add-srt-next-to-videos 0.1.2

Generate .srt subtitle files next to video files using local MLX Whisper and LLM translation

add-srt-next-to-videos

Generate .srt subtitle files next to video files. Point it at a file or folder, get subtitles.

Runs entirely on your Mac — MLX Whisper handles speech-to-text, a local LLM handles translation. Nothing leaves your machine.

Requirements

• Apple Silicon Mac (M1/M2/M3/M4) — MLX only runs on Apple Silicon
• Python 3.10+
• pipx for system-wide install
• For translation: any local LLM server with an OpenAI-compatible /v1/chat/completions endpoint — LM Studio, Ollama, etc.

Install

# Homebrew
brew install yuyangchee98/tap/add-srt-next-to-videos

# From PyPI
pip install add-srt-next-to-videos

# Or with pipx (isolated install, recommended)
pipx install add-srt-next-to-videos

# Or directly from GitHub
pipx install git+https://github.com/yuyangchee98/add-srt-next-to-videos.git

After install, create a .env file in the directory where you run the command (or your home directory):

cp .env.example .env
# edit .env with your model preferences

To uninstall: pipx uninstall add-srt-next-to-videos

For development:

git clone https://github.com/yuyangchee98/add-srt-next-to-videos.git
cd add-srt-next-to-videos
uv venv && uv pip install -e .

Configuration

Copy .env.example to .env and edit:

# Whisper model for speech-to-text (HuggingFace repo)
WHISPER_MODEL=mlx-community/whisper-large-v3-mlx

# LLM for translation (any server with an OpenAI-compatible /v1/chat/completions endpoint)
# Works with: LM Studio, Ollama, vLLM, llama.cpp server, etc.
LLM_URL=http://localhost:1234
LLM_MODEL=qwen3-32b-mlx

Whisper model weights (~3GB) download automatically on first run and cache in ~/.cache/huggingface/.

The LLM is only used for translation. If you only need transcription (source language subs), you don't need an LLM server running.

Usage

# Transcribe Japanese audio -> Japanese subtitles
add-srt-next-to-videos ./videos -l ja --subs ja

# Translate to English
add-srt-next-to-videos ./videos -l ja --subs en

# Multiple languages at once
add-srt-next-to-videos ./videos -l ja --subs ja,en,zh

# Dual-language subs (both languages in one file, for language learning)
add-srt-next-to-videos ./videos -l ja --subs ja-en

# Mix and match
add-srt-next-to-videos ./videos -l ja --subs ja,en,ja-en,ja-zh

# Single file
add-srt-next-to-videos ./episode01.mp4 -l ja --subs ja,en

Works on single files or entire directories (recursive). Existing .srt files are skipped unless you pass --overwrite.

Output files

The sub spec determines the filename:

video.mp4  ->  video.ja.srt       (transcription)
video.mp4  ->  video.en.srt       (translation)
video.mp4  ->  video.ja-en.srt    (dual: japanese on top, english below)

A dual-language .ja-en.srt looks like this:

1
00:00:05,000 --> 00:00:10,679
母ちゃん母ちゃん今日はどこ行くの
Mom, mom, where are we going today?

2
00:00:10,679 --> 00:00:12,060
美容院よ
The beauty salon.

How it works

1. Whisper transcribes the audio once per video — timestamped segments in the source language
2. LLM translates each segment's text to other languages (only if needed)

Whisper handles transcription only. All translation goes through your local LLM for better accuracy.

The tool transcribes once and caches the result per video, so --subs ja,en,ja-en runs Whisper once and translates to English once, even though two outputs use the English translation.

CLI reference

add-srt-next-to-videos [-h] [--language LANGUAGE] [--subs SUBS] [--overwrite] path

Flag	Short	Default	What it does
path		(required)	Video file or directory (recursive)
--language	-l	auto-detect	Source language (ja, en, zh, etc.)
--subs	-s	source lang	What to generate: ja, en, ja-en, or comma-separated
--overwrite		off	Regenerate existing .srt files

Whisper model and LLM settings are configured in .env (see Configuration).

Supported video formats

.mp4 .mkv .avi .mov .webm .flv .wmv .m4v .mpg .mpeg .ts .vob

Notes

• If a run gets interrupted, just re-run the same command — it skips videos that already have their .srt files

• Language codes are ISO 639-1. Whisper supports these languages for transcription:

  af Afrikaans, ar Arabic, hy Armenian, az Azerbaijani, be Belarusian, bs Bosnian, bg Bulgarian, ca Catalan, zh Chinese, hr Croatian, cs Czech, da Danish, nl Dutch, en English, et Estonian, fi Finnish, fr French, gl Galician, de German, el Greek, he Hebrew, hi Hindi, hu Hungarian, is Icelandic, id Indonesian, it Italian, ja Japanese, kn Kannada, kk Kazakh, ko Korean, lv Latvian, lt Lithuanian, mk Macedonian, ms Malay, mr Marathi, mi Maori, ne Nepali, no Norwegian, fa Persian, pl Polish, pt Portuguese, ro Romanian, ru Russian, sr Serbian, sk Slovak, sl Slovenian, es Spanish, sw Swahili, sv Swedish, tl Tagalog, ta Tamil, th Thai, tr Turkish, uk Ukrainian, ur Urdu, vi Vietnamese, cy Welsh

  Translation to any language depends on your LLM's capabilities

License

MIT