bilalkaradeniz
← Back to all work

2025 — now · Creator, backend and app

VoText

Speech to text in 30+ languages, and any text read aloud.

30+ languages, mixed in one recording
PDF, Markdown, SRT and VTT export
Transcripts stay on the device
Stack
Go · PostgreSQL · Speech-to-text · Text-to-speech · iOS · React · Node.js
votext.app ↗
uploadpurchasesjob queueVoText appGoProvider routingSpeech-to-textiOSPostgreSQLText-to-speechuploadpurchasesjob queueVoText appGoProvider routingSpeech-to-textiOSPostgreSQLText-to-speech

Problem

Transcription apps assume one language per recording. Real conversations switch, and long files fail or cost too much.

System

A Go backend accepts uploads, queues speech jobs in PostgreSQL, and routes each job to the right speech-to-text or text-to-speech provider. It handles authentication, credit accounting and App Store purchase verification for the iOS app.

VoText website
votext.app

Hard parts

  • A PostgreSQL-backed queue that survives restarts and retries safely.
  • Routing between providers by language, length and cost.
  • Server-side App Store purchase verification tied to credits.

Result

Coming to the App Store for iPhone and iPad, priced in minutes: 20 free credits to start, Plus and Pro plans, and top-ups that never expire. Recordings stay on the phone, servers delete audio within 24 hours, and there are no ads or trackers. The site is live at votext.app in 30+ languages.

Next projectTusTrack