What I Built
AI Voice Interview Simulator is an intelligent, voice-enabled mock interview platform that helps job seekers practice interviews with real-time feedback, emotion analysis, and performance tracking.
Most interview preparation platforms today are text-based and fail to simulate real conversational pressure or provide actionable feedback on communication skills.
This project solves that by creating a voice-based AI interview experience that:
- conducts real-time mock interviews
- asks adaptive questions based on responses
- evaluates interview performance
- analyzes emotional signals
- tracks progress over time
The goal was to build a more realistic and interactive interview preparation platform rather than a static chatbot experience.
Features
🎤 Voice Interview Simulation
- AI asks questions using text-to-speech
- Users can answer using their microphone
- Automatic speech transcription using Whisper
🧠 Dynamic AI Questioning
The interview flow adapts based on previous answers.
The platform supports:
- General interviews
- Technical interviews
- HR interviews
- Sales interviews
Difficulty levels:
- Entry Level
- Mid Level
- Senior Level
The system also supports resume-based personalized interview generation.
📊 Real-Time Performance Scoring
After every response, the platform evaluates:
- Clarity
- Confidence
- Relevance
Users receive instant AI-generated feedback after each answer.
🎭 Emotion Analysis
The system analyzes emotional signals from candidate responses and detects:
- confidence
- nervousness
- enthusiasm
- hesitation
This helps users understand both technical and communication performance.
📈 Interview History Dashboard
The platform stores previous interview sessions and allows users to:
- review past interviews
- track score improvements
- analyze trends over time
- revisit previous feedback
🔄 Resume Upload Support
Users can upload resumes in PDF format.
The AI then generates personalized interview questions based on:
- skills
- projects
- experience
- technologies mentioned in the resume
Demo
🌐 Live Applications
Prototype Version
https://ai-interview-simulator-prototype-bausbh9a9dsxk9e5uvws8w.streamlit.app/
Full Application
https://ai-interview-simulator-web.streamlit.app
📂 GitHub Repository
https://github.com/KhushiSingla-tech/ai-interview-simulator
🎥 Demo Video
tys-tdxs-dvd (2026-04-28 20_17 GMT+5_30).mp4 - Google Drive
drive.google.com
Tech Stack
| Component | Technology |
|---|---|
| Frontend | Streamlit |
| Workflow Automation | n8n |
| LLM | Groq (llama-3.1-8b-instant) |
| Voice Output | Lemonfox TTS |
| Voice Input | Lemonfox Whisper STT |
| Emotion Analysis | Groq LLM |
| Database | Supabase (PostgreSQL) |
| Backend Deployment | Railway |
| Frontend Deployment | Streamlit Cloud |
Architecture
Workflow 1 — AI Interview
- Receives user answer and conversation history
- Sends context to AI workflow
- Generates next interview question dynamically
Workflow 2 — Answer Scorer
- Receives question and candidate answer
- Scores clarity, confidence, and relevance
- Returns structured feedback
How It Works
User Starts Interview
↓
AI Generates Question
↓
User Answers via Voice/Text
↓
Speech-to-Text Processing
↓
AI Evaluation + Emotion Analysis
↓
Feedback + Next Question
↓
Session Stored in Database
Challenges Faced
Maintaining Conversational Flow
One challenge was preserving enough interview context so that follow-up questions felt natural and relevant.
Real-Time Voice Processing
Handling speech-to-text conversion while maintaining smooth interview flow required careful workflow orchestration.
Consistent Performance Evaluation
Interview scoring can be subjective, so tuning prompts for balanced evaluation across different answer styles required multiple iterations.
Future Improvements
Planned enhancements include:
- real-time voice emotion detection
- multilingual interview support
- company-specific interview modes
- coding round simulations
- analytics dashboards
- progress visualizations
Conclusion
Building AI Voice Interview Simulator demonstrated how conversational AI and voice workflows can improve interview preparation experiences.
The most exciting part of the project was creating adaptive interview interactions that respond dynamically to user answers while also providing structured performance feedback and emotion analysis.
The combination of voice interaction, AI evaluation, and adaptive questioning helped create a more realistic interview simulation platform.
Top comments (0)