AI Meeting Note Taker (Offline, Private & Unlimited!)

🎙️ AI Meeting Note Taker (Offline, Private & Unlimited!)

Tired of short recording limits and privacy concerns with online tools? Transcribe and summarize meetings of ANY LENGTH completely offline on your own machine. This is a one-time purchase for a lifetime of private, unlimited meeting analysis.

• 🔒 Your Data Never Leaves Your Computer: Unlike cloud-based subscription services, this tool processes all your audio files and generates summaries locally. Your sensitive meeting content remains completely private. >
• ⏳ No Time Limits & No Subscriptions: Process hour-long meetings, interviews, or lectures without any restrictions. Pay once and use it forever.

🌟 Key Features

• Offline Audio Transcription: Convert long audio recordings to text (supports `.m4a`, `.mp3`, `.wav`, `.flac`, `.ogg`, `.wma`, `.aac`).
• Private AI-Powered Summaries: Generate concise meeting minutes, identify action items, and extract key decisions using a local AI model (Ollama).
• Handles Large Files: Process audio files up to 500MB, suitable for long-form recordings without issues.
• Network Accessible: Run the service on one computer and access it from any other device on your local network.
• Multiple Export Formats: Download your meeting notes and transcripts as `TXT`, `Markdown`, or `PDF` files.
• Easy-to-Use Web Interface: A simple drag-and-drop interface for uploading and processing your audio files.
• Standalone Scripts Included: Comes with command-line tools for transcription and minutes generation, perfect for automation or integration into other workflows.
• Full Source Code: You get the complete Python Flask application code, which you can modify or extend as you see fit.

🚀 How to Get Started

1. Install Ollama
   • Download and install Ollama from here
   • Pull a model to handle the summarization. 'gemma3:27b' is a great choice. Type the below command in your terminal (You can select smaller model as well.)

     ollama pull gemma3:27b

2. Install Python Dependencies
   • Unzip the package you downloaded from Gumroad
   • Open your terminal, navigate to the meeting-note` directory, and run:

     pip install -r requirements.txt

   • This will install Flask, Whisper for transcription, and other necessary libraries
3. Configure Your AI Model (Optional)
   • The application is configured to use gemma3:27b by default
   • To change this, create a env file by copying the .env.example:

     cp .env.example .env

   • Now, edit the env file to specify a different model

     OLLAMA_LLM_MODEL=another-mode:latest #example: phi4:14b

   • Update line 39 in the app.py file with your chosen ollama model
4. Run the Application

   python app.py

5. Access the Web Interface:
   • Open your web browser and go to http://127.0.0.1:8080 or http://localhost:8080
   • You can now upload your audio files and start generating meeting notes

🛠️ Advanced Usage: Accessing from Other Devices

To use the app from other computers on your local network (e.g., use your laptop to access the app running on your desktop), you need to make one small change:

1. Open the app.py file.
2. Go to the very bottom of the file.
3. Change this line:

   app.run(debug=True, host='127.0.0.1', port=8080)

to

app.run(debug=True, host='0.0.0.0', port=8080

4. Now, run python app.py. You can access the app from other devices using the local IP address of the computer it's running on (e.g., http://192.168.1.10:8080). You may need to allow this connection through your firewall.

💡 Tips for Best Results

• Ensure your audio recordings are clear for the best transcription accuracy.
• Processing time depends on the length of the audio file and the power of your computer. Since all processing is done locally for 100% privacy, it will naturally be slower than cloud-based services, but your data remains secure and confidential.
• To balance speed and accuracy, you can adjust the model sizes.
• For Summarization: Change the OLLAMA_LLM_MODEL to a smaller model you have downloaded (e.g., llama3:8b or phi4:14b, gemma3:4b). Smaller models are faster but may be slightly less detailed. For better performance please use non thinking model such as gemma, llama, phi, etc.
• For Transcription : The web application uses the powerful "large-v3" Whisper model by default. For faster results with a slight accuracy trade-off, you can edit the code directly.
  • Open the app.py file.
  • Find the line (around line 48): WHISPER_MODEL = whisper.load_model("large-v3")
  • Change "large-v3" to a smaller model like "base.en" or "small.en".

📄 License

This project is licensed under the MIT License. You are free to use, modify, and distribute the code.

Thank you for your purchase! If you have any questions, please refer to the contact information on the Gumroad product page.