Smart Dictation

Inspiration

I wanted a fast, reliable way to capture speech on Mac without juggling between separate transcription services and translation tools. Many solutions are either locked behind clunky interfaces or only offer one of the two. Smart Dictation was born from the need for a simple, powerful dictation tool that works natively on Mac and integrates seamlessly for users.

What it does

Smart Dictation transcribes speech into text with high accuracy and supports instant translation into multiple languages. It uses the latest Open AI model (GPT-4o-transcribe) allowing users to record directly on macOS, transcribe in real time, and export or share their notes.

How we built it

I built Smart Dictation in SwiftUI for a clean, native macOS experience. The transcription and translation are powered by the OpenAI API (GPT-4o-transcribe, then using the Whisper model as a fallback if any issues with the newer model, then it uses GPT-4 for translation). For subscriptions and payments, I have of course integrated the wonderful RevenueCat :)

Challenges we ran into

Handling large recordings with high performance whilst keeping the app lightweight.
Designing a user interface that feels simple but still offers advanced features.
Getting subscription logic polished without breaking the user flow.

Accomplishments that we're proud of

Built a fully native macOS app.
Achieved high transcription accuracy with minimal latency.
Created a clean, distraction-free design.
Launched with subscription tiers that make sense for both casual and professional users.

What we learned

The importance of local caching and error handling for a smooth offline-to-online transition.
Better practices to securely handle API keys.
Nuances between iOS and macOS development, as this was my first time building for Mac.