MP4 to MP3 Converter

Save just the audio from an MP4 as MP3: no upload, and big videos open too because only the audio is read

This MP4 to MP3 converter pulls the audio track out of a video file and saves it as an MP3. When you pick a file, it first reads the MP4's index and shows, as a bar, how many bytes belong to video, audio and everything else, then draws a waveform of the audio track. Drag the handles on the waveform to select the part you need, choose a quality and press 'Convert to MP3'.

The whole conversion happens inside this browser, and the video is never sent to a server. Video bytes are skipped and only audio bytes are read, so even a phone video of several gigabytes opens with no file size limit. Convert the example video on the first screen (18 seconds of synth music) at 192 kbps and you get an MP3 of 432,576 bytes (433 KB). The estimate on screen and the real file size match to the byte.

When you only need the sound from a video

A video file is heavy because of the picture, and you have to keep the screen on to play it. Once the sound is saved as an MP3 the file is much smaller, and it works in music apps, car stereos and audio editors.

• Lecture and meeting recordings: keep only the speech from a screen recording or a video call, so you can listen on the go or feed a transcription tool. At 128 kbps one hour comes to about 57.6 MB.
• Concert and rehearsal videos: cut a single song out of a phone video of a gig or a band practice, then keep it as an MP3 or send it to the people you played with.
• Editing prep: pull the sound out of your footage first to clean up noise in an audio editor, or reuse it as background audio for another video.
• Voice records: conversations with your kids, interviews and family events, where the voices matter more than the picture, fit in a small file.

Downloading someone else's video or music and reusing it can infringe copyright. This tool opens only files already on your device, and it has no feature for fetching a video from a web address.

How to use it

1. Choose a file: press 'Choose my MP4' or drop a file anywhere on the workspace. MP4, MOV and M4V files open.
2. See what is inside: the composition bar shows the share of video, audio and other data, plus the audio format (for example AAC · stereo · 48 kHz · 128 kbps). If there is more than one audio track, track buttons appear below.
3. Select a range: drag the handles at either end of the waveform, or type a time such as 1:05.5 in the Start and End boxes. 'Play selection' plays only the chosen part, and clicking the waveform plays from that spot. The first 1.6 seconds of the example video are nearly silent, so moving the start to 0:01.6 gives a 16.4-second MP3.
4. Pick a quality: each of the 128, 192 and 320 kbps options shows the size the current range would have. The default is 192 kbps.
5. Convert and save: press 'Convert to MP3' and the selected range fills in on the waveform from the left. When it finishes, switch between 'Original' and 'MP3' to compare, then download with 'Save MP3'.

An MP4 keeps its video and audio separate

MP4 is a container that holds video, audio, subtitles and other data together, and the audio inside is compressed with its own codec, usually AAC. The MP4 format derives from the ISO base media file format, which in turn derives from Apple's QuickTime (MOV) format, so an iPhone MOV is read the same way. Renaming .mp4 to .mp3 does not change the data inside.

Inside the file, a box called moov works as a table of contents. For each track it lists the size of every sample (stsz), where each chunk of samples sits (stco), how chunks map to samples (stsc) and where playback starts (elst). This tool reads only that index, then reads just the bytes that hold audio samples. The numbers in the composition bar are the sum of the sample sizes in that index.

If a file has video at 16 Mbit per second and audio at 128 kbit per second, the audio is only about 0.8% of the total (128 ÷ (16,000 + 128)). Even in a 1 GB video, the audio actually read is about 8 MB. Whether the phone shot the video as 'Most Compatible' (H.264) or 'High Efficiency' (HEVC) makes no difference to the result.

WebCodecs AudioDecoder is available in Chrome 94, Firefox 130 (desktop) and Safari 26 and later. In a browser without it, such as Firefox for Android, even AAC goes through the Web Audio path, so only audio up to 20 minutes long can be converted, even if you select a shorter range. That is because Web Audio's decodeAudioData accepts only complete file data, so all of the audio has to sit in memory at once.

128, 192 or 320 kbps: which one to pick

kbps is the amount of data used per second, in kilobits. This tool saves at a constant bitrate (CBR), so the file size is set by the audio length and the bitrate. For mostly speech, 128 kbps is often enough. For music or recordings with many sounds at once, 192 kbps or higher is a safe choice. 320 kbps is the highest value you can pick in MP3's MPEG-1 mode.

When the composition bar shows AAC audio, converting to MP3 adds one more round of lossy compression. AAC was designed to compress more than MP3 while improving fidelity, so if the source is AAC at 128 kbps, saving the MP3 at 192 kbps stays closer to the original. On the other hand, going past twice the source bitrate mostly makes the file bigger with little audible gain, so this tool marks those options 'This only adds size'. Sound or noise already lost from the source does not come back at a higher bitrate.

The size is 'duration in seconds × kbps × 1,000 ÷ 8' plus one MP3 frame. MP3 stores 1,152 samples per frame, so with this tool's encoder the frame count is ⌈samples ÷ 1,152⌉ + 1 and the size is frames × 144,000 × kbps ÷ sample rate, in bytes, rounded down. For example, one minute of 48 kHz audio is 2,880,000 samples and 2,501 frames, and at 192 kbps one frame is 576 bytes, which gives 1,440,576 bytes. Audio at 44.1 kHz comes out a few hundred bytes larger for the same length (1 minute at 192 kbps is 1,440,705 bytes). MB in the table means 1,000,000 bytes.

Trimming and previewing

You can drag a handle with the mouse or a finger, or focus it and use the arrow keys for 0.1 seconds, Shift plus an arrow key for 1 second, and Page Up or Page Down for 10 seconds. In the Start and End boxes, type 1:05.5 (1 minute 5.5 seconds) or 65.5, and 1:02:03 once you pass an hour. The end must be at least 0.1 seconds after the start.

'Play selection' plays only the chosen range and returns to the start when it reaches the end. After you make the MP3, press 'Original' and 'MP3' in turn to compare the same spot. If you cannot decide between 128 and 192 kbps, the quickest way is to convert a short range twice and listen to the same passage in both.

MP3 stores audio in 1,152-sample frames, so the last frame is padded with zeros (silence), and the player adds a short delay while decoding. As a result the MP3 is slightly longer than the selected range. The MP3 made from the 18-second example reads as 18.024 seconds.

Videos with several audio tracks

A video can hold more than one audio track. A video with dubbed languages and a screen recording that saves the microphone and the game sound separately are typical cases. Some players play only the first track, but this tool shows the format, channels, sample rate, length and language code (eng, kor and so on) of each track as buttons, and redraws the waveform for the track you choose.

The sample rate and channels of the result follow the source. A 48 kHz source gives a 48 kHz MP3, and mono stays mono, so no stereo width is invented. A sample rate MP3 does not support (96 kHz, for example) is saved at 48 kHz, and 3 or more channels (5.1 and so on) are mixed down to stereo.

How it converts without an upload

The old version downloaded an FFmpeg engine (about 32 MB) and loaded the whole file into memory, so it accepted files only up to 100 MiB. The current version decodes the audio with the decoder already built into your browser and downloads only the LAME encoder (about 0.14 MB) for saving the MP3. Audio is turned into MP3 as it is decoded and the decoded sound is discarded, so even long videos use little memory.

• The file is read only on this device, and the MP3 is created on the device too. The only network traffic is downloading the page and the encoder.
• There is no file size limit. A single conversion covers up to 3 hours (about 432 MB at 320 kbps). For anything longer, convert it in sections.
• The conversion runs as a separate task behind the page (a Web Worker), so the page does not freeze. On a phone, it is safer not to stay away in other apps for long while it converts.

Common mistakes

• Only renaming the extension: a file renamed to .mp3 is still an MP4 inside. Some players open it anyway, but apps and upload forms that accept only MP3 may reject it.
• Previewing and then closing the page: the preview and the 'Done' message are not a save. Press 'Save MP3', then open the file from your downloads list or the Files app to check it.
• Copying a file before the recording finished: a recording that stopped midway or a half-downloaded file has no moov index or a cut-off end. Without the index it cannot be opened, and if only the end is cut, you are told that only the first part was read.
• Result differs from what the player played: in a video with several audio tracks, the track your player chose can differ from the one this tool picks first. Check the language code and channels on the track buttons.
• Using a high bitrate to remove noise: 320 kbps only reduces compression loss. It does not fix noise or a muffled voice.

Limits and cautions

• One file opens at a time. To convert several videos, save one, then choose the next.
• Title, artist and cover art are not carried into the MP3, and neither are video or subtitles.
• It does not boost volume, remove noise or separate voices. The audio is only converted to MP3 as it is.
• Even when the source audio is already MP3, it is compressed again rather than copied out as it is. The bitrate is constant (CBR): 128, 192 or 320 kbps.
• WebM, MKV and AVI are different containers from MP4 and cannot be opened. Protected (DRM) content is not decrypted.
• Refreshing the page or choosing another file clears the current result. Save the MP3 you need first.

As of The standards and browser documentation were checked on 2026-10-20. The size table and example results were measured the same day in Chrome with this tool's encoder.

Sources

• 01W3C — AAC WebCodecs RegistrationCodec string mp4a.40.2 (AAC-LC) and others; the description is an AudioSpecificConfig (ISO/IEC 14496-3)
• 02MDN — AudioDecoderThe WebCodecs interface for decoding audio chunks; available from Chrome 94, Firefox 130 and Safari 26
• 03MDN — BaseAudioContext.decodeAudioData()Accepts only complete file data, and resamples the result to the AudioContext's sample rate
• 04MDN — Web audio codec guideMP3 in MPEG-1 mode: 32 to 320 kbps at 32,000, 44,100 and 48,000 Hz. AAC was designed to compress more than MP3 while improving fidelity
• 05LAME — Technical FAQOne MP3 frame = 1,152 samples, the last frame is padded with zeros, decoder delay is 528 samples. LAME is LGPL-licensed
• 06Apple — QuickTime File Format: Sample size atomDefinitions of stsz (sample sizes), stco (chunk offsets), stsc (sample-to-chunk) and elst (edit list, where playback starts)
• 07MDN — Media container formatsMP4 derives from the ISO base media file format, which derives from the QuickTime format
• 08Apple Support — Using HEIF or HEVC media on Apple devicesCamera formats 'High Efficiency' (HEVC) and 'Most Compatible' (H.264)
• 09wasm-media-encoders — LAME MP3 encoder for WebAssemblyMIT license (the encoder itself, LAME, is LGPL). This tool uses version 0.7.0 as is

Frequently asked questions

Tools to use with this one