Turning multilingual audio into searchable, editable text across iOS and macOS.
CLIENT OR PRODUCT
SoundBridge — multilingual audio transcription product for iOS and macOS.
MY ROLE
Product engineering / client architecture
THE PROBLEM
Teams needed reliable speech-to-text that worked across devices, languages, and real-world audio quality — not a one-off demo notebook.
CONSTRAINTS
- •Dual-platform Apple surface area (iOS + macOS)
- •AI API reliability, latency, and cost trade-offs
- •Editable output UX for non-technical users
- •Production deployment and ongoing iteration
THE INVESTIGATION
Mapped the workflow from capture → transcription → edit → export, and isolated where platform differences and model behavior would break a naive implementation.
THE STRATEGY
Build a native SwiftUI client architecture with clear separation between media handling, AI integration, and editing UX so each layer could be improved without rewriting the product.
WHAT I BUILT
- •iOS and macOS client experiences
- •AI transcription workflow integration
- •Editable, multi-language text output path
- •Product iteration toward production reliability
THE HARD PART
Making AI output feel trustworthy in a consumer-grade editor: handling imperfect audio, multi-language edge cases, and keeping the UI responsive while long jobs run.
THE VERDICT
- •Product available for real users on Apple platforms
- •End-to-end path from audio to editable text
- •Architecture that can absorb model and API changes
RELATED CASES
© 2026 Sidi Li. All rights reserved. | Built with ❤️ and determination. | Thanks for visiting!