📁 AI Meeting Logger - Automate Work Hours with Voice AI

Turn every meeting conversation into structured data—automatically! With FastAPI, Whisper, and LangChain, this project helps consultants and teams log meeting summaries, participants, and key insights directly from voice recordings. All saved to Google Sheets and fully automated via n8n workflows.

---

🌟 Key Features

• ✅ Upload voice notes via API or n8n webhook
• ✅ Transcribe audio to text using Whisper
• ✅ Extract meeting details using LLM-powered insights
• ✅ Save structured data in Google Sheets
• ✅ Send email notifications after each entry
• ✅ Automate workflow seamlessly with n8n

---

🚀 How It Works (End-to-End Workflow)

1. Voice Upload → Upload .mp3 file via API or n8n webhook
2. Transcription → Whisper converts audio to text
3. Info Extraction → LangChain + Ollama extracts meeting details
4. Google Sheets Logging → Meeting data is appended
5. Email Notification → Summary is emailed
6. n8n Orchestration → End-to-end process automation

📌 Fully automated workflow - Just upload a file & let AI handle the rest!

---

🛠 Technologies Used

• FastAPI - Backend API service
• OpenAI Whisper - Audio-to-text engine
• LangChain + Ollama - LLM-driven extraction
• Google Sheets API - Structured database
• n8n - Low-code automation

---

🔧 FastAPI Endpoints

1️⃣ /upload/ – Upload and Process Audio

• Accepts: .mp3 file
• Runs the full pipeline:

  • Transcribes audio

  • Extracts info

  • Adds transcript to result

  • Saves to Google Sheets

  • Sends email

    

2️⃣ /transcribe/

• Input: .mp3 audio file

• Output: transcript (text)

  

3️⃣ /extract/

• Input: JSON with transcript

• Output: Extracted fields like name, start_time, note, etc.

  

4️⃣ /save/

• Input: JSON object with meeting fields

• Output: Saves a row to Google Sheets

  

• Sample of saved data

5️⃣ /notify/

• Input: Same JSON as /save/

• Action: Sends formatted email with extracted info

  

• Sample of email

  

---

🟢 Local Setup Instructions

✅ 1. Clone and Setup

git clone https://github.com/mohannadhendi/ai-meeting-logger.git
cd ai-meeting-logger
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt

✅ 2. Run the App

uvicorn app:app --reload --host 0.0.0.0 --port 8000

✅ 3. Test Endpoints

Use Postman, cURL, or browser to access:

• http://localhost:8000/docs → interactive Swagger UI

---

🔄 Workflow Automation (n8n)

You can trigger the pipeline automatically using n8n.

🔹 Uploading JSON Files to n8n (Meeting_Data_Extraction.json)

1. Open n8n and create a new workflow.

   

2. Send audio file to webhook using API.

   

3. Pass the data through the following nodes:

   • HTTP Request (Transcribe) → Sends audio to /transcribe/

     

   • HTTP Request (Extract) → Sends transcript to /extract/

     

   • HTTP Request (Save) → Saves structured data

     

   • HTTP Request (Notify) → Sends summary email

     

Sample Workflow:

1. Webhook Node - Accepts voice note from client
2. HTTP Request (Transcribe) - Sends audio to /transcribe/
3. HTTP Request (Extract) - Sends transcript to /extract/
4. HTTP Request (Save) & Notify - Parallel saving and emailing

---

📄 Environment Configuration

Create a .env file in root directory:

LANGCHAIN_API_KEY=langchain_api_key
LANGCHAIN_TRACING_V2=true
LANGCHAIN_ENDPOINT=https://api.smith.langchain.com
SERVICE_ACCOUNT_FILE=path_to_your_service_account.json
SPREADSHEET_ID=your_google_sheet_id
WORKSHEET_NAME=Sheet1
EMAIL_SENDER=example@gmail.com
EMAIL_PASSWORD=your_app_password
EMAIL_RECEIVER=target_email@gmail.com
SMTP_SERVER=smtp.gmail.com
SMTP_PORT=587

---

📂 Project Structure

📁 files_project
├── app.py               # FastAPI main file
├── requirements.txt
├── .env
├── uploads/             # Uploaded audio temp storage
├── scripts/
│   ├── speech2text.py
│   ├── extract_data.py
│   ├── save_data.py
│   └── send_email.py

---

📖 References

• 📌 n8n Documentation
• 📌 LangChain + Ollama Documentation
• 📌 FastAPI Documentation
• 📌 Whisper Speech-to-Text
• 📌 Google Sheets API
• 📌 GPT-4 for Research and Document Organization