🎵 SwissHack25-Priceless

SwissHack25-Priceless is an AI-powered full-stack application developed during SwissHack25, aiming to transcribe, analyze, and semantically interpret audio input. It integrates WhisperX for efficient speech-to-text transcription and Ollama models (like LLaMA and Mistral) for advanced language processing, all wrapped within a scalable Dockerized microservice architecture.

---

🚀 Technologies Used

Component	Technology	Purpose
Frontend	Vue 3	User interface with modern JS framework
Backend	Flask	API server handling ML and DB logic
AI/ML	WhisperX	High-performance audio transcription and alignment
AI/ML	Ollama	Local LLM server supporting Mistral and LLaMA
Database	MongoDB	NoSQL database for storing transcription data and semantic vectors
Container	Docker	Environment and deployment management
Orchestration	Docker Compose	Service linking, volume and port management

---

📁 Project Structure

SwissHack25-Priceless/
├── backend/                 # Flask API with WhisperX + Ollama integration
│   ├── app/                 # Application modules
│   │   ├── whisperx_api.py  # WhisperX-based audio transcription
│   │   ├── llm_ollama.py    # Ollama-based semantic parsing
│   │   └── db.py            # MongoDB interaction layer
│   └── requirements.txt     # Python dependencies
│
├── frontend/                # Vue 3 frontend
│   ├── src/
│   │   ├── components/      # UI components
│   │   └── views/           # App views (Home, Result etc.)
│   └── package.json         # JS dependencies and scripts
│
├── data/                    # Data volume and seed data
│
├── compose.dev.yml          # Docker Compose file for development
├── compose.prod.yml         # Docker Compose file for production
└── README.md                # Main documentation

---

🧠 Core Features

• 🎹 WhisperX Transcription: Fast and aligned multi-language speech-to-text.
• 🧠 Semantic Search & Embedding: LLM-generated vector embeddings and summarizations.
• 🔍 Searchability: Indexed and retrievable audio insights stored in MongoDB.
• 📦 Modular Containers: Easily scalable via Docker Compose for dev and prod.

---

⚙️ Local Development

1. Prerequisites

• Docker & Docker Compose
• Node.js (for frontend)
• pyenv (recommended for Python)

2. Clone the Repository

git clone https://github.com/liliaGerber/SwissHack25-Priceless.git
cd SwissHack25-Priceless

3. Development Setup (via Docker Compose)

docker compose -f compose.dev.yml up --build

• Frontend available at: http://localhost:5173
• Backend API on: http://localhost:5000
• MongoDB at: mongodb://localhost:27017

4. Stopping Services

docker compose -f compose.dev.yml down --volumes

---

🧪 Sub-Module Documentation

📁 frontend/

# Frontend - Vue 3

This folder contains the Vue.ts frontend that serves the user interface for the SwissHack25-Priceless app.

## 📦 Setup

```bash
cd frontend
pnpm install
pnpm dev

🔨 Build for Production

pnpm build

🔧 File Structure

• src/components/: Reusable UI components
• src/views/: App views
• main.js: Entry file
• vite.config.js: Vite config

---

### 📁 `backend/`

```markdown
# Backend - Flask + AI Services

This folder contains the Flask-based backend which integrates WhisperX and Ollama APIs.

## 💡 Key APIs

- `/transcribe`: Uses WhisperX to transcribe audio
- `/semantic`: Sends data to Ollama for embedding/summarization

## ⚙️ Setup (Manual - Optional)

```bash
cd backend
pyenv install 3.13.1
pyenv virtualenv 3.13.1 venv
pyenv activate venv
pip install -r requirements.txt

📂 Files

• whisperx_api.py: Speech-to-text pipeline
• llm_ollama.py: Chat and embedding logic with Ollama
• db.py: MongoDB connection and queries

---

### 📁 `data/`

```markdown
# Data Folder

This directory holds volume mounts for MongoDB and example seed data for testing.

## 🧪 Usage

Used during container builds to initialize Mongo collections or persist across runs.

---

👥 Authors

• Lilia Gerber
• Arka Mitra
• Kaja Vujic
• Olha Sirikova

---

📨 Feedback & Issues

Open an issue or contact @liliaGerber.

// For Mac brew install python-tk