README.md

metadata

title: KARTE - 音声カルテ分析
emoji: 🎯
colorFrom: red
colorTo: pink
sdk: docker
app_port: 7860
pinned: false

KARTE - Audio Analysis for Medical Records

KARTE is a powerful application that generates medical records and customer service analysis from audio data. It combines advanced speech recognition with natural language processing to provide comprehensive insights.

Features

• 🎤 Audio Transcription: Converts audio files to text using OpenAI's Whisper model
• 📊 Style Analysis: Evaluates customer service style and communication quality
• 🔄 Flow Analysis: Analyzes conversation flow and structure
• 📝 Medical Record Generation: Creates structured medical records from conversations
• 🔒 Secure Authentication: Basic auth protection for sensitive data
• 📥 Export Functionality: Download analysis reports in JSON format

Technology Stack

• Streamlit: Web application framework
• OpenAI Whisper: Speech-to-text transcription
• Groq: Large language model for analysis
• Python: Core programming language
• Docker: Containerization

Environment Variables

Required environment variables:

• BASIC_AUTH_USERNAME: Username for basic authentication
• BASIC_AUTH_PASSWORD: Password for basic authentication
• OPENAI_API_KEY: OpenAI API key for transcription
• GROQ_API_KEY: Groq API key for analysis

Usage

1. Access the application through Hugging Face Spaces
2. Log in using the provided credentials
3. Upload an audio file (supported formats: MP3, WAV, M4A)
4. Wait for transcription and analysis
5. Review the generated insights and medical record
6. Download the complete analysis report

Deploying to Hugging Face Spaces

Option 1: Using Streamlit SDK (Recommended)

1. Make sure your README.md has the correct metadata at the top: ```

   title: KARTE - 音声カルテ分析 emoji: 🎯 colorFrom: red colorTo: pink sdk: streamlit sdk_version: 1.32.2 app_file: main.py pinned: false

2. Set up your environment variables in Hugging Face Spaces:

   • Go to your Space settings
   • Add the following secrets:
     • OPENAI_API_KEY
     • GROQ_API_KEY
     • BASIC_AUTH_USERNAME
     • BASIC_AUTH_PASSWORD

3. Push your code to the Hugging Face repository:

   git add .
   git commit -m "Deploy Streamlit app"
   git push
   

Option 2: Using Docker (Advanced)

1. Update your README.md metadata to use Docker: ```

   title: KARTE - 音声カルテ分析 emoji: 🎯 colorFrom: red colorTo: pink sdk: docker pinned: false

2. Make sure your Dockerfile is properly configured:

   • Uses port 7860
   • Sets up a non-root user
   • Installs all dependencies including ffmpeg
   • Properly sets environment variables

3. Set up your environment variables in Hugging Face Spaces as in Option 1

4. Push your code to the Hugging Face repository:

   git add .
   git commit -m "Deploy Docker app"
   git push
   

Troubleshooting

If you encounter issues with deployment:

1. Start with the diagnostic app:

   • Change the Dockerfile to use test_app.py instead of main.py
   • This will help identify environment and dependency issues

2. Check the Spaces logs for errors:

   • Go to your Space → "Logs" tab
   • Look for any error messages related to missing dependencies or environment variables

3. Common issues:

   • Missing environment variables
   • Missing system dependencies (like ffmpeg)
   • Incorrect port configuration
   • Memory or resource limitations

Development

To run locally:

1. Clone the repository
2. Install dependencies: pip install -r requirements.txt
3. Set up environment variables
4. Run the application: streamlit run main.py

License

This project is licensed under the MIT License.