Voicer — Real-Time WebRTC Audio Spaces

The Vision

Real-time audio applications face extreme challenges: maintaining sub-150ms latency across global connections, handling symmetric NAT traversal, and dynamically managing audio levels for hundreds of concurrent participants.

Voicer was built to solve these architectural bottlenecks with zero third-party cloud audio SDK lock-in.

Key Highlights

• Custom Signaling Engine: Built on lightweight WebSockets with Redis Pub/Sub backplane for multi-instance horizontal scaling.
• NAT Traversal & TURN Fallback: CoTURN clustering with geo-DNS routing ensuring 99.9% connectivity across restrictive firewalls.
• Adaptive Jitter Buffer: Client-side adaptive buffering algorithm minimizing audio jitter under unstable packet loss conditions.

// WebRTC Peer connection state machine
const peer = new RTCPeerConnection({
  iceServers: [{ urls: 'stun:stun.l.google.com:19302' }]
});

peer.ontrack = (event) => {
  const [remoteStream] = event.streams;
  audioPlayer.srcObject = remoteStream;
};

🚀 Tech Stack

Frontend

• React.js (with Redux Toolkit for state management)
• RTK Query for API integration
• WebRTC for peer-to-peer audio streaming
• TailwindCSS / Chakra UI (for UI)

Backend

• Node.js + Express.js
• MongoDB (user, rooms, and profile data)
• Redis (caching + pub/sub for socket scaling)
• Socket.IO (signaling + realtime presence)
• WebRTC (voice communication)
• TURN/STUN servers for NAT traversal
• (Future) SFU (e.g., mediasoup/LiveKit) for scaling audio rooms

📌 Planned Features

🔐 Authentication

• OTP-based login (via phone/email)
• User profile (full name, profile picture, username, followers/following)

🎧 Rooms

• Public Rooms → Visible & joinable by any authenticated user
• Closed Rooms → Join via invite link only
• Roles:
  • Speakers → Can mute/unmute themselves, choose mic device
  • Listeners → Join in listen-only mode
• Room creator can:
  • Add/remove speakers
  • Delete room

👤 Profile

• Profile page with followers/following count
• Ability to explore rooms & users

🏗️ Project Status

• ✅ Tech stack finalized
• ✅ Initial architecture & backend setup
• ⏳ Working on authentication & room management APIs
• ⏳ Next step: WebRTC integration for realtime audio

Impact & Results

• Achieved consistent 120ms round-trip latency globally across 5 regions.
• Successfully supported 500+ simultaneous listeners in a single interactive room.