JeffiN11-audio-text-app 🎙️

An AI-powered audio transcription API. Upload any audio file and get a full transcript, automatic summary, sentiment analysis, and speaker detection — all running locally with no API keys needed.

Features

• 🎙️ Transcription — Local speech-to-text via OpenAI Whisper
• 📝 Summarization — Auto-summary via Ollama (llama3)
• 😊 Sentiment Analysis — Positive / Negative / Neutral scoring
• 🗣️ Speaker Detection — Detects multiple speakers
• 📁 File Support — mp3, mp4, wav, m4a, ogg, flac, webm (max 50MB)
• 🐘 PostgreSQL — Stores all transcriptions
• 🐳 Docker — One command to run everything
• ✅ CI — GitHub Actions lint and tests

How It Works

1. Upload an audio file to POST /transcriptions/
2. API saves it instantly and returns status: processing
3. In the background: Whisper transcribes → Ollama summarizes → Ollama analyzes sentiment
4. Poll GET /transcriptions/{id} to get the completed result

Tech Stack

Layer	Technology
API Framework	FastAPI 0.115 (async)
Transcription	OpenAI Whisper (local)
Summarization + Sentiment	Ollama + llama3 (local)
Database	PostgreSQL 16 + SQLAlchemy 2
Containerization	Docker + Docker Compose
Testing	pytest + pytest-asyncio
CI	GitHub Actions

Quick Start

git clone https://github.com/JeffiN11/JeffiN11-audio-text-app.git
cd JeffiN11-audio-text-app
docker compose up --build

API: http://localhost:8000 Docs: http://localhost:8000/docs

API Endpoints

Method	Endpoint	Description
GET	/	Health check
POST	/transcriptions/	Upload audio file
GET	/transcriptions/	List all transcriptions
GET	/transcriptions/{id}	Get single transcription
DELETE	/transcriptions/{id}	Delete transcription

Example

curl -X POST http://localhost:8000/transcriptions/ \
  -F "file=@meeting.mp3"

Response:

{
  "id": 1,
  "filename": "meeting.mp3",
  "status": "processing",
  "transcript": null,
  "summary": null,
  "sentiment": "pending"
}

After processing:

{
  "id": 1,
  "filename": "meeting.mp3",
  "status": "completed",
  "transcript": "Good morning everyone, let us get started with the weekly sync...",
  "summary": "A weekly team sync discussing project progress and upcoming deadlines.",
  "sentiment": "positive",
  "sentiment_score": 0.82,
  "speakers_detected": 3,
  "language": "en",
  "duration": 342.5
}

Running Tests

pip install -r requirements-dev.txt
pip install fastapi uvicorn sqlalchemy asyncpg pydantic httpx python-dotenv python-multipart alembic
pytest -v