I built a tool that detects fake lossless and replaces it from Soulseek - free, open source
I got tired of finding out half my FLAC collection was just re-encoded MP3s. You know the deal, file says 1411 kbps but the spectrum cuts off at 16 kHz.
So I built DDD. It scans your library and ranks every file by its real spectral cutoff (FFT analysis, not tags). Anything that's fake or below your bar, it goes and finds a real version on Soulseek automatically. FLAC/WAV/AIFF first, MP3 320 fallback. Then it re-checks the download by spectrum before keeping it. The declared format is only used for the search, never for the keep/reject decision.
It also sorts your tracks by genre from the audio itself, not the filename. Uses a local ML model so even an untagged Track_01.flac ends up in the right folder.
Three presets: DJ Club (>=18 kHz, includes real 320), Audiophile (>=20 kHz), Purist (full lossless only). MP3s below 320 always rejected.
Windows app, double-click and go. No Python, no ffmpeg, no terminal. CLI too if that's your thing.
Free, open source, no account, no cloud.