v1.2.0

What's new

• Queue is now a table (file / size / live per-file status); double-click a row to reveal it in Explorer, and dropping a folder enqueues its audio files
• Header band, segmented MVSep API / local engine switch, model filter for the 117+ model list, a determinate N / M progress bar with an elapsed timer and an Open folder button
• Stem downloads run in parallel (3 at a time) instead of one by one

Fixed

• Model option dropdowns (Karaoke model type, extraction type, ...) could read empty: the option values were never sent as add_opt1/2/3, leaving only the server defaults
• No more pointless re-encode: when no mono fold / resample is requested a stem lands byte-identical instead of being recoded
• Elapsed timer no longer rewrites the status line; a bundled ffmpeg.exe next to the app is preferred over PATH; a malformed value in config.json no longer crashes startup

v1.1.0

Auto-fetch the model list on launch when the local cache is empty.

v1.0.0

Windows installer (bundles ffmpeg 9.0). Installs to C:\Program Files\Matteo Barni Photography.