subgen

subgen creates subtitles from audio and video for personal media libraries. It's useful for films and shows that lack usable subtitles, with automatic generation tied to your media server.

• Media-server automation: Bazarr can use subgen as a Whisper provider, while Plex, Jellyfin, Emby and Tautulli can trigger generation when media is added or played.
• Transcription and translation: Generate SRT subtitles or LRC files from speech in its original language, or translate foreign-language audio into English.
• Subtitle timing: When it can access the source video, subgen compensates for audio stream start offsets that would otherwise make subtitles appear too early.
• Compatible audio API: Open WebUI, compatible Obsidian plugins and other clients that use the OpenAI Whisper API can request transcription or English translation, with text, subtitle or timestamped responses, including word-level timing.

Speech recognition runs locally through faster-whisper and stable-ts, using a CPU or an NVIDIA GPU with CUDA. subgen runs in Docker or as a standalone Python application and is open source under the MIT license. Its audio API uses the configured Whisper model rather than the model requested by the client.