WAV vs MP3: which one do you actually need?

One of these discards sound on the assumption you will not miss it. The other keeps everything and charges you about ten megabytes a minute for the privilege. Which is right depends entirely on what happens to the file next.

Updated August 1, 2026

What each one actually is

MP3 is a lossy format. Its encoder decides which parts of the sound you are least likely to notice and throws them away, then compresses what is left. That is where the roughly tenfold size reduction comes from — not from clever packing, but from deletion.

WAV is not really a compression format at all. A WAV file is a 44-byte header followed by the raw samples, one after another. That is the whole design. It stores what the microphone measured, unmodified.

So the comparison is not "which compresses better". One of them compresses and the other one does not.

The size difference is arithmetic, not opinion

A WAV file's size follows directly from three numbers: how many samples per second, how many channels, and how many bytes per sample. At CD quality that is 44,100 × 2 × 2 = 176,400 bytes every second — about 10.6 MB per minute, or 42 MB for a four-minute song.

The same song as a 192 kbps MP3 is around 5.8 MB. Nothing about the music changes that ratio; it is fixed by the format.

This predictability is occasionally the point. A WAV file's length tells you exactly where every sample sits, which is why editors and analysis tools like it.

Converting MP3 to WAV does not restore anything

This is the single most common misunderstanding about these two formats. The detail an MP3 encoder discarded is not hidden somewhere in the file waiting to be decompressed — it is gone. Decoding to WAV produces a larger file containing exactly the same audio.

That does not make the conversion pointless. Plenty of software only accepts WAV, and giving it one is a perfectly reasonable thing to do. Just understand you are re-packaging, not repairing.

It is also why our audio extractor never re-encodes. Pulling audio out as M4A copies the original compressed data into a new container byte for byte; pulling it out as WAV decodes it once to raw samples. Neither path adds a second generation of compression on top of the first.

When WAV is the right answer

Anything where the file will be processed again. Every trip through a lossy encoder adds another generation of damage, and those stack — so if you are going to edit, mix, filter or feed the audio to a model, you want the version that has not been through the mill.

Our stem separator writes WAV for precisely that reason. Its four outputs exist to be loaded into an editor, and handing you MP3s of an estimate would compound one approximation with another.

Archiving is the other case. If this is the only copy that will ever exist, keep the one that has not thrown anything away.

When MP3 is the right answer

Listening, sending, and storing a lot of it. At 192 kbps and above, on ordinary equipment, most people cannot reliably pick the MP3 out of a pair — and it is one tenth the size.

Compatibility is the other half. MP3 opens in essentially everything, including hardware old enough to predate whatever else you were considering.

We will not tell you what your ears can hear; that is not something we can measure for you. We can tell you the file is ten times smaller and that the difference is deletion rather than degradation.

One awkward thing about MP3 files themselves

MP3 has no container. It is a bare run of frames, each with its own small header, laid end to end. There is no index and no table of contents, so software that wants to work with an MP3 has to scan for frame boundaries rather than look anything up.

We had to write that scanner ourselves, and two details caught us out. An ID3v2 tag at the front of the file stores its own length in "syncsafe" bytes that only use seven bits each, so reading it as an ordinary number lands you in the middle of the audio. And the very first frame is often a Xing or Info frame carrying metadata rather than sound — decode it and you get a burst of silence at the start.

The practical consequence is that MP3 is poor at precise seeking and its reported duration is frequently an estimate. WAV has the opposite property: no compression to unpick, so every position is exact.

Do it now, without uploading anything

Is WAV higher quality than MP3?

WAV is unmodified, which is not quite the same claim. If the WAV was made from an MP3, its quality is the MP3's quality in a larger file. WAV only sounds better when it never went through a lossy encoder in the first place.

If I convert WAV to MP3 and back, do I get the original WAV?

No. You get a WAV of the MP3. The round trip is not symmetric — the first step deletes information and the second step cannot invent it again.

Which format should I extract audio from a video as?

M4A if you just want the sound as it already exists, because that path copies the compressed audio without touching it. WAV if the file is going into an editor or an analysis tool that needs raw samples.