MP3 vs WAV: When Compressed Audio Is Good Enough

A four-minute song is about 40 MB as a WAV, 9.6 MB as a 320 kbps MP3, and 5.8 MB at 192 kbps. In a blind test on normal headphones, almost nobody reliably tells the 320 kbps version from the WAV. So is WAV pointless? No — but the reason to use it has almost nothing to do with how it sounds, and everything to do with what happens to the file next.

What the Two Formats Actually Are

WAV is essentially no format at all: raw, uncompressed PCM samples in a thin wrapper. CD-quality stereo (44.1 kHz, 16-bit) runs about 10 MB per minute, every minute, regardless of content. Nothing is discarded, ever — a WAV is a bit-perfect copy of the signal.

MP3 is perceptual compression: it models what human hearing can’t resolve — quiet details next to loud ones, certain frequency masking effects — and discards exactly that. The discarding is controlled by bitrate, measured in kilobits per second. More bits, less discarded.

The size math is simple and worth knowing:

What Bitrates Actually Sound Like

• 128 kbps — audibly compressed on music: cymbals get swishy, stereo image flattens. Acceptable for voice, dated for anything else.
• 192 kbps — the practical sweet spot for speech. Podcasts, lectures, recorded meetings: transparent, and half the size of 320.
• 256–320 kbps — transparent for music to almost all listeners on almost all equipment. This is where the WAV-vs-MP3 listening difference effectively disappears.

One rule saves a lot of wasted megabytes: an MP3 can’t sound better than its source. Encoding a mediocre recording at 320 kbps just makes a bigger mediocre file. Bitrate preserves quality; it never adds it.

When WAV Actually Matters

WAV earns its 10× size in exactly the situations where audio gets processed rather than played:

• Editing and mixing. Every MP3 encode discards information, and the losses stack: edit an MP3, export an MP3, and you’ve compressed twice. Audibly worse by the second or third generation — the same generation-loss problem delivery video codecs have. Editing in WAV means loss happens once, at final export.
• Effects and cleanup. Noise reduction, EQ, and time-stretching amplify compression artifacts that were designed to hide below normal hearing thresholds.
• Sampling and music production. DAWs and samplers want PCM; pitching an MP3 artifact up an octave makes it very audible.
• Archiving masters. You can always make an MP3 from a WAV. You can never get the WAV back from an MP3.

Notice what’s not on the list: listening. For playback, sharing, podcasts, and phone storage, a well-encoded MP3 is genuinely good enough — and its universal compatibility (every player, car stereo, and app since the late ’90s) is the reason it’s still the default despite newer codecs like AAC and Opus existing.

The WAV Settings That Matter: Sample Rate and Bit Depth

“Uncompressed” still involves two choices, and they explain the numbers on the tin:

• Sample rate — 44.1 kHz is the CD/music standard; 48 kHz is the video standard, which is why audio extracted from video is usually 48 kHz. Both capture the full audible range; don’t resample without a reason.
• Bit depth — 16-bit is transparent for finished audio. 24-bit exists for recording and mixing, where the extra headroom keeps quiet signals clean through processing. A 48 kHz / 24-bit stereo WAV runs about 17 MB per minute versus 10 for CD-quality.

And the neighboring formats, briefly: FLAC is lossless like WAV but compressed to roughly half the size — excellent for archiving listening libraries, less universal than WAV in DAWs and samplers. AAC (the audio inside MP4 files) beats MP3 slightly at the same bitrate; MP3 keeps winning on the only axis left — absolutely everything plays it.

The One Rule: Never Re-Encode Lossy to Lossy

Converting a 128 kbps MP3 to 320 kbps doesn’t restore anything — it re-compresses already-damaged audio into a bigger file. The workflow that avoids every trap:

1. Capture or extract at the highest quality available.
2. Process in WAV.
3. Export to MP3 exactly once, as the last step.
4. Keep the WAV if you might ever return to it.

Extracting Audio from Video

The most common place this decision actually comes up isn’t music — it’s video. A recorded lecture, a concert clip, a meeting, a podcast published as video: you want the audio track, and you have to pick a format on the way out.

The rule from above applies directly. If you just want to listen — pull the audio as MP3: MP4 to MP3 in the browser lets you pick the bitrate (192 kbps for speech, 256–320 for music) and trim to just the segment you need. Browser-based meeting and screen recorders usually produce WebM instead, where the audio is Opus — a fine codec that car stereos and many players won’t take as a standalone file — so the equivalent WebM to MP3 extraction is the move for recorded calls and webinars. Both run entirely on your device, which matters more for audio than almost anything else: meetings and interviews are exactly the files you don’t want on someone’s server.

If the audio is headed into processing — editing, transcription, sampling — extract MP4 to WAV instead. It hands your tools an uncompressed signal with no extra lossy generation stacked on top, at the usual price: about 10 MB per minute, so trim to the section you need first.

The Bottom Line

WAV is a working format; MP3 is a delivery format. Confusing the two costs you either 10× the storage or a generation of quality — and once you frame it that way, the choice makes itself. The same working-vs-delivery logic runs through every media format on this site, including the video container question.