Back
Echoly
An iPhone app that turns long recordings into readable text with speaker labels, timestamps and full-text search. Built to handle a two-hour lecture or an entire meeting in one pass.
Problem
Most transcription apps are built for short clips. But the recordings that actually need transcribing are the long ones: a two-hour lecture, a full interview, an entire meeting. Three problems surface at once with these. The app can't keep recording in the background, so locking the screen ends the session. The output arrives as one undifferentiated wall of text, and who said what is lost. And a single crash takes the whole session with it. So people either keep typing it out by hand, or never open the recording again.
Solution
We built the app around the long-recording case from the start. Recording continues with the screen locked, and audio is written to disk incrementally, so a crash never costs a session. Upload and transcription happen server-side, so device power isn't the constraint. The transcript is designed to be read: generous line spacing, adjustable text size, and the current word highlighted as audio plays. Tap any word to jump to that moment; search the full transcript and move between matches. Speaker separation keeps interviews and meetings readable. Output exports as plain text or SRT subtitles. Privacy was a deliberate product decision: audio is uploaded only when the user requests a transcript, recordings kept on device never leave it, and user data isn't used to train models. Users can delete a recording — or their whole account — from inside the app. The product runs on a free tier (30 minutes per month) plus two subscription levels, with App Store subscription infrastructure integrated.
Results
· Live on the App Store, English and Turkish support
· Two-hour recordings processed in a single session
· Revenue model built on a free tier plus two subscription tiers
Tech Stack
SwiftPostgreSQL
