MP3 or M4A


Here’s something I found that puzzles me: I am usually operating under the assumption that saving music as M4A is “better” than saving it as MP3.

Well, I decided to convert some WAV music (taken directly from a disc I own) into a slightly smaller file. I used Audacity to export the WAV as (1) MP3 (both 256 and 320 kbps), and (2) as M4A (257 and 320 kbps). I then used audacity’s “plot spectrum” on the WAV and the four smaller files. The results are clear: the MP3 is definitely better than M4A (“better” in the sense that the plot shows higher frequency sounds in MP3 than in M4A (both, at 256 and 320 kbps). Unsurprisingly, the WAV is best, but not by much when compared to the 320 MP3.

This seems more objective than a hearing test. I got the same results in another musical piece I tried, so - at least when the source is WAV, I am perplexed. (I guess it’s a different story when the source is MP3 or M4A)

Can anybody un-perplex me?


When the M4A format was invented, it was said to provide better quality than MP3 at the same bit-rate.
That “may” have been true in the 1990s (or it may have been marketing hype).

The LAME MP3 encoder has been in continuous development since then, and has substantially improved.
The most recent large scale double blind test results that I’ve seen were from around 2019, and concluded that for 256 kbps and higher, M4A, AAC, MP3 and OGG were all “transparent” (not audibly distinguishable for most listeners from the source audio). In other words, they are all about on par when it comes to high bit-rates.

(“Opus” format is the clear winner at very low bit-rates, being much better quality than MP3, OGG, M4A or AAC at very low bit-rates)

The first suspect is Plot Spectrum which, by design, aligns itself with the work, making it very nearly impossible to compare two or more readings. There is no way, that I know of, to fix the left and bottom edge calibrations.

Original Recordings, Edit Masters and Archival Work should all be in WAV. Running on the Beach can be in whatever format sounds the best to you and gives you the file sizes you like. MP3 is universal and M4A less so, so if you’re going to send files to someone or post work on-line, the success is most likely with MP3.

I guess it’s a different story when the source is MP3 or M4A

Never do production in MP3 or M4A. Compressed formats get their small files by re-arranging musical tones and leaving some of them out. If you make an MP3 from an MP3, it rearranges more tones and leaves more out. By the third pass, most people can hear the damage. Making a 128 quality MP3 from a 128 quality MP3 does not give you a 128 quality MP3. Quality always goes down. Making a WAV from an MP3 doesn’t cure it. It just postpones the damage until later when you forget the original was an MP3.

It’s downhill all the way and you can’t stop it.

M4A is reputed to be better at this multi-pass dance than MP3.

320 MP3.

Any reason you’re up in the compression stratosphere? What you’re supposed to do, and every techie does at least once, is make an MP3 from a perfect quality WAV and reduce the quality until you can hear the damage. You may find (as every techie does) that a mono sound file falls apart at 32 and stereo at 64.

That’s why for years the Audacity default was 128 Constant. For single-pass stereo, nobody can hear the damage. I don’t think it is any more. I think it’s one of the variable formats now.

You should pay attention to the job. Audiobook Submissions need to be MP3, 192 Constant, minimum, and it’s not optional.


Yes that is “reputed”, though it is probably no longer true, even if it was true in the '90’s.
It is certainly NOT true that M4A provides the best audio quality at very low bit-rates - Opus beats M4A by a mile at very low bit-rates. Unfortunately Opus format is not very widely supported at present (Opus support was only recently added to Audacity).