Inspect audio

Fake FLAC Checker & Spectrogram

View a local audio spectrogram and estimate its frequency cutoff to look for signs of a lossy source inside a FLAC or WAV file.

Runs locally in your browser

Local workspace

Process a local file

No media upload

Choose a file Select from your device or drop it here.

On-device

Your selected media stays on this device. AudioFetcher loads its processing engine, reads the file in browser memory, and creates the result locally.

Temporary

Refresh or close the tab and the selected file and generated result leave browser memory.

Honest limits

A cutoff is a clue, never proof of authenticity or a particular bitrate. Natural roll-off, noise, quiet passages and mastering can mislead the estimate. Files up to 64 MB; at most the first 3 minutes are analysed.

How to use Fake FLAC Checker & Spectrogram

1
Choose a local audio file up to 64 MB. Keep the original; this checker does not modify it.
2
Select Analyse spectrogram. The tool samples at most the first three minutes and shows progress while decoding and analysing.
3
Read the cutoff hint alongside the coloured frequency plot. Compare several recordings before drawing conclusions.
4
Download the PNG if you want to compare the image later. The image includes the frequency and time axes.

What this tool changes

View a local audio spectrogram and estimate its frequency cutoff to look for signs of a lossy source inside a FLAC or WAV file.

How to read the audio spectrogram

A spectrogram displays the contents of an audio signal across time. Time runs from left to right; frequency runs from the bottom to the top. The upper edge is the Nyquist frequency, half the analysis sample rate. A 44.1 kHz analysis therefore stops at 22.05 kHz. Bright areas contain more energy than dark areas. A sustained note appears as a horizontal line, while a drum hit can produce a broad vertical stripe.

This display uses short overlapping analysis windows, a logarithmic energy scale and a fixed 90 dB range below the strongest observed band. The colours are relative to that passage, so two pictures with different loudness should not be treated as calibrated measurements. The first three minutes are represented by a limited number of time columns to keep memory use modest. Very brief events can fall between sampled windows. The Audacity spectrogram manual explains the trade-off between time and frequency detail if you want a closer inspection.

What the estimated cutoff tells you

The estimate looks for frequency bands that repeatedly contain energy above an estimated floor, rather than using one isolated bright pixel. A visible shelf means energy drops abruptly above a particular frequency. The checker only offers a lossy-source hint when there is enough broadband material and a substantial drop. A pure tone, silence or a narrow voice recording is usually inconclusive.

A sharp edge around 15–16 kHz can resemble some roughly 128 kbps lossy encodings. An edge near 19–20 kHz can resemble a higher-bitrate lossy source. These are comparison clues, not a table for identifying an encoder. Codec, encoder settings, sample rate and the original recording all affect the result. An MP3 encoder can retain or discard different frequencies at the same bitrate. Full-looking energy near 21–22 kHz means there is no obvious shelf in that passage; it does not certify lossless provenance.

Why converting MP3 or YouTube audio to FLAC cannot restore it

FLAC preserves the audio samples supplied to its encoder. That says nothing about how those samples were created. Decoding an MP3 and saving it as FLAC prevents another lossy encode, but any information removed earlier remains missing. The new file can be a valid FLAC while still having a lossy source. The same applies when a compressed stream is converted from a video platform.

Do not confuse the compression format with the history of the recording. A large file, high sample rate or 24-bit tag is not evidence that the music started that way. Upsampling and adding empty precision do not reconstruct the original signal. See YouTube to FLAC and its quality limits for that workflow, or use the audio bitrate checker to inspect how a file is encoded. This tool deliberately avoids stamping a recording “real” or “fake”.

Limits and misleading results

A genuine master may roll off early because of its microphones, recording equipment, filtering or artistic choices. Some older recordings contain little high-frequency sound. Quiet intros are poor test passages. Added hiss, noise or dither may fill the upper spectrum and hide an earlier lossy cutoff. A file assembled from several sources may change quality halfway through; examining the beginning cannot establish the rest.

Channels are averaged for analysis. Opposite-phase material can partially cancel during that mix, so an unexpected result should be checked in a desktop editor with channels viewed separately. The source bit depth is shown when a readable header or decoder report supplies it; otherwise it remains unavailable. Average file bitrate includes tags and container overhead. The analysis rate can differ from the original rate when decoding or a browser resamples the file. Check both figures before interpreting the top edge.

Other ways to investigate a suspicious file

For a valuable purchase or archive, compare the file with a trusted release rather than relying on a single graph. Spek provides a dedicated desktop spectrogram. Audacity lets you change the window size, inspect separate channels and choose another passage. Those options are useful when this quick checker reports inconclusive results or you need to examine a long recording in detail.

Keep notes about where the file came from, listen for audible problems and preserve your original. Changing formats to force a better-looking picture defeats the purpose of the check. If the recording itself is narrow-band, no software can infer frequencies it never contained. This local tool gives you a practical visual starting point; provenance and a careful comparison remain more useful than a confident label produced from one cutoff.