Is that file really 320? The spectrogram doesn't lie.
Lossy encoders cut the top off the spectrum, and re-encoding can't put it back — a '320kbps' file made from a 128 still carries the 128's scar. Drop a file and read its real history.
Free · no account · runs in your browser — nothing is uploaded; your audio never leaves your device.
Know your sources. Keep the good ones forever.
Producers live and die by source quality. AudioEncode in the studio keeps your verified files organized, converted, and tagged in one permanent library. Sign up: 3 full packs free.
How to read the spectrogram
Time runs left to right; frequency runs bottom to top; brightness is energy. Real full-quality audio fades out gradually toward the top. A lossy encode leaves a hard horizontal ceiling instead — a shelf where the encoder said "nobody will miss this" and deleted everything above it. The shelf's height is the file's true birth certificate: around 16 kHz says 128 kbps, around 19 kHz says 192–256, and energy pushing past 20 kHz says genuine 320 or lossless.
The verdict card does that reading for you and calls out the one thing sellers of re-encoded files hope you never check: the label on the box can be rewritten; the shelf cannot. Sample diggers, DJs buying packs, and anyone archiving a collection should make this a habit — thirty seconds here saves a lifetime of dull hi-hats.
The exact rule this page measures with
The tool analyzes the file in 4096-sample frames and sums energy into 1 kHz bands. It takes the mean energy of the 2–8 kHz region as the reference level for the file. A band counts as present while it stays within 55 dB of that reference. The cutoff is the highest 1 kHz band still inside that window, reported as a whole kHz.
So the answer is a band index, not an exact frequency. A shelf that truly lands at 17.4 kHz is reported as 17 kHz. A file whose real edge sits between two bands reads as the band below it, and nothing between bands is guessed.
The shelf is then drawn onto the spectrogram itself: a dashed line at the measured cutoff, the region above it dimmed, and the number written on the image. That line is the evidence, not a caption — it is placed on the same linear frequency axis the spectrogram was drawn on, so you can see for yourself where the energy stops.
The second number on the page is the container's average bitrate, computed as file size × 8 ÷ decoded duration. That is the figure a player prints in its info panel, and it is a claim: it describes how many bits are in the file, not what those bits contain. The page reports both numbers and states when they disagree. It never rewrites, repairs or re-encodes anything.
| Measured cutoff | Verdict the page shows | What it usually means |
|---|---|---|
| 20 kHz, or 1 kHz under Nyquist | Full bandwidth — genuine high quality | True 320, lossless or a clean master |
| Under 20 kHz, in a lossy file averaging 256–400 kbps | Re-encode: a 320 label over a 16 kHz shelf | The container is big; the audio never was |
| 18–19 kHz | ≈192–256 kbps class | A good middle-bitrate encode |
| 15–17 kHz | ≈128 kbps class — check the label | The 128 kbps signature, or an old master |
| Under 15 kHz | Low bandwidth — heavily compressed source | Low bitrate, voice codec, old rip |
a 320 label over a 16 kHz shelf
Take a file sold as 320 kbps whose energy shelves at about 16 kHz. The page reads the container's average bitrate from size ÷ duration, gets roughly 320, measures a 16 kHz shelf, and returns the re-encode branch: a 320 label over a 16 kHz shelf. The same line is drawn on the spectrogram, so the picture and the number agree. A genuine 320 keeps real content past 20 kHz, so it lands in the full-bandwidth branch and no re-encode banner appears.
The label is a claim written into the file; the shelf is what the audio actually contains. A mismatch between the two is the tell you are looking for, and comparing them is what this page now does for you. It reports the average bitrate and the measured shelf and names the mismatch. It does not — and cannot — tell you the bitrate the file was originally encoded at, and it never claims a specific source.
Why a local check beats a portal
Your file is decoded inside the tab and never leaves the device. Nothing is uploaded, so there is no queue to wait in and no copy on a server. The spectrogram stays on screen while you scrub it and listen for where the shelf bites.
When you have the answer, press Download verdict card. You get a 1200 × 630 PNG carrying the container bitrate, the measured shelf, the verdict, the file name and a strip of the analysed spectrogram with the shelf line drawn across it. Post it in a thread where someone is selling a pack, or keep it beside the file as evidence.
Where the method is blind
A legitimately low-passed source reads low without being a bad encode. An old master, vinyl-sourced material or a synth-only mix may stop near 16 kHz by design. The verdict cannot tell that apart from a cheap 128 kbps shelf — a low-shelved file honestly labelled and a low-shelved file sold as 320 look the same here, which is why the re-encode branch names the numbers rather than declaring an origin.
The container bitrate is inferred from file size, so embedded cover art or a large ID3 tag inflates it, and a variable-bitrate file only ever yields an average. The 256–400 kbps window is deliberately loose for that reason: it detects "this claims to be a 320" without pretending to be an exact meter. Re-encode wording appears only for lossy containers (MP3, M4A/AAC, OGG/Opus); a WAV or FLAC is never accused, because a lossless file's size always matches its duration.
A file shorter than about a tenth of a second produces no frame at all, and the page refuses to guess. Anything just above that is judged from a single 4096-sample frame. A file with a loud midrange raises the reference and can drag the cutoff under the true edge. A noisy or dithered master can read high, because broadband hiss lifts the top bands.
Read the number as a measurement of this file's spectrum, not as proof of where it came from.
How to check an MP3's real quality
- Load the fileDrop the MP3 (or any audio file) on the checker.
- Read the cutoffThe tool measures where the file's high-frequency content stops and shows the spectrum.
- Read the verdictCompare the measured cutoff with the bitrate the file claims to be.
- Save itPress Download verdict card to keep a PNG of the result.
Frequently asked questions
How can you tell an MP3's real quality?
Lossy encoders discard high frequencies to save space: 128 kbps typically cuts everything above ~16 kHz, 192 around ~18–19 kHz, and 320 keeps content past 20 kHz. That cutoff shelf is permanent — the tool measures where your file's energy actually stops and compares it to what the file claims to be.
What is a 'fake 320' or upscaled MP3?
A low-bitrate file re-encoded at a higher bitrate. The new file is bigger and labeled 320 kbps, but the audio still ends at the old cutoff — bigger box, same loss. They're everywhere on file-sharing sites and old sample folders.
Can a real 320 kbps file still fail the check?
Rarely, honestly: some legitimate sources are low-pass filtered (old masters, vinyl rips, certain synth-only tracks with little natural high end). The verdict card reports the measured cutoff and lets you eyeball the spectrogram — the shelf's shape usually settles it.
Does this work on WAV and FLAC files?
Yes — and it's the most fun there. A WAV or FLAC with a hard 16 kHz shelf was made from a lossy source; 'lossless' packaging doesn't launder a lossy past.