Transcribr macOS app main interface showing drag-and-drop audio transcription with model selection panel
macOS App

Transcribr — Private AI Transcription for macOS

Drop any audio or video file, hit record, or paste a YouTube link. Transcribr turns it into accurate, timestamped text in seconds — all on your Mac, nothing leaves your machine.

Available on Mac App Store

App Screenshots

Transcribr main UI with drag-and-drop zone, file picker, record button, and URL paste field
1 / 4

Project Overview

Transcribr is a native macOS app that brings powerful AI transcription to your desktop without any cloud dependency. Everything runs locally using Whisper-based models — your audio, your transcripts, your privacy.

Drop an MP4, MOV, MP3, WAV, M4A, FLAC, or WebM file, record live audio with the built-in recorder, or paste a YouTube, Instagram, or TikTok URL to pull audio directly from the web. Transcribr handles it all.

The model selector lets you pick the right speed/accuracy tradeoff — Tiny (77MB) for near-instant results, Base (147MB) for most use cases, Small (486MB) for high accuracy, or the advanced models for professional-grade output. Auto-language detection covers English, Spanish, French, German, Italian, Portuguese, Chinese, Japanese, Korean, Arabic, Hindi, Russian, Dutch, Polish, Turkish, Persian, and more.

The Ollama Cleanup toggle routes your transcript through a locally-running LLM (Gemma4 by default) for grammar fixes, filler-word removal, and a clean final output — still 100% offline. Results can be exported with or without timestamps, copied to clipboard, or remixed with AI. A system-wide Dictation Shortcut (Hold Ctrl by default) means you can trigger recording from any app.

Key Features

  • Fully offline — no data leaves your machine, ever
  • Drag-and-drop transcription for MP4, MOV, MP3, WAV, M4A, FLAC, WebM
  • Live recording with real-time transcription display
  • YouTube, Instagram, and TikTok URL import
  • Four model tiers: Tiny (77MB) to advanced for speed vs. accuracy control
  • Auto language detection across 15+ languages
  • Ollama LLM cleanup — local grammar correction and filler-word removal
  • Timestamped transcript segments with export and copy options
  • System-wide dictation shortcut for use across any app
  • Remix with AI — further refine transcripts without leaving the app

Tech Stack

SwiftSwiftUIWhisperOllamaCore AudioAVFoundationyt-dlp

Platforms

macOS

Project Details

Year

2025

Duration

2 months

Team Size

2 specialists

Metrics

Users

100% local — zero cloud dependency

Performance

Whisper + Ollama on-device AI

Market Position

Privacy-first macOS transcription

Business Impact

Transcribr shows DevX Group's ability to ship polished, production native macOS apps with sophisticated on-device AI — Whisper for transcription and Ollama for LLM cleanup — while keeping user privacy non-negotiable.

Interested in a Similar Project?

Let's discuss how we can build something great for your business.