Vocat — AI Interview Agent

A real-time AI interview agent with a Google Meet-style interface. Upload your resume, join the call, and the AI conducts a live voice interview — no typing, no lag, just a natural conversation.

How It Works

1. Upload your resume — drag & drop or click to upload (PDF/TXT)
2. Join the call — the AI interviewer greets you and starts asking questions based on your resume
3. Talk naturally — voice activity detection handles turn-taking automatically
4. Get interviewed — behavioral questions, technical deep-dives, follow-ups — just like a real interview

Architecture

Mic → WebRTC → VAD → Whisper STT → GPT-4o → ElevenLabs TTS → WebRTC → Speaker

• WebRTC for full-duplex, low-latency audio streaming
• Voice Activity Detection for automatic turn-taking (no push-to-talk)
• Whisper for speech-to-text
• GPT-4o with streaming for fast, context-aware responses
• ElevenLabs Turbo for natural text-to-speech
• Sentence-level TTS — starts speaking before the full response is generated

Tech Stack

Layer	Tech
Frontend	React + Vite + Tailwind CSS
Transport	WebRTC (aiortc)
VAD	webrtcvad
STT	OpenAI Whisper
LLM	GPT-4o (streaming)
TTS	ElevenLabs (turbo v2.5)
Backend	Python + aiohttp

Setup

Prerequisites

• Python 3.10+
• Node.js 18+
• FFmpeg
• OpenAI API key
• ElevenLabs API key

Install

# Clone
git clone https://github.com/modelpath-dev/Vocat.git
cd Vocat

# Python dependencies
pip install -r requirements.txt

# Frontend dependencies
cd frontend && npm install && cd ..

# Environment variables
cp .env.example .env
# Edit .env with your API keys

Run

# Terminal 1 — Backend
export $(cat .env | xargs) && python vocatl2_backend.py

# Terminal 2 — Frontend
cd frontend && npm run dev

Open http://localhost:5173 in your browser.

Configuration

VAD

VAD_AGGRESSIVENESS = 3        # 0-3, higher = less sensitive to background noise
VAD_SILENCE_TIMEOUT_MS = 600  # ms of silence before processing speech

Voice

Change the voice in vocatl2_backend.py:

voice_id="EXAVITQu4vr4xnSDxMaL"  # Sarah — see ElevenLabs voice library

Interview Prompt

Edit prompts/interviewer.md to customize the interviewer's behavior, question style, or focus areas.

Project Structure

vocat/
├── vocatl2_backend.py       # WebRTC server + AI pipeline
├── frontend/
│   ├── src/
│   │   ├── App.jsx          # Google Meet-style UI
│   │   ├── hooks/
│   │   │   └── useWebRTC.js # WebRTC client hook
│   │   └── index.css
│   ├── vite.config.js
│   └── package.json
├── prompts/
│   └── interviewer.md       # Interview system prompt
├── .env.example
├── requirements.txt
└── README.md

License

MIT