P.E.T.E.

Inspiration

P.E.T.E. – AI Crisis Response System (Devpost Project Story)

Inspiration Every second matters in crisis response. Whether it's a suicide prevention hotline, emergency dispatch, or mental health support, human operators often face extreme stress while trying to provide the right response. We wanted to build a system that enhances, not replaces, human intervention, ensuring faster, more accurate, and emotionally intelligent responses.

Our team envisioned P.E.T.E. (Personal, Emergency, Technical, Expert)—an AI-driven assistant that listens, analyzes distress signals, and provides real-time response suggestions to crisis operators.

What It Does P.E.T.E. is an AI-powered assistant designed for emergency response teams. It uses natural language processing (NLP) and voice analysis to detect distress, key phrases, and emotional cues in real-time conversations. Key features include:

✅ Keyword Detection: Identifies crisis-related words to guide the operator's response. ✅ Live Speaker Differentiation: Distinguishes between the caller and the operator. ✅ Emotional Tone Analysis: Detects stress, urgency, and emotional distress. ✅ Real-Time Guidance: Provides AI-generated suggestions to help operators respond effectively.

How We Built It We integrated several cutting-edge technologies:

🔹 Ollama – AI model for generating responses. 🔹 WebRTC + VAD (Voice Activity Detection) – Real-time voice stream analysis. 🔹 Python (multi-threading & streaming) – Handles simultaneous calls. 🔹 Whisper AI – Speech-to-text conversion.

Challenges We Ran Into ⚠️ Real-time Audio Processing – Ensuring smooth, low-latency processing without lag. ⚠️ Accurate Emotion Detection – Fine-tuning AI models to differentiate stress, anger, and sadness. ⚠️ Multi-Speaker Separation – Handling overlapping conversations effectively.

Accomplishments We’re Proud Of 🚀 Successfully integrated real-time speech analysis + AI response generation. 🚀 Created a scalable, lightweight system that runs efficiently in crisis centers. 🚀 Built an operator-friendly interface to help responders act faster under pressure.

What’s Next for P.E.T.E.? 🔜 Multi-Language Support – Expanding accessibility for global crisis hotlines. 🔜 Machine Learning Feedback Loop – Improving accuracy based on real-world data. 🔜 Mobile & Cloud Deployment – Enabling broader reach and integration.

We believe P.E.T.E. can revolutionize emergency response, mental health support, and crisis intervention. By empowering human responders with AI, we can save more lives and provide better care when it’s needed most.