"Prerana team on AssemblyAI - Voice Agent Hackathon Hackathon"
Team Idea
Team Prerana designs and deploys production-grade, low-latency voice AI agents built to handle real-time conversational streaming under high-pressure hackathon deadlines. We bridge multimodal language models, bidirectional audio pipelines, and edge tooling to create intuitive, human-like voice interfaces. Core Technical Stack & Real-Time Skills * Streaming Voice Architecture: Implemented duplex audio streaming using WebSockets and WebRTC, cutting turn-taking latency below 500ms for natural, interruptible conversations. * Pipeline Integration: Built end-to-end voice processing by chaining real-time Deepgram Speech-to-Text, Gemini Flash function calling for conversational logic, and Cartesia streaming Text-to-Speech. * Dynamic Tool Calling: Wired voice agents to external REST APIs and localized SQLite databases, allowing the assistant to trigger real-time actions, check live schedules, and fetch contextual data mid-sentence.
Team Members
.png&w=256&q=75)
.png&w=256&q=75)
.png&w=256&q=75)
Submission
Team Leader hasn't made a submission yet