Home

Transcribe CLI

Offline meeting transcription (lwt) from audio and video using faster-whisper. Runs 100% on your machine.

Everything goes through lwt — no manual pip extras, env vars, or huggingface-cli.

Wiki

Page	Description
Installation	Install package, ffmpeg, PATH fixes
First-Time Setup	lwt setup wizard and model download
Usage	Basic transcription examples
Commands	Full CLI reference
HuggingFace Diarization	Required for --diarize — account, licenses, token
Ollama	Cleaning, translation, summarization
GPU and CUDA	lwt install cuda, GPU acceleration
Configuration	config.toml and lwt config
Troubleshooting	Common errors and fixes
Development	Tests and project structure

Quick start

git clone https://github.com/Klema4/transcribe-cli.git
cd transcribe-cli
pip install -e .

lwt setup
lwt transcribe meeting.mp4

Features

• Audio (WAV, MP3, FLAC, …) and video (MP4, MKV, AVI, …)
• Export: TXT, SRT, VTT, JSON
• GPU via CUDA (auto fallback to CPU)
• Optional speaker diarization (--diarize) — see HuggingFace Diarization
• Optional clean / translate / summarize via Ollama — see Ollama
• Rich terminal UI

License

MIT — see repository