MP3Gain Command Line Guide: Every Flag That Matters

The Windows program is a friendly shell over a command-line engine — backend version 1.5.2 — and on Linux and macOS that engine is all you get (see how to install it). The good news: the CLI does everything the GUI does, and it scripts beautifully. Every flag below is documented in the official mp3gain(1) man page; this guide translates that page into the options you will actually use, in the order you will reach for them.

The apply flags: make it louder, make it quieter

One interaction to memorize: if you pass both -r and -a, only the second one takes effect — the man page says so explicitly, and it is the most common cause of “I ran album mode but got track mode” confusion.

Targeting and clipping: aim, and don’t overshoot

Tags and undo: where the memory lives

mp3gain stores its analysis, the applied gain and the undo information in tags — APEv2 by default. The -s family controls that storage:

Workflow flags for scripts and big libraries

• -q — quiet mode, no status messages; pair with -o for machine-readable, tab-delimited output you can pipe into a spreadsheet or log.
• -t — write the modified MP3 to a temporary file, then replace the original (the safer behavior, and the default in Debian builds). Its opposite, -T, modifies the original in place.
• -p — preserve the original file timestamps, so your library’s sort order survives processing.
• -x — report only the maximum amplitude of each file, a quick scan for tracks that are already at full scale.
• -f — force mp3gain to treat input as MPEG Layer III without checking (for mis-named files); -w handles the rare edge case where a gain change would wrap past the frame field’s 0–255 range; -v prints the version; -? prints the built-in help.

Three real jobs, three commands

1. One album, leveled as an album. Change into the album’s folder and apply album gain with clip protection to every file in it:

mp3gain -a -k *.mp3

The -a makes mp3gain compute one shared change for the whole set; -k guarantees the loudest track’s peaks cap the change before anything clips.

2. A whole mixed library, track by track, unattended. Recurse through a music tree and level every file independently:

find ~/Music -iname "*.mp3" -exec mp3gain -r -k {} +

-r is right here because the files are unrelated singles; -k means no file can clip and nothing stops to ask questions halfway through a 5,000-file run. Run it on a backup copy first if the library is irreplaceable — the full library workflow explains the safety routine.

3. Undo everything in a folder. Changed your mind about a processed album? Restore the original volumes from the stored undo data:

mp3gain -u *.mp3

This only works while the tags carrying the undo information are intact — which is why -s d (delete tag info) should always be the last thing you ever do to a file, if you do it at all.

Frequently asked questions

Does running mp3gain with no options change my files?

No. Without an apply flag (-r, -a, -g or -l), mp3gain only analyzes each file and writes the recommended gain into a tag — APEv2 by default. The audio data is modified only when you explicitly apply a gain.

Should I use -r or -a?

Use -r (track gain) for unrelated files — singles, playlists, a folder of mixed downloads — so each song is leveled on its own. Use -a (album gain) for a real album, so one shared adjustment preserves the intended loudness differences between its tracks. If you pass both, only the second one takes effect.

Why does mp3gain stop and ask about clipping?

Because without -c, it asks for confirmation before applying a gain that might push a file into clipping. You have three choices: answer the prompt, add -c to apply anyway (the file may clip), or — usually best — add -k, which automatically lowers the gain to the highest clip-free value.

Can I undo a gain change made from the command line?

Yes. Run mp3gain -u on the files and mp3gain reverses its earlier changes using the undo information stored in the files’ tags. The undo stops being possible if those tags were deleted (-s d) or stripped by another tool, so keep the tags until you are sure you are done.

MP3Gain on Linux: Install the CLI and GUI Front-Ends

MP3Gain was born as a Windows GUI wrapped around a portable command-line engine. On Linux there is no official graphical version — the command-line backend (version 1.5.2) is the real product, and everything else is a front-end driving it. This guide explains why your package manager probably can’t install it anymore, and the three routes that still work: building the official CLI from source, using a GUI front-end, or switching to its modern rewrite.

Why apt install mp3gain fails

If you learned MP3Gain a decade ago, you may remember it being one apt command away. That ended in 2015: Debian removed mp3gain from its repositories because the package no longer had an active maintainer (the removal traces to Debian bug #761847), and Ubuntu dropped it at the same time. Nothing has reversed that decision — current Debian and Ubuntu releases do not carry the package, and Ubuntu’s security tracker lists mp3gain as “not in release” for every release since. Old .deb files and man-page mirrors still circulate, which is why search results make it look installable, but on a modern Debian or Ubuntu system the package simply isn’t there.

A few distributions never dropped it. openSUSE Tumbleweed, for example, still packages the classic command-line tool — its current man-page index carries mp3gain(1) — so openSUSE users can install it directly with sudo zypper install mp3gain. Everyone else should use one of the routes below — none of them is a compromise: the command-line tool was always the fully-featured half of MP3Gain, and the Windows interface is best understood as a convenience layer over it.

Route 1: build the official CLI from source

The project’s SourceForge files page still hosts the backend source as mp3gain-1_5_2-src.zip (our source link points at the official project). It is plain C with no exotic dependencies, and it builds the same way it always has:

1. Install a compiler toolchain — on Debian/Ubuntu that is sudo apt install build-essential unzip; use your distribution’s equivalent otherwise.
2. Unzip the source archive and open a terminal in the folder it creates.
3. Compile and install: run make, then sudo make install. The result is a single mp3gain binary.
4. Verify with mp3gain -v, which prints the version. You now have the identical 1.5.2 engine the Windows GUI drives — our command line guide covers every flag that matters.

Building from source is the route we recommend for the classic tool: it does not depend on any distribution’s packaging decisions, and the resulting binary has been stable for a decade.

Which version are you actually getting?

Linux documentation makes the versioning look stranger than it is. The man-page mirrors you will find online were written for the Debian package and reference files under /usr/share/doc/mp3gain-1.4.6/ — 1.4.6 was simply the backend version Debian froze on. The final backend release is 1.5.2, which is what the source archive in Route 1 builds, what openSUSE packages, and what sits inside the Windows GUI; we untangle the whole family tree (including the 1.3.x beta line and the download-site noise around it) in which MP3Gain version is real. For practical purposes the differences are negligible: the ReplayGain analysis and the 1.5 dB frame-step mechanism are identical across these versions, and files processed by one are understood — and undoable — by the others.

Route 2: a GUI front-end over the CLI

If you want the Windows program’s point-and-click workflow — file list, analysis columns, big Apply button — two front-ends are worth knowing:

• QMP3Gain — a Qt-based recreation of the original MP3Gain GUI, developed on Linux. Its SourceForge page offers a Linux AppImage (version 0.9.2) that needs no installation: make it executable and run. One gotcha the project states plainly: the AppImage does not include the mp3gain backend, which you must install separately — Route 1 above is the dependable way to get it.
• wxMP3gain — a wxWidgets front-end distributed as source from its own SourceForge project (version 3.2). It bundles the analysis-and-apply workflow into a native window, but you compile it yourself, so it suits users already comfortable building Route 1.

There is also easyMP3Gain, an older GTK front-end for MP3Gain, VorbisGain and AACGain that the project’s Wikipedia page still links to. It works where its dependencies survive, but it is the least maintained of the three; try QMP3Gain first.

Using it once it’s installed

The binary you get from any of these routes behaves identically, because the front-ends and packages all drive the same engine. A typical first run on an album folder looks like this:

mp3gain -a -k *.mp3

That applies one shared album gain (-a) with automatic clip protection (-k). Run it with no apply flags and it only analyzes and tags; add -u later and it reverses itself from the stored undo data. The complete flag reference — track vs album mode, the -s tag family, scripting options — is our command line guide. One Linux-flavored tip: because the CLI changes files in place (or via temp files), it drops naturally into shell scripts and cron jobs in a way the Windows GUI never could; leveling new downloads automatically is a five-line script, not a chore.

Format note: the engine only speaks MP3. The long-standing companion tool for AAC files in MP4 containers is AACGain, a fork of the same code that most MP3Gain front-ends can be pointed at instead — so a QMP3Gain setup from Route 2 can often be taught to handle your M4A files too, by swapping the backend binary.

Route 3: mp3rgain, the modern rewrite

If the “build a 2010 codebase” part bothers you more than the workflow, look at mp3rgain: an open-source (MIT) rewrite of mp3gain in Rust, available for Windows, macOS and Linux. It uses the same trick as the original — adjusting the global gain field in each MP3 frame, with the original values stored in APEv2 tags so changes stay reversible — but with maintained code and modern analysis. It installs through Cargo or a Homebrew tap. It is a young project, so test it on copies of your files first; but for a Linux user in 2026 it is the most future-proof way to get exactly MP3Gain’s lossless behavior. (If you only need ReplayGain tags rather than modified files, tag-only scanners such as Loudgain fill that narrower job.)

MP3Gain for Android: No Official App — What Actually Works

Search Google Play for “MP3Gain” and you will find volume boosters, equalizers and apps borrowing the name — but not MP3Gain. There is no official MP3Gain app for Android, and there never has been: MP3Gain is a Windows desktop program by Glen Sawyer, distributed through SourceForge. This page covers the one historical port you may have heard of, why it is no longer a real option, and the two routes that genuinely work today.

The DroidGain story: a real port that time left behind

In the early 2010s an unofficial Android port called DroidGain did exist. By all surviving accounts it copied the desktop program’s workflow: point it at MP3 files, let it analyze loudness, and apply the same lossless frame-gain adjustment in bulk. For a while it was the answer to “MP3Gain for Android.”

It is not the answer anymore. DroidGain has been unmaintained for years and, as of October 2026, it has no Google Play listing — searching the store turns up only unrelated booster apps, and the remaining mentions of DroidGain live on auto-generated “alternatives” pages describing it in a present tense that is long out of date. Copies of the APK still float around third-party download mirrors, and we do not recommend installing them: the app predates modern Android storage rules, so it is unlikely to handle a current phone’s library correctly, and sideloading an ancient, unverifiable binary whose one job is to rewrite your music files is a poor trade. If DroidGain is still installed on an old device of yours, it will probably still run — just don’t hunt down a copy for a new phone.

Route 1 (recommended): level the files on a computer, then sync

The most reliable way to get MP3Gain results on Android is to do the processing where MP3Gain actually lives — on a Windows PC — and then copy the finished files to your phone. This works because of how MP3Gain applies gain: it changes the gain value inside the MP3 frames themselves, in 1.5 dB steps, rather than asking the player to compensate. Once a file is processed, every Android player plays it at the corrected volume — the stock player, a streaming app’s local-file mode, your car stereo over Bluetooth. No tag support, no special settings, nothing to configure on the phone.

1. Process the library on your PC. Run MP3Gain (or follow the full workflow in our guide to making all MP3s the same volume). Track Gain suits mixed folders and playlists; Album Gain keeps an album’s internal dynamics intact.
2. Copy the processed files to the phone by USB, cloud drive or your usual sync tool. Processed files replace the originals on the phone; keep the PC copies as your master library.
3. Play them anywhere. Because the gain lives in the audio data, the leveling survives every player and every future phone upgrade.

The changes also stay reversible: MP3Gain stores an undo record in each file’s tags, so if you change your mind, the desktop program can restore the original volume exactly — see how undo works. And if your real complaint is that songs jump in volume specifically on your phone, our phone volume guide separates the file problem from the player problem before you process anything.

Route 2: keep files untouched and use a ReplayGain-aware player

The alternative is to leave the MP3s alone and let the player even things out at playback time using ReplayGain values stored in tags. Tools like foobar2000 — or MP3Gain itself in its analyze-only mode — can write those values without touching the audio. The catch is that this only works in players that read the tags, and on Android the honest shortlist is:

• AIMP for Android — actively maintained, and its Play Store listing explicitly advertises “volume normalization using replay gain or peak-based normalization.” Point it at a tagged library, enable the normalization, and track-to-track jumps mostly disappear.
• Poweramp — its Play listing includes replay gain among its features, and it reads ReplayGain values written to ID3v2 tags.
• foobar2000 mobile — from the same team as the desktop scanner, so tags it writes itself are tags it reads.

Two caveats before you commit to this route. First, tag formats matter: MP3Gain stores its data in APEv2 tags by default, which not every Android player reads; ID3v2 ReplayGain fields have the widest support. Second, tag-based leveling exists only inside that one player — send the same files to a car stereo or a different app and you are back to uneven volume. That is why Route 1 is the recommendation whenever a computer is available: it fixes the files, not one app’s playback.

One trap: don’t level the same files twice

Whichever route you choose, pick one system and stick to it. If you level the audio data with MP3Gain on your PC (Route 1) and then also enable replay-gain normalization in a tag-aware Android player (Route 2), the player may apply a second correction on top of the first — especially if the files still carry ReplayGain values suggesting a change that has, in effect, already been made. The usual symptom is a library that sounds too uniform in one app and oddly flat or quiet in another. The file-based route needs no player help at all: if you go with Route 1, leave player normalization off. If you go with Route 2, leave the files untouched and let the tags do the work. Our phone volume guide walks through diagnosing which of the two is already active on your device.

What about the “MP3 volume booster” apps on Play?

Most of them fall into two groups, and neither is MP3Gain. Playback boosters add gain or compression inside the player, which raises quiet songs but also loud ones, and does nothing outside the app. File-processing boosters typically decode and re-encode your MP3s through an FFmpeg-style pipeline at a higher level — that works, but re-encoding an MP3 costs audible quality, is not reversible, and throws away the one thing MP3Gain exists to protect. If a file is simply too quiet and you don’t care about re-encoding, that is a legitimate choice; just make it knowingly, and keep the original.

MP3Gain for Mac: No Official Version — Here Are Your 3 Real Options

Let’s start with the fact most “MP3Gain for Mac download” pages won’t tell you: there is no official MP3Gain for macOS, and there never has been. MP3Gain by Glen Sawyer is a Windows program. Any page offering you an official-looking “MP3Gain-Mac.dmg” is offering something else under a borrowed name. The good news: Mac users have three genuine routes to the same job, and they are all below.

Option 1: MP3Gain Express — the real Mac port

MP3Gain Express, by developer Paul Kratt (Sappharad), is the established Mac port of MP3Gain. It does the same job the same way: ReplayGain analysis, lossless gain adjustment written into the MP3 frames, batch processing, and undo — in a native macOS application. It also bundles AACGain support, so it can normalize AAC/M4A files, something the Windows original cannot do at all. The most recent release, version 2.4, dates from 2020; it is mature rather than actively evolving, which for a solved problem like this is mostly fine, though very new macOS versions are always a “test first” proposition with software of this age. You will find it on the author’s site, sappharad.com. This is the closest thing to “MP3Gain for Mac” that honestly exists, and for most Mac users it is the answer.

Option 2: wxMP3gain — the cross-platform open-source GUI

wxMP3gain is a free, open-source graphical front-end built on the same MP3Gain engine, written with the wxWidgets toolkit so it can run across Windows, Linux, and macOS. In spirit it is the most “official-adjacent” alternative: same analysis, same lossless method, same 89.0 dB target world. In practice it asks more of you than MP3Gain Express — distributions for macOS have historically been less polished and may require more manual setup — so it suits users who prefer open-source tooling and don’t mind tinkering. (One similar name to skip: QMP3Gain, a Qt front-end, explicitly supports Linux and Windows only — its own project page states there is no Mac support.)

Option 3: A modern alternative, if you want more than MP3Gain

If you are on a Mac anyway, it is worth asking whether a 2001-design tool is what you want. Modern normalizers such as waveAlign run natively on current macOS, measure loudness with the LUFS (BS.1770) standard used by streaming and broadcast instead of ReplayGain, and handle AAC, WAV, and FLAC alongside MP3 — for a fee (€79 one-time at the time of writing), where MP3Gain and its ports are free. The other familiar name, Audacity, can normalize too, but by decoding and re-encoding the audio, which is a different, lossy trade-off — fine for editing, not a substitute for lossless gain adjustment. Choose by need: same free lossless MP3 workflow → Option 1; open-source purist → Option 2; multi-format, current-standard, supported software → Option 3.

What not to download

Not a “MP3Gain for Mac” installer from a download portal, and not the file this page used to point to. Until this guide existed, our own download page’s Mac row linked to a cloud mirror labelled as a Mac build — there is no such official build, so we have replaced that link with this page. That is the general rule: for Mac, the chain of trust runs through the port author’s own site, never through a generic .dmg on a file host. Our safety guide shows how to apply the same skepticism on Windows, and the MP3Gain overview explains the Windows original these ports are based on.

Still want the original?

If you also have a Windows PC — or a Windows virtual machine on your Mac — the original 1.2.5 remains the reference implementation, free from the official SourceForge project via our download page, with a portable ZIP build that needs no installation. Mac on the desk, Windows in a VM, MP3s in between: plenty of libraries are maintained exactly that way.

What about running the Windows original on your Mac?

Technically possible, rarely the best answer. A Windows virtual machine (Parallels, VMware Fusion, or UTM on Apple-silicon Macs) will run the genuine 1.2.5 — the portable ZIP is the convenient build for this, since nothing needs installing inside the VM. Compatibility layers in the Wine family can also carry older Windows utilities on macOS, with the usual tinkering tax. Weigh that honestly against Option 1: MP3Gain Express gives you the same analysis and the same lossless result in a native app, in less time than a VM takes to boot. The VM route makes sense when the Mac is a side machine and your whole library workflow already lives in Windows; as a way to avoid a Mac port that exists and works, it is effort spent in the wrong place.

Your files don’t care which side leveled them

One reassurance for mixed households: the gain change itself is platform-blind. An MP3 leveled by MP3Gain Express on a Mac carries the same frame-level adjustment, and its undo tag works the same way if the file later meets the Windows original. There is no Mac dialect and no Windows dialect of a leveled MP3 — which means you can normalize on whichever machine is in front of you, with whichever tool above you chose, and never create a compatibility problem for future you.

Which option fits which Mac user?

• “I just want my MP3s leveled, like the old days”: MP3Gain Express. It is the port, it behaves like the original, and it is free.
• “My library is half M4A from the iTunes years”: also MP3Gain Express — its bundled AACGain support covers the AAC files the Windows original never could.
• “I run a mixed Mac/Linux/Windows household”: wxMP3gain, accepting the extra setup as the price of one open-source tool everywhere.
• “I prepare tracks for streaming or DJ software and care about LUFS”: the modern alternative — you are buying current loudness standards and multi-format support, not nostalgia.

Whichever you land on, the destination is the same library state the Windows crowd reaches with 1.2.5: every track at one honest volume, originals recoverable, nothing re-encoded. The tool is a detail; the leveled library is the point.