About this project

Traccia is a browser-based editor for correcting subtitles on a timeline divided by speaker. It specializes in human correction following automatic transcription and includes the following features: ### Main Features - **Intuitive Editing**: Efficient editing is possible via drag-and-drop for duration adjustment and speaker changes, as well as shortcut keys for splitting, merging, and aligning playback positions. - **Validation Functions**: Displays warning marks for overlapping subtitles within the same speaker or for display speeds that are too fast to read. - **Transcription Integration**: Supports paid transcription via Gemini API, as well as free local transcription using faster-whisper and pyannote (via the `transcribe` command). - **Version Control and Comparison**: Raw output, finalized versions, and versions imported from Filmora (.wfp) can be saved separately and compared side-by-side to check differences. - **Export**: Allows exporting .srt files for each speaker. - **Analysis and Learning**: It is possible to measure conventions, such as line length, from corrected subtitles and reflect these in future transcriptions. ### Workflow 1. Prepare .srt files (created via existing files, Gemini API, or local commands). 2. Launch the editor with `traccia edit` and make corrections in the browser. 3. Export as speaker-specific .srt files and finalize using editing software such as Filmora.