Translation Configuration

Configure translation backends for multilingual subtitle generation. Choose between AI-powered and built-in translation services.

Translation Backends

Backend	Quality	Cost	Speed	Notes
LLM	Highest	Paid (API)	Moderate	Context-aware, reflection support
Bing Translate	Good	No key needed	Fast	No setup required
Google Translate	Good	No key needed	Fast	May require VPN in some regions
DeepLX	High	Self-hosted	Fast	Requires self-hosted DeepLX server

LLM Translation (Recommended)

Uses your configured LLM for context-aware translation. The LLM understands the full meaning of each subtitle and produces natural, fluent translations.

Terminal

videocaptioner subtitle input.srt --translator llm --target-language ja

Reflection Translation

The "translate-reflect-translate" mechanism performs two passes:

First pass: Initial translation
Reflection: LLM reviews and identifies issues (redundancy, unnatural phrasing)
Second pass: Refined translation incorporating reflections

Terminal

videocaptioner subtitle input.srt --translator llm --target-language zh --enable-reflection

Cost note

Reflection translation uses 2-3x more tokens but produces noticeably better results, especially for nuanced content. Worth enabling for important projects.

Built-in Translation Options

These providers work without an API key:

Terminal

# Bing Translate (no API key, no setup)
videocaptioner subtitle input.srt --translator bing --target-language en

# Google Translate (no key, may need VPN)
videocaptioner subtitle input.srt --translator google --target-language fr

Supported Languages

VideoCaptioner supports translation to 37 target languages including: English, Chinese, Japanese, Korean, French, German, Spanish, Portuguese, Russian, Arabic, Hindi, Thai, Vietnamese, Indonesian, Turkish, Polish, Dutch, Italian, and more.