How to convert MP4, MOV, and WebM audio to MP3

People often call this extracting the audio. OfflineMP3 does not lift the original audio frames out of the video. It decodes the track you select and encodes a new MP3. This page is the path from a file on your computer to that MP3, including the points where the current Beta stops.

Open the converter

What you need before you start

Use a video you own, or one the rights holder has allowed you to copy, convert, and download. The converter does not open a streaming link, a page on another site, or a file that sits behind a login. You choose a file that is already on your computer. The content site at https://www.offlinemp3.com/ explains the tool. The converter itself is https://app.offlinemp3.com/.

The file has to be a real MP4, an audio-only M4A or M4B, a QuickTime MOV, or a WebM file when the page looks inside it. The letters at the end of the name are only a hint. An MP4, M4A, M4B, or MOV file needs AAC-LC on every audio track. A WebM file needs Opus on every audio track. Mono, stereo, and a layout labelled 5.1 are the channel layouts that pass. The formats guide lists the refusals in one place.

Open the converter in a desktop copy of Chrome or Edge. The processing page reads the browser's user-agent string. It starts a task only when that string looks like desktop Chrome, Chromium, or Edge, and when the browser exposes a Worker and WebAssembly. Phones and tablets are blocked, including mobile Chrome. You can still read this page in other browsers.

The steps on the processing page

Choose the file. The page prepares the converter, which means it downloads the conversion program from the processing site if it is not already loaded. That download is the program, not your file. If the program cannot be loaded, the page says so and asks you to check the connection, then try once more. It tries a failed load of the program three times, with a short pause between tries.

Inspection comes next. The page reads the whole file in the browser and asks the on-device tools what container, tracks, codecs, channel labels, and duration it contains. It also tries to decode each audio track all the way through. A truncated track fails here, before any MP3 is written. If the file has several audio tracks, you see the language, title, channel layout, and default flag when the file stored them. A declared default is already selected. A file with one audio track selects that track even when no default flag is set. If several tracks exist and none is the default, the convert button stays disabled until you pick a track. Selecting a track does not start the encode.

Press Convert selected track. Encoding shows elapsed processed time and an approximate percentage. That percentage stays at or below 99% until verification has succeeded, so a full-looking bar during encoding is not a finished file. Preparation, inspection, and verification do not invent a percentage at all.

What the MP3 contains

The encoder writes MPEG-1 Layer III at 48 kHz and 192 kbps constant bitrate. There is no menu for another rate. Mono stays one channel. Stereo stays two. A labelled 5.1 mix is folded to stereo with the fixed downmix-v1 matrix, and the low-frequency effects channel is not included in that mix. Tags, chapters, and cover art are dropped on purpose. The suggested name is the video's base name, cleaned of characters that are awkward in a file name, cut at 120 characters, with .mp3 on the end. An empty cleaned name becomes converted-video.mp3.

Because the source codecs that pass inspection are themselves AAC-LC or Opus, the MP3 is another lossy generation. It is not a copy of the soundtrack as it was stored in the video. The bitrate guide and the lossy-audio guide explain what that does and does not preserve.

Verification, then save

After encoding, the page decodes the source track and the new MP3 and compares them. It checks that every MPEG frame is 48 kHz, 192 kbps, MPEG-1 Layer III, that there is a single MP3 stream with the expected channel count, and that chapters and inherited tags are absent. The decoded MP3 has to contain the same number of samples per channel as the decoded source. If any of that fails, there is nothing to save, and the message tells you not to keep the result.

Those checks do not look for a special tone, a test pattern, or any particular number of loud segments. Ordinary speech, music, and other audio are saved when the file is supported and the MP3 is non-empty, decodes completely, and has a decoded duration that matches the selected track. Sample counts come from the decoded audio, not from the container duration. A decode error, a missing track, an unsupported codec, a size or memory limit, or a sample-count mismatch still stops the save. Retrying the same broken file meets the same rule.

When verification does pass, Save MP3 uses the browser save dialog where the browser provides it. Save completed means the write finished and the file was closed. Download MP3 only means the browser accepted a download. Dismissing the save dialog keeps the verified result so you can save again without a second encode. Discard result releases it.

Stopping, retrying, and leaving

Cancel is immediate during preparation and inspection. During conversion and verification the page asks you to confirm, and the focused action is to continue. Confirming cancel destroys the worker and keeps no partial MP3. There is no resume if you refresh, navigate away, or the tab crashes. The page may ask the browser to warn you before leaving. The browser does not have to show that warning.

After a failure or a cancellation you can press Retry from the beginning once. That starts a new task from the same file. A second failure of that input is the end of the product's retries. A file that was refused during inspection does not get that button, because the same file will be refused again. The troubleshooting guide matches each reason code to a next step.