Why is my MP3 so much quieter than everything else?

It's almost always one of four things: the track was mastered at a lower level than the rest of your library, your player is normalizing it down (ReplayGain, Sound Check, Spotify's loudness matching), the file came from a low-level rip or capture, or the recording was genuinely quiet at the source. Measuring the file tells you which one in about two seconds. A track that peaks at -0.3 dBFS but still sounds small has a loudness problem you can only partly fix. A track that peaks at -18 dBFS is just sitting low, and you can raise it 17 dB with nothing lost at all.

Measure it before you guess

Two numbers separate all four causes. Drop the file into the free in-browser volume booster and hit Smart Boost. It scans every sample and reports:

• Sample peak (dBFS) — the single loudest sample in the file. 0 dBFS is the ceiling; nothing goes above it without clipping.
• Integrated loudness (LUFS) — a real ITU-R BS.1770-4 measurement: K-weighted, 400 ms blocks with 75 % overlap, gated at -70 LUFS absolute and -10 LU relative. This is the number that tracks how loud something actually sounds across the whole song.
• Headroom (dB), plus the exact boost percentage that produces zero clipped samples. It aims for 1 dB below full scale and leaves that margin alone.

Peak tells you whether there's free space. LUFS tells you whether the track is quiet because it's low-level or quiet because it's dynamic. Read them as a pair.

One ceiling to know before you start: the slider runs 0 to 500 %, and 500 % is +13.98 dB. That covers almost everything. If a file needs more than about 14 dB, export the WAV and run it through a second pass.

Cause one: the track was mastered quietly

This is the most common answer, and it doesn't mean your file is broken. Mastering engineers pick a level. Before roughly the mid-1990s, and still today in classical, jazz, folk and most soundtrack work, that level leaves the average well below where a modern pop master sits.

The gap is bigger than people expect. A 2020s pop single often lands around -7 or -8 LUFS integrated. A 1985 CD master might sit at -18 LUFS. A classical recording can be -23 LUFS or lower. That's about 15 dB of perceived loudness between two files that peak in the same place. Fifteen dB is not subtle. It's the difference between "loud" and "did the song stop?"

You'll spot it instantly: peak near 0 dBFS, LUFS low. The file has no headroom to give you. Matching a loud master means squashing the dynamics, which is a trade, not a repair.

Cause two: your player is turning it down

If the file measures fine on its own but plays quietly in one app, something is normalizing it. Loudness normalization is on by default in a lot of places:

• Spotify targets roughly -14 LUFS. A -8 LUFS master gets pulled down about 6 dB.
• Apple Music's Sound Check does the same job with its own measurement.
• YouTube attenuates loud uploads toward a similar target and never boosts quiet ones.
• ReplayGain and EBU R128 tags written into ID3 or APEv2 metadata by ripping tools. Foobar2000, VLC and MusicBee honour them; plenty of other players ignore them entirely, which is why one file plays at two different volumes on two devices.

The tell for "one song much quieter than the rest" is that it only happens in one app. Play the file somewhere else, or measure it in the browser tool where no normalization exists, and it reads normally. The fix is to turn normalization off in that player's audio settings, or strip the ReplayGain tags with a tag editor. Boosting the file itself is the wrong move here, because the normalizer will just pull your louder version straight back down.

Cause three: a low-level rip or capture

Bitrate is not the culprit people think it is. A 128 kbps MP3 is not quieter than a 320 kbps one. It's less detailed, with smeared cymbals and a swirly top end, but the level is identical. What does produce a genuinely quiet MP3:

• Analog captures. Anything through a line-in, a cassette deck or a turntable preamp with the input gain set low. These usually peak somewhere around -15 to -25 dBFS.
• Screen and app recordings. Phone and desktop capture tools often record well below full scale on purpose so nothing clips.
• Batch converters with a safety pad. Some apply a couple of dB of attenuation on encode so the decoder can't overshoot.
• A file that got normalized down and re-saved. Once it's been re-encoded at the lower level, that's baked in.

These are the easy wins. The peak sits far below 0, so the gain costs nothing — you're multiplying samples, not squashing anything. Smart Boost gives you the exact percentage that lands 1 dB below full scale with zero clipped samples, and one click applies it.

The fix for each case

1. Low peak, low LUFS. Load it, run Smart Boost, take the suggested percentage, export. If it says 380 %, that's about +11.6 dB and nothing clips. Under a minute.
2. Peak near 0, low LUFS. You need loudness, not gain. Push past what the peak allows and let the soft limiter catch the overs. It's a tanh-knee waveshaper: transparent below 0.7 of full scale, asymptotic to full scale above it, so it rounds peaks instead of chopping them square. It's a limiter, not a compressor — no attack or release to set. It's on by default. Go gently. 150 % on a dynamic master is usually fine; by 300 % the transients are flattened and the track sounds hard.
3. Fine in one player, quiet in another. Don't touch the file. Turn off normalization in that player.
4. Quiet and dull or thin. Level isn't the whole story — a dull recording reads as quieter than it measures. Try the five-band equalizer. The Voice preset is -4 dB at 60 Hz, -2 at 230 Hz, +3 at 910 Hz, +5 at 3 kHz and +2 at 14 kHz, so it trades low end for presence rather than just adding level everywhere. If the problem is thin rather than dull, the bass booster is a 120 Hz low shelf up to +12 dB. Both add energy and can push peaks into the ceiling, so re-check the clip warning after. Use the level-matched A/B button while you judge: it trims the processed side back to the original's peak, so "louder" can't trick you into thinking "better".

One note on precision: the tool reports sample peak, not true peak (dBTP). Inter-sample peaks after D/A conversion can run roughly 0.5 to 1 dB above the highest sample, and MP3 decoding isn't bit-exact across decoders, so two players can produce slightly different peaks from the same file. The 1 dB of margin Smart Boost leaves is there for exactly that.

Making a whole library match

If you're chasing consistency across hundreds of files, don't set every track to the same peak. Peak-normalizing a library is precisely what leaves classical tracks sounding quiet next to pop: same peak, 12 dB apart in perceived loudness. Match on loudness instead. Pick a target, read each file's integrated LUFS, apply the difference.

Sensible targets: -14 LUFS if the material is headed for Spotify or YouTube anyway, -16 LUFS for spoken content going to Apple Podcasts, -18 or -20 LUFS if you want to keep the dynamics and only tame the obvious outliers. The tool measures and tells you the gain needed to hit those numbers, but it won't re-master to a target behind your back. You read the number and move the slider, or click the suggestion. That's deliberate — automatic loudness targeting is how quiet, well-made albums get flattened by accident.

Files stay on your machine throughout. No upload, no server, no account; turn Wi-Fi off after the page loads and it still runs. Export is WAV at 16, 24 or 32-bit float, rendered offline faster than real time. There's no MP3 export, so if you need MP3 back you'll re-encode elsewhere and lose a generation of quality — worth knowing before you start. For fixing quiet tracks on your phone, the Increase Volume app for iPhone keeps a library of your files with per-track settings saved, boosts to 200 % (+6.02 dB), and works offline.

What boosting can't fix

Gain multiplies everything in the file, including the parts you don't want. On a quiet capture with audible hiss, +18 dB of boost raises the hiss +18 dB too. The signal-to-noise ratio doesn't move. If a recording was made at -35 dBFS in a room with an air conditioner running, you get a louder air conditioner. No volume tool removes noise, separates stems, or repairs a file whose peaks were already clipped flat at the source.

And nothing here raises your device's system volume. If your phone is at maximum and everything still sounds quiet, that's a hardware ceiling — no browser page and no iOS app can move it. What you can do is raise the level inside the file so it uses more of that fixed ceiling, which is exactly what a track peaking at -18 dBFS is failing to do. Video works the same way: the video volume booster plays MP4, MOV and WebM with the audio boosted live and exports the boosted audio as WAV. It does not re-encode the video file, so you'll be pairing that WAV with the original video in an editor.