FLAC vs MP3: What Is Different, and Can You Hear It?

The row that matters most is the first one. FLAC is a container that promises to give back exactly what it was given — which is why it is the right format for the copy you never want to touch again. MP3 makes no such promise, and in exchange it fits a whole album into the space of three FLAC tracks.

An MP3 encoder is not a compressor in the ZIP sense. It runs a psychoacoustic model — a mathematical description of what human hearing fails to notice — and spends its limited bits only on the parts of the signal you are likely to perceive. Three mechanisms do most of the work:

• Masking. A loud sound hides quieter sounds near it in frequency and in time. The encoder identifies what is masked and stores it crudely or not at all. This is where most of the savings come from, and it is why the result can be indistinguishable despite deleting the majority of the data.
• The lowpass filter. Everything above a cutoff is discarded outright. LAME sets that cutoff around 16 kHz at 128 kbps, roughly 19 kHz at 192 kbps, and about 20.5 kHz at 320 kbps. Adult hearing usually rolls off well before 20 kHz, which is why the highest setting effectively costs nothing.
• Joint stereo. Instead of two independent channels, the encoder can store a mid channel plus the difference between the two. Since most music is heavily correlated between left and right, this frees up bits for detail elsewhere.

None of this is reversible. The discarded data is not hidden somewhere in the file — it is gone, which is also why converting an MP3 back to FLAC restores nothing and only inflates the file.

“Transparent” is the term for an encode that listeners cannot separate from the source under blind testing. It is not a marketing word; there is a testing procedure behind it. Comparing MP3 against FLAC is really a question of which bitrate:

Note what the argument is not about. It is not about the format label, the price of your DAC, or whether the source was 24-bit. Those points dominate forum threads and lose blind tests. Bitrate is the variable that decides whether an MP3 is separable from its FLAC.

Transparency at 320 kbps is a statement about typical music, not a guarantee about every waveform. The failure cases are consistent, and knowing them tells you when keeping the FLAC actually matters:

• Sharp transients. Castanets, applause, harpsichord, glockenspiel, and triangle hits are the classic problem material. MP3 encodes audio in blocks, and a sudden attack inside a block can smear quantisation noise backwards in time — an artifact called pre-echo, heard as a faint swish just before the strike.
• Solo instruments in silence. Artifacts hide inside dense mixes. A single voice or acoustic guitar with nothing else to mask the noise is a much harder test than a loud rock chorus.
• Deliberate killer samples. The Hydrogenaudio community collected a set of clips engineered specifically to break lossy codecs. They defeat MP3 at 320 kbps. They are also not music anyone listens to for pleasure.
• Anything you will process further. Pitch-shifting, time-stretching, heavy EQ, and re-encoding — DJ sets, video editing, sampling — amplify artifacts along with everything else. Work from the FLAC in those cases and encode to a lossy format exactly once, at the end.

Sighted comparisons are worthless — knowing which file is playing is enough to make anyone hear a difference. The fix is an ABX test, where the software hides the answer from you:

1. Convert one FLAC you know well to 320 kbps MP3, then decode that MP3 back to WAV so both files are the same format and the comparison is purely about the audio.
2. Install foobar2000 and its ABX Comparator component on Windows, or an equivalent ABX tool on macOS or Linux.
3. Load both files, run at least 16 trials, and use headphones in a quiet room. Loop a short passage with cymbals or applause rather than a whole track.
4. Score 12 or more out of 16 and you genuinely hear a difference — keep that kind of material lossless. Score around 8 and you are guessing, which is the result almost everyone gets at 320 kbps.

Either outcome is useful. One tells you where to spend storage; the other frees you to fill a phone with music instead of arguing about codecs.

Fidelity is where the debate happens; compatibility is where people actually get stuck. FLAC support has improved a lot — Windows, Android, modern browsers, and recent Apple devices all handle it — but the gaps sit exactly where you listen:

• Car head units. USB playback on anything more than a few years old is frequently MP3 and WMA only. A folder of FLACs simply does not appear.
• The Apple ecosystem. iOS plays FLAC through the Files app, but the Music library still will not import it, so a FLAC collection cannot be synced as a library the way MP3 or ALAC can.
• Older portable players, TVs, and set-top boxes. Anything with a fixed codec list from the MP3 era plays MP3 and stops there.
• Bluetooth headphones. Every Bluetooth codec — SBC, AAC, aptX, LDAC — is itself lossy, so your phone decodes the FLAC and re-encodes it before it reaches the headphones. On wireless gear the source format barely registers.

That last point is worth sitting with. A large share of listening now happens over a lossy wireless link, which means FLAC's advantage often disappears somewhere between the phone and the ears, regardless of what the file on disk contains.

The FLAC-or-MP3 question assumes you have to choose. You do not, and the people with the best-organised libraries never do. Keep FLAC as the archive — on a drive, a NAS, wherever storage is cheap — because it is the copy you can re-encode to any future format without accumulating damage. Generate MP3s from it for the phone, the car, and the old player, and treat them as disposable: if you want a different bitrate next year, delete them and make new ones in minutes.

The only mistake in this arrangement is deleting the FLACs after converting. Every other decision here is reversible; that one is not.

For step-by-step methods — browser, VLC, fre:ac, ffmpeg — see How to Convert FLAC to MP3 (4 Free Ways), and for how the desktop tools compare on batch handling and tags, the seven-converter comparison.