Spectro
Detect fake lossless audio files on macOS
After getting burned by a fake WAV at a gig — a 320kbps MP3 repackaged as AIFF that sounded brittle on a Funktion-One — I decided to build a tool to catch this automatically.
The problem: Record pools and digital stores routinely distribute lossy audio in lossless containers. The file looks right, the size is plausible, but the frequency ceiling gives it away. A real lossless file has energy up to 22kHz. An upconverted MP3 has a flat cutoff at a suspiciously round number — 16kHz, 18kHz, 19.5kHz — that no natural recording produces.
The existing tool for this (Spek) hasn't been updated in years. So I built Spectro.
What I shipped:
FFT-based classifier: Lossless / Lossy / Fake Lossless
Batch analysis — drag a folder of 100 tracks, get verdicts in seconds
Finder extension + Quick Look plugin (the macOS integration took longer than the core algorithm)
100% offline — relevant when checking unreleased promo tracks
$39 one-time, no subscription
FREE TRIAL: 25 tracks to see how it works!
What I learned:
The hardest part wasn't the FFT — it was tuning the verdict thresholds for edge cases. Old recordings, vinyl rips, and heavily compressed masters can look like fake lossless when they aren't. I've been iterating on real-world test cases and the false positive rate on modern material is very low, but older recordings are still tricky.
The macOS distribution pipeline (Developer ID, notarization, Sparkle for auto-updates, Lemon Squeezy for payments) took more time than expected for a first native app. Plenty of undocumented edge cases.
Where I'm at: launched today. First sales coming in. Focused now on building awareness in DJ and audiophile communities where the problem is felt most acutely.
Happy to answer questions about the spectral analysis approach, the macOS distribution setup, or anything else. And if you work with audio and want to check your library, https://getspectro.app
2 Comments
DJs often don't realize they are playing low-quality files until they hear that brittle sound on a high-end club system. It is incredibly frustrating to pay for premium lossless tracks only to find out they were just upconverted MP3s with a fake frequency ceiling. Since tuning the thresholds for vinyl rips and old recordings is so tricky how do you ensure the classifier doesn't accidentally flag a genuine vintage track as fake?
hey! Don't use the frequency ceiling as your primary signal. Instead, look at what's above it. A genuine vinyl rip always has analog noise texture up there tape hiss, surface noise even when the musical content has rolled off. An upconverted MP3 has near digital silence above its encoding cutoff. That distinction is reliable regardless of how low the ceiling sits.
If you need a second signal, check whether the cutoff aligns with a standard MP3 bitrate boundary 16khz for 128kbps, 19khz for 192kbps. A vintage track rolls off at a frequency determined by the original recording chain, not by a codec spec. Those two clues together noise floor texture plus suspiciously "round" cutoff frequency, let you flag fakes without penalizing old records for being old.