⊘
SCREEN TOO NARROW

Please open on a larger screen to view the IDE interface.

Or email: hello@itsbor.is

selected_projects › voz42.md
voz42.md

voz42

A private, local-first macOS app for dictation and transcription.

  • Speak, and the text appears in whatever Mac app is focused
  • Auto-translate your speech and adjust the transcription style
  • Transcribe files, videos, YouTube/Vimeo links, and podcasts
  • Record and transcribe meetings
  • Edit transcripts and export the results

voz42 main transcription window

The speech-to-text market is full of players now, so why do we need yet another app?

I wanted to build something free and privacy-first. All ASR work happens locally, with Whisper, Silero VAD, Core ML, and native macOS audio APIs doing most of the heavy lifting. Most data lives in local SQLite databases. Release builds do not keep diagnostic audio, and history can be paused or cleared.

voz42 studio transcript editor

voz42 settings and workflow view

Tech Stack

  • App: Swift + SwiftUI
  • Dictation insertion: macOS Accessibility APIs + global hotkey event handling
  • Audio capture: AVFoundation for microphone input; Core Audio process taps for system audio recording
  • Voice activity detection: Silero VAD converted to Core ML
  • Media processing: FFmpeg for audio extraction and conversion
  • URL import: URLSession for direct media and podcasts; yt-dlp for platform links like YouTube
  • Translation: Apple Translation framework / TranslationSession on macOS 15
  • Persistence: SQLite databases for history, projects, segments, speakers, snippets, glossary, and corrections
  • Distribution: signed and notarized macOS .dmg

Features

  • Speak anywhere: global hotkey starts dictation and inserts text into the focused Mac app
  • 20+ languages for local transcription workflows
  • Studio mode: import audio/video files, edit transcripts, merge/split segments, and work with timestamps
  • System audio recording: capture meetings, calls, browser audio, and other Mac output
  • URL transcription: import direct media links, podcast feeds, and YouTube-style platform URLs
  • Speaker diarization: detect and label different speakers in longer recordings
  • Auto-translation: translate supported languages to English with Apple's system translation framework
  • Personal dictionary: glossary and correction rules help preserve names, brands, and technical terms
  • Snippets and history: reuse repeated phrases and recover previous transcriptions
  • Exports: TXT, SRT, WebVTT, JSON, Markdown, PDF, and DOCX
main itsbor.is