Speech Analysis for 9-1-1 Calls and Surveillance Audio

This research project has two goals. The first goal is to design and test software that will analyze the content of transcribed 9-1-1 emergency calls. We will use Natural Language Processing (NLP) and Artificial Intelligence (AI) to perform an evaluation of the operator performance during the 9-1-1 call and generate an evaluation form in the format required by the Alberta government. The second goal is to improve existing software that automatically transcribes long surveillance audio clips into English text. Whisper, a new AI-based Speech Recognition package from OpenAI, will be applied to segments of calls and then adapted to better handle unique features of surveillance audio, such as noisy backgrounds, unintelligible speech, conversational speech, numeric values, and code words. The research project directly addresses the Edmonton Police Service’s two most pressing needs for cheaper transcription that cannot be met by existing software.

Faculty Supervisor:

Howard Hamilton

Student:

Partner:

City of Edmonton

Discipline:

Computer science

Sector:

Administrative and support, waste management and remediation services; Public administration

University:

University of Regina

Program: