To make an MP3 louder without distortion, raise its gain by no more than its headroom: the gap between the loudest peak and about -1 dBFS. If you need more than that, put a limiter after the gain so the peaks are held down while the rest of the track comes up. A quiet recording can usually take +10 to +15 dB cleanly. A song that is already mastered loud has only 1 to 2 dB to give, and pushing harder adds clipping much faster than it adds loudness.
We measured exactly how this plays out on a public domain recording, using the same gain and limiter chain that runs the SearchSound.cloud volume booster. Below are the numbers, a step-by-step method for the browser, Audacity and FFmpeg, and the loudness targets worth aiming for.
Photo by Al Elmes on Unsplash
Why does making audio louder cause distortion?
Digital audio has a hard ceiling called 0 dBFS (decibels relative to full scale). Any sample pushed above it cannot be stored, so it is cut flat at the ceiling. That flattening is clipping, and it sounds like crackle, fuzz or harsh, smeared drums.
Gain is multiplication: +6 dB multiplies every sample by about 2 (20 × log10(2) ≈ 6.02 dB). Quiet passages and loud peaks are multiplied by the same amount, so the peaks hit the ceiling first. Distortion is therefore not caused by "too much volume" in general but by peaks with nowhere to go.
Three terms that come up in every guide:
- Headroom: how far the loudest peak sits below 0 dBFS. A file peaking at -9 dBFS has 9 dB of headroom.
- Peak level vs loudness: the peak is the single highest sample. Loudness is how loud the whole file sounds, measured in LUFS (loudness units relative to full scale) under the ITU-R BS.1770 standard. Two files with the same peak can differ in loudness by 10 LU or more.
- Limiter: a fast, high-ratio compressor that turns down only the moments that would cross a ceiling. The Audacity manual describes it as passing signals below the threshold unaffected "while preventing the peaks of stronger signals from exceeding this threshold".
How much can you boost an MP3 before it clips?
With plain gain, you can boost by exactly the headroom. A file that peaks at -15.6 dBFS can take +14.6 dB to reach -1 dBFS. To check your peak, open the file in Audacity (the Amplify dialog shows the amount of gain that would bring the peak to full scale, per the Amplify manual page) or run FFmpeg's volumedetect filter, which prints max_volume.
A limiter lets you go past the headroom, but only by trading dynamics for loudness. Our measurements show where that trade stops being worth it.
What did we measure?
We took 60 seconds of "by the across" by Thomas Park (Public Domain Mark), decoded to 44.1 kHz stereo (2,646,000 frames), and prepared it at two levels. We then rendered both in Chromium with the exact Web Audio chain the volume booster uses: a gain stage followed by a DynamicsCompressorNode set as a limiter (threshold -1 dB, ratio 20:1, hard knee, 3 ms attack, 250 ms release). For comparison we also applied the same gain with no limiter. Loudness and true peak were measured with FFmpeg's EBU R 128 meter after converting to 16-bit, which is what the downloaded WAV holds.
Case 1: a quiet recording
This is the track turned down by 14 dB, a stand-in for a quiet phone recording or an unmastered demo: -29.5 LUFS, peak -15.6 dBFS.
| Boost | Loudness after | True peak | Clipped frames with limiter | Clipped frames, plain gain |
|---|---|---|---|---|
| 0 dB | -29.0 LUFS | -15.0 dBTP | 0 | 0 |
| +6 dB | -23.0 LUFS | -9.0 dBTP | 0 | 0 |
| +12 dB | -17.0 LUFS | -3.0 dBTP | 0 | 0 |
| +14 dB | -15.0 LUFS | -1.0 dBTP | 0 | 0 |
| +18 dB | -11.2 LUFS | +0.2 dBTP | 35 | 2,167 |
| +24 dB | -7.6 LUFS | +0.3 dBTP | 5,912 | 136,227 |
Up to its headroom, the quiet file got louder decibel for decibel with nothing clipped. At +18 dB the limiter kept clipping to 35 frames, a 98% reduction compared with plain gain. At +24 dB it still clipped 5,912 frames (0.2% of the file) and the loudness gain started to fall behind the slider.
Case 2: a song that is already loud
The same track at its original level: -15.5 LUFS, peak -1.6 dBFS, typical of a finished master with little headroom left.
| Boost | Loudness after | True peak | Clipped frames with limiter | Clipped frames, plain gain |
|---|---|---|---|---|
| 0 dB | -15.0 LUFS | -1.0 dBTP | 0 | 0 |
| +3 dB | -12.0 LUFS | +0.1 dBTP | 5 | 144 |
| +6 dB | -9.7 LUFS | +0.2 dBTP | 581 | 25,032 |
| +12 dB | -7.0 LUFS | +0.3 dBTP | 11,565 | 287,207 |
Here +12 dB bought only 8.5 LU of extra loudness and clipped 0.44% of the frames, while the plain-gain version clipped almost 11%. The limiter helps a lot, but it is not a brick wall: with a 3 ms attack, the very first instant of a sharp drum hit gets through before it reacts.
Two details worth knowing from these runs:
- The 0 dB rows are about 0.5 dB louder than the input. That is not a bug in the measurement: the Web Audio specification gives every
DynamicsCompressorNodea fixed makeup gain computed from its threshold, ratio and knee. For these settings it works out to about +0.57 dB. - The quiet file at +14 dB landed at -15.0 LUFS with a -1.0 dBTP peak, which is close to where streaming services want music to sit (see below).
How do you make an MP3 louder online for free?
On SearchSound.cloud you can boost a file in the volume booster, which runs entirely in your browser, so the file is never uploaded.
- Open the volume booster and drop in your MP3, WAV, FLAC, M4A, AAC or OGG file.
- Set the Volume slider. It starts at +6 dB and goes from -12 to +24 dB.
- Press play and listen to the loudest part of the track, usually the chorus or the drops. If drums start to crackle or lose their snap, back off by 1 to 2 dB.
- Pick MP3 (320 kbps), WAV or FLAC and download.
As a rule of thumb from the tables above: for a quiet recording, boost by the headroom and stop. For a finished song, stay within +3 dB. If the beginning or end has silence or noise you don't need, cut it first with the audio trimmer so you can hear the part that matters while you adjust.
Normalize, amplify or limit: which should you use?
They solve different problems, and the right choice depends on whether you care about peaks or perceived loudness.
| Method | What it changes | Best for | Can it clip? |
|---|---|---|---|
| Amplify (gain) | Every sample by the same amount | Quiet files with plenty of headroom | Yes, once you pass the headroom |
| Peak normalize | Gain so the highest peak hits a target | Making the peak exactly -1 dBFS | No, but one spike limits the result |
| Loudness normalize | Gain so the file hits a LUFS target | Matching songs in a playlist or podcast | Yes, if the target needs more than the headroom |
| Limiter + gain | Holds down peaks, raises everything else | Getting louder than the headroom allows | Rarely, but dynamics get flatter |
Peak normalization is the trap most people fall into. If a recording has one loud cough or a single snare hit near 0 dBFS, normalizing does almost nothing because that one peak is already at the target. Audacity's Normalize sets peak amplitude; its separate Loudness Normalization effect targets LUFS instead and is based on EBU R 128.
How do you make an MP3 louder in Audacity or FFmpeg?
In Audacity, the cleanest sequence is:
- Select the track and open Effect > Volume and Compression > Amplify. The dialog proposes the gain that would bring the peak to 0 dB. Lower it by 1 dB and apply.
- If it is still too quiet, apply Effect > Volume and Compression > Limiter with a -1 dB limit and some makeup gain, then listen.
- Export as MP3.
With FFmpeg on the command line:
- Find the peak:
ffmpeg -i input.mp3 -af volumedetect -f null -and readmax_volume. On our loud test track it reported -1.5 dB, so it had no room to grow. - Raise by the headroom minus 1 dB. For a file peaking at -9 dB:
ffmpeg -i input.mp3 -af "volume=8dB" output.mp3. - Or let FFmpeg aim for a loudness target with a true-peak ceiling:
ffmpeg -i input.mp3 -af "loudnorm=I=-14:TP=-1" output.mp3. Without options,loudnormdefaults to -24 LUFS and -2 dBTP, which is a broadcast level and will usually make music quieter.
Both tools give finer control than a one-slider page, such as release time or two-pass loudness analysis. The trade-off is installing software and, with FFmpeg, typing commands.
How loud should an MP3 be?
For music, around -14 LUFS with peaks no higher than -1 dBTP is a sensible target. Spotify's loudness normalization page says it adjusts tracks to -14 dB LUFS according to ITU 1770, turns louder masters down, and only turns softer ones up as far as their headroom allows, keeping 1 dB of headroom for lossy encodings. Its example: a track at -20 LUFS with a -5 dBFS true peak is only lifted to -16 LUFS.
That has a practical consequence: making a finished song much louder than -14 LUFS gains nothing on Spotify, because playback turns it back down, and the clipping you added stays. Boosting is most useful for files that will be played as-is, like voice memos, lecture recordings, sound effects, DJ edits played next to louder tracks, or audio for a video.
For speech and broadcast, targets are lower: the Audacity manual notes -24 LUFS for US television and -23 LUFS from the EBU.
What if your audio still sounds quiet after boosting?
Then the problem is usually not gain. Common causes:
- One loud peak: a single spike limits the whole file. Trim it out, or let a limiter handle it.
- Thin low end on small speakers: phones and laptops reproduce little bass, so a track can measure loud and still sound weak. A gentle bass boost (under More effects on the same page) can help, but it uses headroom too.
- Already at the ceiling: if
volumedetectreports a peak near 0 dB and the file measures above -12 LUFS, it is mastered loud. Further boosting will mostly add distortion, as Case 2 shows. - Converting formats: turning the MP3 into WAV makes it bigger, not louder or better, as we measured in does converting MP3 to WAV improve quality.
FAQ
Can you increase MP3 volume without re-encoding?
Only with special MP3 editors that change the gain stored in each frame instead of decoding the audio, and those cannot add a limiter, so they are limited to the headroom. Any tool that applies a limiter, including the volume booster on SearchSound.cloud, decodes and re-encodes the file. Exporting at 320 kbps MP3 or as WAV/FLAC keeps that extra generation loss small.
Does normalizing audio reduce quality?
Peak and loudness normalization are just gain changes, so they don't add distortion as long as the result stays below 0 dBFS. Saving the result as MP3 again adds one more lossy encode, which is the only real quality cost. Use WAV or FLAC if you will edit the file further.
How many dB louder is twice as loud?
+6 dB doubles the signal amplitude, which is what the slider adds. Perceived loudness grows more slowly, so a boost that doubles the amplitude does not sound twice as loud. For judging loudness, compare LUFS values instead of the slider number.
Is +24 dB on a volume booster safe?
Only for very quiet sources. In our test, +24 dB on a file that peaked at -15.6 dBFS clipped 5,912 frames even with the limiter. Find the headroom first and boost by about that much; go further only while it still sounds clean.
Why does my boosted audio sound flat or pumping?
That is the limiter working hard. When the boost goes well past the headroom, it turns down every loud moment, so drums lose punch and the level can seem to breathe. Lower the boost by a few dB, or trim a single loud spike that is forcing the limiter to act.
If you are building a free toolkit for edits like these, see our guide to a sustainable music production workflow with free tools, and for another browser effect with measured settings, how to make nightcore.