Why your MP3 sounds different after converting from MP4

People assume that pulling the audio out of a video is a copy operation. It usually is not. An MP4 file holds a video track and an audio track, and that audio track is almost always AAC. MP3 is a different codec, so the conversion cannot just lift the track out and rename it.

What actually happens

The converter decodes the AAC stream back to raw samples, then encodes those samples as MP3. Both AAC and MP3 are lossy, meaning each one throws away detail it judges you will not hear. The problem is that they disagree about what to throw away. The MP3 encoder is making its decisions about audio that has already had detail removed by AAC, and it cannot put back what is missing.

This is generation loss. It is the same reason a photocopy of a photocopy looks worse than the original, and it applies every time you convert between two lossy formats.

• AAC at 128 kbps to MP3 at 128 kbps sounds worse than the original, despite the matching number
• Cymbals, applause, and reverb tails degrade first, because that is where both codecs cut hardest
• Converting the same file back and forth a few times makes the damage obvious

What to do about it

Give the MP3 encoder more room than the source had. If the AAC track is 128 kbps, encoding to 192 kbps MP3 leaves enough headroom that the second pass has little to cut. The file gets bigger, which feels wrong, but the bitrate is not describing quality here. It is describing how much space the encoder has to work in.

Match or exceed the source bitrate when converting between two lossy formats. You cannot recover detail, but you can avoid removing more.

The short version

The other option is to skip the round trip. If you only need the audio and the player supports it, extracting the AAC track into an M4A container copies the stream without re-encoding. Nothing is decoded, nothing is thrown away, and the result is bit for bit what was inside the MP4.