Getting audio out of a video without losing quality
The honest answer is that MP4 to MP3 always loses something — and that you almost never have to accept it, because the audio can come out exactly as it went in.
Free on iPhone and iPad, for the videos already in your camera roll. Works offline; one free conversion, then a subscription.
Almost every page that answers this question says "choose 320 kbps". That is the wrong answer, and it is wrong in an interesting way: the setting cannot fix the problem, because the problem is the format you picked before you got to the settings.
Why MP4 to MP3 always loses something
A video's soundtrack is already compressed. Your phone recorded AAC, your browser recorded Opus, the DVD carries AC-3. Every one of those threw away detail when it was created — that is what lossy compression means.
When you ask for MP3, that compressed audio has to be decoded back to raw samples and then compressed again by a different encoder with different ideas about what to discard. The second encoder cannot see what the first one removed, so it makes its own cuts on top. This is generation loss, and no bitrate setting prevents it: at 320 kbps you get a much larger file that is still a second-generation copy.
The way out is a copy, not a conversion
A video file is a container. Inside it, the soundtrack sits as a self-contained stream. If the target format can legally hold that same stream, nothing needs to be decoded at all — the stream is lifted out and given a new wrapper. The audio that comes out is the audio that was in the video, sample for sample.
| Your video | Soundtrack inside | Untouched extraction |
|---|---|---|
| MP4, MOV, M4V | AAC | M4A — a straight copy |
| WebM | Opus | .opus — a straight copy |
| AVI (older files) | MP3 | MP3 — a straight copy, the one case where MP3 is free |
| MKV | AAC, FLAC, Opus… | Whichever matches: M4A, FLAC, .opus |
| MKV, TS, VOB | AC-3, DTS | None of our formats hold these — FLAC keeps every sample instead |
| Camera / editing master | Uncompressed PCM | WAV or FLAC — both lossless |
The extractor on this site picks the copy automatically whenever you choose a format that allows it, and the panel says "Nothing is re-encoded" before you press the button. When it cannot, it says why in the same place.
How we know it is really identical
Because it is tested on every build, against real files rather than assumptions. The test suite creates a set of videos — an MP4 with AAC, a QuickTime master with 24-bit PCM, an MKV with three soundtracks, a WebM with Opus, an AVI with MP3, an MPEG-TS capture, a WMV — runs the site's own extraction plans against them, and then compares the audio stream that came out with the audio stream that was inside the video. A plan that claims "copy" fails the build if those differ by a single byte.
One honest refinement, because it is the kind of detail that gets rounded off elsewhere: the file is not byte-identical to anything. FFmpeg writes its own container around the copied stream, with its own metadata and timestamps. The audio is identical; the wrapper is new. That is the claim, and it is the one you can check.
The cases where a copy is refused
- You folded the audio to mono. That changes the samples themselves, so it has to be re-encoded. The panel switches its wording the moment you choose it.
- You picked a lower bitrate on purpose. You have asked for a smaller file, and handing back a copy would not be doing the job.
- The codec cannot live in that container. AAC cannot go into a WAV. FFmpeg will actually let you try this and report success — the resulting file plays nowhere — which is why this site keeps its own whitelist of legal pairs rather than trusting the tool to refuse.
"But WAV is lossless, so surely that is better?"
WAV is uncompressed, which is not the same as restorative. Converting a 128 kbps AAC soundtrack to a 24-bit WAV gives you a file forty times the size containing exactly the audio that the AAC decoded to — every flaw preserved at enormous cost. WAV is the right answer when an editor or a transcription tool demands uncompressed input, and a waste of disk space otherwise.
FLAC is worth knowing here: it is genuinely lossless compression, so it holds the same samples as the WAV at around half the size. If your video's audio is uncompressed PCM — a camera master, a capture card — FLAC keeps all of it and is the better file.
Four rules that cover everything
- If the file is for you, take the format that copies. M4A for phone video, .opus for WebM.
- If something else demands MP3, use MP3 once, at 192 kbps or the source bitrate — whichever is lower.
- Never ask for a higher bitrate than the soundtrack already has. It cannot add detail; it only adds bytes.
- Only reach for WAV when a tool insists on it.
The extractor picks the copy automatically when the format allows it, and says so before it runs.