Skip to content

CLI Reference

= edited this page Feb 10, 2026 · 1 revision

CLI Reference

Install:

cargo install transcriber-cli

Basic usage

# Transcribe a URL
transcriber-cli https://youtube.com/watch?v=...

# Transcribe a local file
transcriber-cli recording.mp3

# Output as SRT subtitles
transcriber-cli video.mp4 --format srt

# Save to a file
transcriber-cli video.mp4 --format srt --output subtitles.srt

All flags

Flag Default Description
--format / -f text Output format: text, srt, vtt, json
--output / -o stdout Write output to file
--model / -m large-v3 Whisper model (see Models)
--language / -l auto Language code or auto for detection
--translate off Translate to English
--word-timestamps off Include word-level timing
--no-gpu off Force CPU-only
--gpu-device 0 GPU device index
--threads auto CPU thread count
--no-vad off Disable voice activity detection
--temperature 0.0 Sampling temperature
--beam-size greedy Beam search width
--cache-dir system default Model cache directory
--dc-offset off Remove DC offset from audio
--normalize off Peak-normalize audio
--trim-silence off Trim leading/trailing silence

Info commands

# List available models with sizes
transcriber-cli --list-models

# Show cached models
transcriber-cli --list-models   # shows cached models at the bottom

# Download a model without transcribing
transcriber-cli --download-model large-v3-turbo

# List supported languages
transcriber-cli --list-languages

Examples

# German podcast, save as WebVTT
transcriber-cli podcast.mp3 --language de --format vtt --output subs.vtt

# YouTube video, translate to English, with word timestamps
transcriber-cli https://youtube.com/watch?v=... --translate --word-timestamps --format json

# Quick draft with a small model on CPU
transcriber-cli meeting.wav --model small --no-gpu

# Noisy recording — clean up audio first
transcriber-cli noisy.mp3 --dc-offset --normalize --trim-silence

Clone this wiki locally