Subline

Subline

Generate subtitles and transcribe meetings.
Privately on your Mac.

Download for Mac Install with Homebrew

v0.4.10 · macOS 13+

Subline uses OpenAI's Whisper to transcribe speech with remarkable accuracy across 99 languages. Generate subtitles for videos, transcribe meeting recordings with speaker identification, or convert any audio to text. Runs natively on Apple Silicon with Metal GPU acceleration. No cloud, no API keys, no subscriptions.

Get Subline

macOS 13 or later. Apple Silicon and Intel.

Direct download (recommended)

Subline.dmg macOS 13+ · Apple Silicon & Intel

Open the DMG and drag Subline to your Applications folder. Includes the GUI app and subline CLI.

Homebrew — GUI app

$ brew install --cask fenelon/tap/subline

Installs the app and CLI. Updates with brew upgrade --cask.

Homebrew — CLI only

$ brew install fenelon/tap/subline

Command-line tool only. For batch processing, piping, and scripting.

Subline downloads the Whisper model (~500 MB) on first run. Speaker diarization models are downloaded separately when needed. Everything is cached locally.

Features

99 languages

Auto-detects the spoken language or set it manually. From English to Urdu, Whisper handles it.

GPU accelerated

Leverages Metal on Apple Silicon for fast inference. A 1-hour video transcribes in minutes.

Speaker diarization

Identifies who said what. Supports 2 to 5 speakers with automatic voice separation.

Drag and drop

Drop video or audio files onto the window. Batch process entire folders at once.

SRT, VTT, TXT

Output in the format you need. SRT and VTT include timestamps for video players.

Fully local

Everything runs on your machine. Your audio never leaves your computer.

App

Subline — drag and drop interface with model, language, format and speaker settings
Speaker diarization settings with dropdown
Transcription in progress with ETA
Multi-track audio selection dialog
Parallel transcription and speaker identification

CLI

$ subline [options] <path...>
-m, --modelWhisper model: tiny, base, small, medium, turbo, large turbo
-l, --languageLanguage code — auto-detects if omitted
-f, --formatOutput format: srt, vtt, or txt srt
-S, --speakersNumber of speakers — enables diarization, implies txt
-p, --promptInitial prompt for Whisper (vocabulary, punctuation context)
-a, --audio-trackAudio stream index auto
-o, --output-dirDirectory to write output files next to source
-s, --skip-existingSkip files that already have a subtitle file
-v, --verboseShow detailed model loading and engine output

Examples

# Transcribe a video — generates video.srt
$ subline video.mp4

# Meeting transcript with 3 speakers
$ subline --speakers 3 meeting.mp3

# Batch process a folder in Spanish
$ subline --language es ~/Videos/