video-translation

The video-translation skill translates the spoken content of a video from one language into another and produces a dubbed version while preserving the original video stream. It is designed for situations where users want to consume video content in a preferred language without manually creating subtitles, generating voiceovers, or editing media files. The workflow downloads a video and its subtitles, translates the subtitle content into a target language, generates timeline-aligned speech using text-to-speech (TTS), and replaces the original audio track with the newly generated dubbed audio.

The skill provides an end-to-end workflow for video localization. It can download videos and subtitle files through the youtube-downloader skill, translate subtitle text while preserving SRT structure and timing, generate dubbed audio from the translated subtitles using the tts skill, and merge the generated audio back into the original video. The workflow supports specifying source and target languages, uses subtitle timestamps to maintain synchronization, and can leverage the original video audio track as a reference for voice cloning and emotion matching. It also supports an optional reference audio source for voice cloning scenarios.

This skill is intended for users who want to watch or share videos in a different language, including multilingual audiences, content consumers, educators, and creators who need localized video versions. Common use cases include translating foreign-language YouTube videos, creating dubbed versions of existing content, and making video material more accessible to speakers of different languages. The skill is most suitable when subtitle data is available and a complete translated video with replaced audio is required.