addition to those required to manage the accident, victims, and on-going traffic. Later on, this difficulty contributes to errors, missing information, and incorrect data entries.

Intellectual property

Early Stage


Patent Information:
For Information, Contact:
Kegan Mcmullan
Licensing Manager
University of Nevada, Las Vegas
Jee Woong Park
Cristian Arteaga-Sanchez
Computer Science & Software
Computer Science & Software - Apps
Engineering & Manufacturing - Transportation
© 2024. All Rights Reserved. Powered by Inteum