Audio Extractor

Lossless formats do not restore quality

The thing to understand before choosing a format, because it saves you from a common and expensive mistake.

The audio inside a video file is already compressed. In an MP4 it is almost always AAC, encoded once when the video was made, and some detail was discarded at that point. Extracting it means decoding that audio and encoding it again.

So choosing WAV or FLAC does not recover anything. You get a lossless copy of a lossy source — a file ten times the size containing exactly the information the AAC track had, and no more. The quality ceiling was set when the video was made and nothing here can raise it.

What lossless does give you is no further loss. That matters if the audio is going to be edited, processed and exported again, because each lossy encode compounds on the last. Extract to WAV, do your work, and export to MP3 once at the end.

The same logic explains the bitrate. 192 kbps is a sensible default and it is what you get. Pushing to 320 on audio that came out of a video mostly produces a larger file rather than a better one — you cannot encode detail that is not in the source. Go higher only if the video was made from a high-quality master.

Which format to pick

MP3 for listening, sharing and anything that has to work everywhere. Nothing refuses to play an MP3, which is why it is the default.

WAV if the audio is going into an editor. Uncompressed, universally supported by editing software, and the bit depth is yours to choose — 16-bit is CD quality and right for almost everything; 24-bit is worth it only when the source genuinely has more dynamic range than 16 bits can hold, which audio from a video rarely does.

FLAC when you want lossless but care about size. It compresses without discarding anything, typically to around half of WAV, and most modern players handle it.

AAC is technically better than MP3 at the same bitrate and is the native format in the Apple ecosystem. Worth choosing if you know what will play it.

OGG Vorbis is the open, patent-free option, common in game engines and on the web.

Be ready for the size. A five-minute WAV at 16-bit is roughly 50 MB, where the same audio as a 192 kbps MP3 is about 7 MB.

Loudness normalisation is not a volume slider

The option most worth turning on, and the one that needs explaining.

Turning the volume up multiplies everything, so it lifts the quiet parts and the loud parts together — and the loud parts clip and distort before the quiet parts get usable. Loudness normalisation measures how loud the audio actually sounds across the whole track, using a broadcast standard built around human hearing rather than raw signal peaks, then adjusts it to a consistent target level.

Two situations where it earns its place. Quiet dialogue — a lecture, an interview, a phone recording where the speaker was too far from the microphone — comes back at a comfortable level without the loud moments tearing. And several clips that need to match: extract each one with normalisation on and they arrive at the same perceived loudness, instead of you riding the volume between them.

Leave it off for music you are happy with, and for anything going into an editor where you would rather do the levels yourself. It is a useful automatic adjustment, not a mastering chain.

Trim before extracting, not after

If you only want a section — the two minutes of a talk that matter, one song out of a recorded set, a sound effect from the middle of a clip — set the range in the tool rather than extracting everything and cutting it afterwards.

It is faster, because only that section is processed. The file is smaller. And it is one step rather than two, which on a long video is the difference between a few seconds and a wait.

Times go in as minutes and seconds, so 2:30 to 5:00 gives you two and a half minutes from two and a half minutes in.

Nothing is uploaded

The extraction happens on your own machine. That matters for the obvious reason — recorded meetings, interviews, lectures, client footage and personal video are exactly the files that should not go to a stranger's server — and for a practical one: there is no upload wait, no queue, and no file size limit set by someone else.

The cost is that the audio engine has to load into your browser the first time you use it, which is a one-off download of a few tens of megabytes. After that it is cached.

Related tools

Video to MP3 is the simpler version of this — one click, MP3 out, no settings to consider. Use it when you just want the audio and this one when the format, the level or the section matters. Trim Video cuts the video itself, and Remove Audio from Video does the opposite job: keeping the picture and dropping the sound.