9 DISC

VOICE RECOGNITION SOFTWARE TO MITIGATE CYBER FRAUDS

Current Scenario: Many of the cyber frauds happen over voice communication. Fraudsters communicate with potential victims either through GSM or VoIP calls. On several occasions the victim is able to record the voice of the fraudsters. The voice sample has certain noise and distortions in it. However, this is the only identifier with which the crime or attempted crime can be connected to the criminal. The goal of this project is to design and develop a Voice Recognition System (VRS) that can assist Law Enforcement Agencies (LEAs) in identifying and tracking individuals who have committed or attempted to commit cyber fraud through voice communication. The VRS will be used to analyze voice samples that have been recorded by victims of cyber fraud, which may contain background noise and distortions. The VRS will be able to store voice samples of different individuals of interest that can be uploaded by LEAs and match them with test samples in order to identify suspects. The system should have the following features: The ability to store voice samples of different individuals of interest, along with additional information that can be imported through an API from another system. The ability to group voice samples based on their similarity. The ability to match individual test samples with the samples available in the database, and display near matches in decreasing order of similarity. The ability to eliminate background noise and other distortions from the voice samples. A high level of voice fingerprinting and matching accuracy to ensure reliable results. This system will help in identifying the criminals and can be used as a evidence in the court of law.

Challenges