Voicify — Speech to Text Mobile App
Privacy-first mobile app that converts speech to text in real time using Flutter and Google's Speech Synthesis library — no audio ever leaves the device.
Problem
Existing transcription apps store audio on remote servers — a privacy concern for classrooms, legal settings, and personal use. Users needed a simple, reliable transcription tool that keeps data local.
Solution
A Flutter mobile app using Google's default speech synthesis services for on-device transcription. Real-time text output, local-only storage — no audio is recorded or transmitted. Clean, minimal UI focused on the transcription task.
System Architecture
Flutter mobile app. Speech-to-text via Flutter's speech synthesis library backed by Google's on-device recognition. All transcribed text stored locally on device only. No backend, no server, no data transmission.
# system diagram
User (iOS / Android)
└─ Flutter app
├─ Speech synthesis library → on-device recognition
├─ Google Speech API → transcription
└─ Local storage only → no server, no upload
DevOps & Deployment
Deployment
App Store · Play Store
CI/CD
GitHub Actions with Fastlane for automated store submissions
Challenges
Speech recognition accuracy varies across devices and accents. Surfaced confidence indicators in the UI so users know when to re-speak a phrase rather than silently getting wrong output.
Key Learnings
On-device ML capabilities in modern phones are powerful enough for real-time transcription without any server infrastructure. Privacy-by-architecture is a genuine product differentiator.