Skip to main content
Back to Projects
🎙️
Case Study Live

VocalHub

A voice-first community platform for live audio rooms, podcasts, and creator-driven conversations — built for low-latency engagement.

Role:Fullstack Solutions Architect
Next.jsTypeScriptNode.jsWebRTCPostgreSQL
<200ms
Audio Latency
Low-latency WebRTC voice rooms
3
Room Types
Live rooms, podcasts, 1:1 chats
500+
Concurrent Listeners
Per room with stable streaming
Auto
Recording
Replay-ready audio archives

Key Engineering Achievements

The engineering decisions and challenges behind this project

Real-Time Voice Room Engine

Challenge

Hosting live audio with multiple speakers and hundreds of listeners without dropped packets or noticeable lag.

Solution

Built a WebRTC-based audio pipeline with SFU (Selective Forwarding Unit) topology, adaptive bitrate, and a fallback to HLS for high-listener-count rooms.

Impact

Sub-200ms speaker-to-listener latency with stable streaming for 500+ concurrent listeners.

WebRTCSFUHLSAdaptive Bitrate

Creator-First Room Controls

Challenge

Hosts need full moderation power without breaking the conversation flow.

Solution

Designed a host control panel with raise-hand queue, speaker promotion, mute-all, and one-click muting with visual speaking indicators.

Impact

Hosts run rooms with 50+ participants smoothly without third-party moderation tools.

ModerationHost ToolsUX

Replay & Podcast Archive

Challenge

Audiences who miss live rooms still want to consume the content.

Solution

Implemented automatic room recording with chapter markers, transcript generation, and a podcast-style replay feed with playback speed controls.

Impact

Live rooms turn into evergreen podcast content with zero extra effort from the creator.

RecordingTranscriptsPodcast Feed

Project Analytics

A data-driven look at the project's architecture and performance

Feature Priority

Voice Engine95
Host Controls80
Recording75
Discovery65
Profiles60

Architecture Split

100%
Audio Pipeline35%
Frontend30%
Backend API20%
Database & Storage15%

Developer's Note

VocalHub taught me that real-time audio is unforgiving — even 300ms of lag breaks the conversational flow. WebRTC tuning, SFU topology, and graceful fallback to HLS were the difference between a usable platform and a frustrating one.

GN
George Njoroge
Fullstack Solutions Architect