Development release. The notes below record changes at this version; consult current help for today’s availability and limitations.
Track-by-track AI remaster action with an automatic brief, local sampled spectral profile, bounded suggestions, rendering and output checks before application. Manual adjustment, preview-first suggestions, matched comparison and Undo retained.
Source format, loudness range, short-term maximum, over-full-scale percentage and editable WAV genre/date context added to the consented request. Unknown context stays unknown; no audio or identity tags are sent.
Studio processing and a -1 dBTP (or stricter) measured ceiling enabled by the automatic action; Undo restores prior quality and ceiling choices.
Added tonal/anti-phase/rate fixtures, metadata parsing and privacy boundaries, gateway validation and automatic browser workflow regressions. Provider remains simulated in tests; listening and live-provider quality evaluation are outstanding.