README.md

metadata

title: Catherine's Catering Voice Simulator
emoji: 🎤
colorFrom: blue
colorTo: purple
sdk: streamlit
sdk_version: 1.37.1
app_file: app.py
pinned: false
license: mit
short_description: Voice role-play simulator for business interviews

Catherine's Catering Voice Simulator

A real-time, voice-to-voice interview simulator featuring Catherine, owner of Catherine's Catering. This application uses OpenAI's Realtime API to provide an interactive role-play experience for business consulting students.

Features

• Voice-only interaction using OpenAI's Realtime API over WebRTC
• Role-play simulation with Catherine, a non-technical small business owner
• Time-limited sessions (5 minutes per group)
• Group-based authentication (groups 1-7)
• Real-time audio with automatic voice detection

How to Use

1. Set up your OpenAI API key in the Space's secrets
2. Login using group credentials (group1-group7, password same as username)
3. Start a conversation and speak directly to Catherine
4. Interview Catherine about her catering business challenges
5. Session automatically ends after 5 minutes

Setup for Local Development

1. Create and activate a virtual environment:

   python -m venv .venv && source .venv/bin/activate
   

2. Install dependencies:

   pip install -r requirements.txt
   

3. Set your OpenAI API key:

   export OPENAI_API_KEY=sk-...
   

4. Run the application:

   streamlit run app.py
   

Environment Variables

• OPENAI_API_KEY: Your OpenAI API key with Realtime access
• OPENAI_REALTIME_MODEL: Model to use (default: gpt-4o-realtime-preview-2024-12-17)
• OPENAI_VOICE: Voice selection (default: shimmer)

Technical Details

• Backend: FastAPI server for token management
• Frontend: Streamlit with embedded HTML client
• Audio: WebRTC for real-time voice communication
• Authentication: HTTP Basic Auth for group access
• Time Management: 5-minute session limits per group

Troubleshooting

• Ensure microphone permissions are granted
• Check that your OpenAI API key has Realtime access
• Some corporate networks may block WebRTC connections
• Verify the token server is running at the expected port