Self-host a music library long enough and you end up with a pile of FLACs — from your own rips, from downloads, from the questionable corners of the internet where hi-res music lives. FLAC means lossless. Except a file’s extension is a label, not a promise, and once you start looking, a surprising number of them are lying.

There are two lies, and they’re both quietly common.

The first: a FLAC can be laundered lossy audio. Someone takes an MP3 — which already threw away most of the data — decodes it, and re-encodes the result as FLAC. The container is now perfectly honest: it is a lossless copy. Of a thing that already lost. You’ve wrapped a compressed file in a lossless box and slapped “lossless” on the box.

The second: “hi-res” is easy to fake. A 24-bit/96kHz file sounds like it should hold more than a 16-bit/44.1kHz CD — more bits, more frequencies. But you can upsample a plain CD rip to those numbers, padding in bit depth and sample rate that were never recorded. The label is real. The information behind it isn’t.

The metadata lies; the spectrogram doesn’t

The tell is in the frequencies. A genuinely lossless file carries real energy up to the Nyquist limit of its sample rate. A lossy source leaves a brick wall — a hard horizontal shelf on the spectrogram where the encoder guillotined everything above roughly 16–20kHz to save space. Fake hi-res shows that same wall sitting far below where a true 96kHz file should reach: all that headroom, nothing in it.

You can’t hear the difference reliably on most gear. You can see it instantly. The metadata can claim anything; the picture of the actual signal can’t.

So I built the thing that checks

That verification is the seed of trueres — a set of browser-based audio tools I built, all of it client-side, because when the files in question are your own library the last thing you want is to upload them somewhere to find out if they’re real. It draws the spectrogram, flags the frequency cutoff, verifies whether a “hi-res” file has any actual high-resolution content, and does the rest of the finicky audio-forensics work — loudness (LUFS), clipping, format checks — that answers one question: is this file what it says it is?

I care about that question because I stepped off the streaming treadmill and keep my own library, on my own hardware — a self-hosted setup feeding a Sony STR-DN1080 receiver and a Dayton Audio 7.2 room with a 4K TV in front of it. Owning your media instead of renting a stream is mostly a gift, but it hands you one job the platform used to do: guaranteeing the file is genuine. When you own it, “is this the real thing?” is yours to answer. trueres is me answering it. (And yes — there is a drawer of USB-C DACs and dongles behind this sentence. The rabbit hole is real and slightly ridiculous, and I went all the way down.)

Most people will never need any of this. A laundered FLAC sounds completely fine on earbuds, and the difference I’m chasing lives right at the edge of audibility — this is a tool for the rounding error, and I’m the rounding error. But if you’ve ever downloaded something labeled “24-bit/96kHz FLAC,” felt a flicker of doubt, and wished you could just look, that’s what this is. It runs in your browser and your files stay on your machine.

The file says lossless. The spectrogram says otherwise. I built the thing that tells you which one to believe.

Sources & further reading

• Shannon — Communication in the Presence of Noise — the sampling theorem behind the Nyquist limit.
• Brandenburg — MP3 and AAC Explained — why lossy encoders low-pass the highest frequencies.
• Sony LDAC — up to 990 kbps, and still a lossy link.