Realtime-Audio-Transcriber

🎧 Audio Transcription Service

A high-performance, real-time streaming audio transcription backend built with Spring Boot, WebFlux, and Google Gemini API.
Designed for low latency, network resilience, and efficient resource usage, this service streams audio chunks and returns partial transcriptions instantly.

---

🚀 Features

🔹 Core Capabilities

• ✅ Real-time audio streaming (no buffering)
• ✅ Bi-directional communication using WebSockets
• ✅ Low-latency transcription with async, non-blocking processing
• ✅ Network resilience with circuit breakers & retries
• ✅ Scalable architecture using Spring WebFlux
• ✅ Concurrent session management
• ✅ Health checks & metrics monitoring

---

🧰 Tech Stack

Layer	Technology
Framework	Spring Boot 3.2.1
Reactive	Spring WebFlux, Project Reactor
Streaming	WebSockets (SockJS fallback)
AI Engine	Google Gemini 2.0 Flash API
Resilience	Resilience4j
HTTP Client	OkHttp (connection pooling)
Build Tool	Maven
Java	Java 17+

---

📸 Screenshots

Below are sample screenshots demonstrating the service in action:

Home page:

Live Transition:

---

📦 Prerequisites

System Requirements

• Java 17+
• Maven 3.8+
• Internet access to Google Gemini API

---

🔑 Gemini API Key

1. Visit 👉 https://aistudio.google.com
2. Create an API key
3. Set it as an environment variable:

# Windows
set GEMINI_API_KEY=your-api-key-here

# Linux / macOS
export GEMINI_API_KEY=your-api-key-here