Turning Video Into MP3: The Quality Settings That Actually Matter

Here's the fact that saves you the most disk space: the audio track inside a typical MP4 is AAC at somewhere between 128 and 256 kbps. When a converter offers you "320 kbps MP3 - highest quality!", it's offering to wrap that same audio in a fatter container. The extra kilobits store nothing. You cannot pour a litre out of a half-litre bottle.

Once that clicks, every quality decision about extracting audio gets easier. This is a guide to making those decisions with actual numbers instead of vibes.

What's inside your video's audio track

It depends on where the video came from, and the differences are big:

• Phone recordings: AAC, usually 96-256 kbps depending on the phone and settings. Voice memos sit at the low end.
• Videos downloaded from streaming platforms: typically 128-160 kbps AAC or Opus. This is the ceiling no converter can raise.
• Camera and DSLR footage: 256 kbps AAC or better; some record uncompressed PCM.
• Screen recordings: all over the map — OBS defaults to 160 kbps AAC, QuickTime uses 128+.

The source bitrate is your quality budget. Every choice downstream can only spend it, never grow it.

The one rule: lossy to lossy always costs something

AAC and MP3 are both lossy formats, and they discard different things. Converting between them means decoding one set of compromises and re-encoding into another. Done once at a sensible bitrate, the loss is inaudible to almost everyone. But it's real, it accumulates with every generation, and it's why you match or exceed the source bitrate when converting — never undercut it.

Practical version: audio that started at 160 kbps AAC should become a 192 kbps MP3, not a 128.

Picking the MP3 bitrate: a cheat sheet

• Speech — lectures, interviews, podcasts: 96-128 kbps. Voice has narrow frequency content; 128 kbps is transparent for talk. A one-hour lecture lands around 55 MB.
• Music, general listening: 192-256 kbps. The range where most people stop hearing differences in blind tests. About 1.4-1.9 MB per minute.
• 320 kbps: justified only when the source itself is high-bitrate or lossless. From a 128 kbps web video, it's pure padding — 2.4 MB per minute storing silence about silence.
• Mono switch: if it's a single voice, mono halves the file with zero practical loss. Worth remembering for long recordings.

The smarter option nobody mentions: don't re-encode at all

If your goal is just "audio file out of a video file", there's a lossless route: copy the AAC track straight out of the MP4 into an .m4a file. No decoding, no re-encoding, no generation loss, and it's nearly instant because nothing is recomputed. FFmpeg does this with the stream-copy flag, and the resulting m4a plays on any modern phone or computer.

So why does MP3 still exist as the default? Compatibility with the long tail: car stereos, older hardware, some audio editors, and podcast platforms with strict ingest requirements. MP3 plays on literally everything manufactured in the last 25 years. When you need that guarantee, converting is the right call — you can pull an MP3 out of an MP4 online without installing anything, pick the bitrate, and be done in under a minute.

Sample rates, and the 48 kHz question

Video audio runs at 48 kHz; music files traditionally use 44.1 kHz. Converters resample between them silently, and despite forum lore, a single competent resample is not something human ears detect. Where people DO get burned is loudness: video mixes often sit quieter than music mixes, so an extracted track can sound weak next to songs in a playlist. That's not a conversion defect — it needs loudness normalisation, which is a separate step.

One more myth worth retiring while we're here: converting the result to WAV afterwards doesn't upgrade anything. A WAV made from a 128 kbps source is the same 128 kbps of information in a container ten times the size.

When to bother installing FFmpeg

Honest split: for one video a week, an online converter wins — nothing to install, works on a locked-down office machine, done before FFmpeg would have finished downloading. FFmpeg earns its place when you're batch-processing a lecture series, working with files over upload limits, or need stream copy specifically. It's a command-line tool with a learning curve; there's no shame in not wanting that for a single file.