Arré Voice

Arre, one of India's most recognised digital media companies, had identified a massive opportunity in social audio. Clubhouse had proven the format, but no Indian platform was building specifically for women creators — a demographic that was underserved across every social media platform.

The vision was ambitious: build a short-form social audio platform where women could create 30-second audio clips (Voicepods), host live group conversations (Jampods), and grow an audience — all within a safe, moderated environment. The app needed a built-in Creator Studio with professional audio tools, real-time audio streaming for live rooms, content discovery algorithms, and social features like following, commenting, and sharing.

The constraint was time. The social audio window was closing fast — Spotify, Twitter (Spaces), and Discord were all shipping audio features. Arre needed to go from wireframes to a production app on both iOS and Android within 14 weeks, with a budget that ruled out building two separate native apps. The product needed to feel native on both platforms, handle real-time audio without latency or drops, and scale to handle the traffic surge from a high-profile launch backed by celebrity collaborators including musician Ankur Tewari.

We chose Flutter for cross-platform development — a single codebase for iOS and Android that could deliver native performance within the 14-week timeline. The backend was built with NestJS (Node.js) and GraphQL, giving the mobile app a flexible, efficient API layer that minimised over-fetching and made rapid iteration possible.

The core audio engine was built on the Agora SDK, providing low-latency real-time audio streaming for live Jampod rooms supporting up to 50 concurrent speakers and thousands of listeners. We implemented WebSocket connections for real-time presence indicators, live reactions, and hand-raise queuing — giving hosts full control over their rooms.

The Creator Studio was the product's differentiator. We built a mobile-first audio recording and editing interface with background music tracks (licensed through Jam 8), voice effects, volume mixing, and one-tap publishing. The 30-second constraint was intentional — we implemented a countdown timer with visual feedback that created urgency and encouraged concise, punchy content.

Content discovery was powered by a recommendation engine that combined collaborative filtering (users who liked X also liked Y) with content-based signals (audio transcription tags, creator categories, engagement velocity). The home feed was personalised from the first session using onboarding preferences.

Moderation was critical for a women-first platform. We built a multi-layered moderation system: automated audio content screening, community reporting with priority queues, live room monitoring tools for hosts, and an admin moderation dashboard with user suspension and content removal capabilities.

The infrastructure was deployed on AWS — EC2 for compute, S3 for audio file storage, CloudFront for CDN delivery, and Redis for real-time caching. Firebase handled push notifications and analytics. The entire CI/CD pipeline was automated through GitHub Actions with separate staging and production environments.