Product Platform
VoiceaRoo
Where Memories Speak
US voice-greeting-card platform for individuals and B2B gifting, with AI transcription and animated delivery
The problem
A printed card is thrown away; a group WhatsApp message disappears. VoiceaRoo’s premise is that voice carries emotion text cannot, and that a collected set of voices from everyone who matters is a genuinely different gift. Delivering that meant solving hard media problems in the browser: reliable one-tap recording across devices, accurate transcription that syncs to the audio, background music mixing, animated presentation that renders beautifully on any screen, and a group-collection flow where dozens of contributors can add a message without any of them creating an account. All of it had to feel effortless to a non-technical gift-giver.
How we built it
8 decisionsBuilt a browser-native one-tap recorder with device permission handling, waveform feedback, retake and trim, no app install required from contributors.
Integrated AI speech-to-text so every message is transcribed and displayed in sync with the audio, making cards readable, accessible and skimmable.
Engineered a media processing pipeline for normalisation, noise reduction, format conversion and mixing of the recorded voice with the selected background music track.
Developed an animated card renderer with themed templates, photo support and music-timed transitions for a premium unboxing moment.
Created a frictionless group-collection flow: the organiser shares one link, contributors record without signing up, and the platform assembles the collection automatically.
Delivered link and QR-code distribution so a card can be handed over physically or sent digitally.
Built B2B capability for corporate gifting: bulk card generation, branded templates and campaign-level administration for teams and employers.
Containerised the entire stack with Docker and deployed on AWS with S3-backed media storage and CDN delivery for fast global playback.
System architecture
10 layersEvery layer designed, built and operated for this system to run in production.
Frontend
React / Next.js, responsive mobile-first UI, Web Audio and MediaRecorder APIs
Backend
Node.js with Express: MERN architecture
Database
MongoDB for cards, contributors, orders and template configuration
Caching & Queues
Redis for session state, rate limiting and asynchronous media-job queues
Media Processing
FFmpeg-based normalisation, mixing, transcoding and rendering pipeline
AI Services
Speech-to-text transcription with timestamp alignment
Cloud Infrastructure
AWS: EC2 compute, S3 object storage, CloudFront CDN distribution
Containerisation & DevOps
Docker images, environment-based CI/CD, health checks and monitoring
Payments
Stripe checkout and order lifecycle handling
Security
Signed media URLs, expiring share links, access-scoped card retrieval
What was delivered
12 modules- 01
One-Tap Voice Recorder
Cross-browser capture with permissions handling, waveform preview, retake and trim.
- 02
AI Transcription & Sync
Automatic speech-to-text rendered in time with playback, with editable corrections.
- 03
Card Designer
Themed templates, colour schemes, photo upload and layout customisation.
- 04
Background Music Library
Curated licensed tracks with automatic ducking against the voice track.
- 05
Animation & Presentation Engine
Music-timed animated reveal optimised for mobile and desktop playback.
- 06
Group Card Collection
Organiser-led invitation flow, unlimited contributor recordings, no-signup contribution.
- 07
Share & Delivery
Shareable links, QR codes, scheduled delivery and recipient view tracking.
- 08
B2B / Corporate Module
Bulk creation, branded templates, campaign administration and team-level billing.
- 09
Payments & Checkout
Per-card purchase for individual and group tiers with secure card processing.
- 10
Media Storage & CDN
Durable object storage for audio, images and rendered cards with edge delivery.
- 11
Admin Console
Content moderation, template management, order administration and support tooling.
- 12
Analytics
Creation-to-delivery funnel, open and playback rates, tier conversion.
Outcomes
5 resultsZero-friction contribution: recipients of an invite can add their voice without downloading anything or creating an account, which is the single biggest driver of group-card completion.
Asynchronous media pipeline so heavy audio work never blocks the user interface, keeping the creation flow fast.
One product serving two very different buyers: an individual sending a birthday card and an enterprise running an employee-appreciation campaign, from a shared core.
Global playback performance through CDN-backed media delivery.
Fully containerised deployment giving the client reproducible environments and simple horizontal scaling.
Value delivered
For consumers: a keepsake gift that carries real emotion and costs less than a bouquet.
For corporate buyers: a scalable, on-brand appreciation channel for employees, clients and milestones.
For the business: high-margin digital product economics with a natural viral loop built into every group card.
Client and product names are used with reference to publicly available product information. Engagement details covered by non-disclosure agreements are described at capability level only.