AI Powered Voicemail Prioritisation System

Abstract

Managing voicemails in critical sectors such as healthcare, finance, and customer service is often hindered by delays, inefficiencies, and missed urgent messages. This project introduces an AI-powered voicemail system  designed to intelligently capture, analyze, and prioritize voice messages, ensuring that high-urgency cases are addressed immediately. Using Twilio for voicemail capture and Google SpeechRecognition for transcription, the system combines text and audio analysis to assess both semantic content and vocal tone. BERT-based text classification evaluates urgency and routes messages to the appropriate department, while OpenSMILE and Random Forest models analyze acoustic features to detect stress, pitch, and speed, indicating urgency levels. PII is automatically redacted using SpaCy to maintain privacy and data security. By merging speech processing, NLP, and machine learning, the system offers a scalable, reliable, and efficient solution for modern communication management.