# youtube-multi-language-audio > Open-source Python CLI (MIT) that turns translated subtitle files (SRT/VTT) plus the original video into dubbed, time-synced audio tracks for YouTube's multi-language audio feature: one upload-ready track per language, exactly as long as the video, in a cloned voice (the creator's own, or one per speaker), with re-timed captions and a YouTube Studio upload checklist. Voices are generated by the Miragea voice-cloning engine (https://miragea.ai). ## Install and run - `uvx --from git+https://github.com/RTY2345RR/youtube-multi-language-audio ytmla dub VIDEO video.es.srt video.de.srt --voice my-voice.wav` - Needs `MIRAGEA_API_KEY` (create at https://miragea.ai/workspace/api). Agent setup: Install https://miragea.ai/skills/miragea-audio/install.md - ffmpeg is bundled through imageio-ffmpeg. ## Commands - `ytmla dub VIDEO SUBS... [--voice [NAME=]CLIP|URL] [--voice-text [NAME=]TEXT] [--clone-from-video [NAME=]START-END] [--actor [NAME=]ID] [-o DIR] [--format m4a|wav|mp3|flac] [--max-speedup 1.3] [--two-pass overflow|always|off] [--background STEM] [--multitrack] [--json]` - `ytmla check VIDEO SUBS...`: offline estimate of lines too long for their time slot (no API key). - `ytmla doctor`: checks ffmpeg, key and voice engine. - Language comes from the file name (`video.pt-BR.srt`) or `pt-BR=file.srt`. A folder of subtitle files also works. - `[Name]`, `Name:` or `(Name):` labels in cues select per-speaker voices. ## Output (per language folder) - `