How to tell if a FLAC is fake

A file can say 24-bit/96kHz on the label and still be an MP3 that somebody re-encoded into a FLAC container. Your player shows the badge and has no idea. Here is what actually gives it away.

Last updated 16 September 2026

What a "fake FLAC" is

A fake FLAC is a file in a lossless container whose audio came from a lossy source. Somebody decoded an MP3 or an AAC and re-encoded the result as FLAC. The container is genuinely lossless — it will reproduce the audio inside it perfectly. The problem is what is inside: everything the lossy encoder threw away is already gone, and re-encoding does not bring any of it back.

The same trick happens one level up, with sample rate and bit depth. A 44.1kHz/16-bit master resampled to 96kHz/24-bit and sold as hi-res is upscaled: the numbers on the label are real, the information is not.

Most of the time you cannot hear it. A 320 kbps source on ordinary gear sounds fine, which is exactly why these files circulate for years. Detection is a measurement problem, not a listening test.

The signals that actually give it away

1. Where the top end stops

Lossy encoders discard the highest frequencies first, and they do it with a hard limit. A 320 kbps MP3 usually stops dead around 20kHz; 128 kbps stops nearer 16kHz. On a spectrogram that looks like a ruled line across the top of the image with nothing above it.

2. Whether that stopping point makes sense

A 96kHz file can physically carry content up to 48kHz. If it carries nothing at all above 16kHz, the extra sample rate is doing no work. The question is never "where is the cut-off" on its own — it is "does this cut-off make sense for what the file claims to be".

3. Whether the edge sits in the same place all the way through

This is the signal most single-number tools miss. Real recordings breathe: a cymbal crash reaches higher than a quiet passage, and the top of the spectrum wanders from second to second. An encoder's limit does not wander. If the same edge sits at the same frequency in the intro, the chorus and the fade-out, something applied it mechanically.

Why a single cut-off number lies

Early versions of almost every detector — including ours — judged on cut-off alone, and they all share the same failure: they call genuine live recordings fakes. A field recording made with modest microphones may hold nothing above 17kHz because there was nothing above 17kHz to record. That file is lossless and honest, and a cut-off rule brands it a transcode.

A detector that cries wolf is worse than no detector, because you start deleting real files. The useful behaviour is to weigh several signals, and to say "I cannot tell" when the evidence does not point either way — rather than guessing to look decisive.

Tools that check for you

Prices and platforms as of September 2026.

The desktop tools are the mature ones, and if your library lives on a computer they are the obvious place to start. The gap they leave is the phone: the files you actually carry around, bought from a store on the device itself, have until recently had nothing to check them with.

You found a fake. Now what?

• If you bought it, ask for a refund and say what you measured. Stores handle this better than people expect, and hi-res sellers depend on their reputation for authenticity.
• If it came from a rip you made, check the source. A scratched disc or a failing drive can produce a file that reads as suspicious for reasons that have nothing to do with transcoding.
• Do not delete on one signal. Look at what the verdict was actually based on before throwing away something you cannot replace.
• Re-encoding will not fix it. Converting a transcode back to a bigger format only makes a bigger transcode.