A protracted rant by the author of Ennuicastr

The digital audio hobbyist and podcasting communities are stuck in a time warp. Trapped, using a format from 1992, under the bizarre misapprehension that either it's the best format for the job, or that they need to use MP3 because of its universal support. But, the community got stuck like this at the turn of the century, when those reasons were true and valid. It's been over twenty years now. It's time to ditch from our common knowledge the idea that MP3 is an acceptable format.

MP3 is terrible.

It is far more terrible than you probably think that it is.

There is absolutely no reason to use MP3, ever. No, really: ever.

Why? Well, let's rant.

First of all, let's make one thing clear: you've almost certainly heard of MPEG-4, and probably assumed that MP3 is its immediate predecessor, MPEG-3. You may have even assumed that MP3 was the audio component, and MP4 the video component. This is all completely wrong; MP3 is an abbreviation of MPEG-1 Layer 3. That's right, it's not MPEG-3. It's MPEG-1. To drive that point, I will usually be calling it MPEG-1 audio from now on in this document.

MPEG-1 audio was created in 1992, and the MPEG-1 standard codified in 1993. It was created to be the state of the art for hardware audio encoding... in 1992. It had some restrictions and decisions that made sense for hardware audio encoding and decoding; in particular, constant bitrate, which is designed to work well with magnetic tapes run at a constant speed. As a consequence, even at the time, it was far from the best that the industry had to offer. It was the best that could reasonably be managed on digital magnetic tape.

In 1995, a new revision was released as part of the MPEG-2 standard, MPEG-2 Layer 3. MPEG-2 Layer 3's enhancements were fairly trivial, but still remain uncommon in “MP3” in the wild; MP3 remains MPEG-1 Layer 3, not MPEG-2 Layer 3. So, MPEG-1 audio was formally obsoleted in 1995, but continued to be used. MPEG-2 also included a superior container format (file format), but that's named .mp2, so everyone assumes it's worse than MP3 (since 2 is less than 3). Yes, MP2 files are better than MP3 files. Go figure. Actually, it's even more confusing than that, since some .mp2 files are MPEG-2 files, but other .mp2 files are MPEG-1 Layer 2 files...

In 1997, MPEG-2 was extended with MPEG-2 AAC (Advanced Audio Coding). AAC vastly improved over MPEG-1, and in particular, was no longer hampered by constant bitrate, which is a terribly inefficient way to encode anything, a carryover from magnetic tape and ISDN lines. At the time, AAC was a bit too advanced for most hardware, so it didn't get much use. So, the successor to MPEG-1 audio was itself formally obsoleted, but MPEG-1 audio continued to be used. (Note: In practice, some “MP3” files do not have a constant bitrate, but this isn't spec compliant, and can confuse many audio players.)

In 1999, the first revision of MPEG-4 was released, including MPEG-4 AAC. This improved MPEG-2 AAC in fairly minor ways while largely retaining backwards compatibility. But, more importantly, hardware was catching up; by 2004, essentially all hardware produced could play MPEG-4 AAC, and among industry professionals, MPEG-1 was long dead. But, MPEG is an abbreviation of “Misanthropic Patent Extortion Gang”¹, so hobbyist use of AAC didn't pick up much because of its expensive patent situation. Technically, the patent situation was the same for MPEG-1, but MPEG's modus oprandi is to gain more and more irrelevant patents over time to make every new technology have more and more litigious organizations behind it, and as a consequence, older technology is a bit safer. So, MPEG-1 audio limped along, now thrice-obsoleted. In 2000, it was still perfectly reasonable to use MPEG-1 audio.

In 2000, Xiph released Ogg Vorbis, a patent-royalty-free audio format. Vorbis has comparable quality to MPEG-1 audio, so it didn't make much of a splash among enthusiasts, but it actually became hugely popular in video games, which appreciated a compressed audio format for which they wouldn't need to pay patent royalties. At this point, Ogg Vorbis was the right moral choice, but didn't offer enough technologically to supplant MPEG-1 audio, and so MPEG-1 audio continued.

In the mid-2000s, DivX was hot for video, but actually, DivX was just MPEG-4 Part 2 video. This is worthy of note only because in the early-to-mid-2010s, the entire digital video community, hobbyist and professional, moved from MPEG-4 Part 2 to MPEG-4 Part 10 (H.264), demonstrating that this kind of change is possible. And yet, audio hobbyists, for some awful reason, continued to use MPEG-1.

In 2003, AAC's successor, HE-AAC, was released. HE-AAC is mostly an improvement at low bitrates, so there's little purpose in using it for high-quality audio. HE-AACv2 was released in 2006. HE-AACv2 is the current MPEG standard, so MPEG-1 audio's successor's successor's successor's successor is obsolete. At this point, the continued use of MPEG-1 audio was getting absurd. Nobody was putting MPEG-1 audio in videos, so why the Hell were they using MPEG-1 audio for just-audio releases?

Also in 2003, Apple introduced the iTunes Store. All audio on the iTunes store has been in AAC since its inception. iPods never “wanted” MPEG-1 audio; they were always designed to use MPEG-4. The same is true of every other music device from the last 20 years. If you upload your audio to any of Apple's services as MPEG-1 audio, they're just going to convert it to MPEG-4.

It's worth noting that MPEG, stimied by their own anti-innovative policies, hasn't developed anything of any value to anyone since H.264 SVC in 2007.

In 2012, the Opus group (IETF, Xiph, and others) released the Opus codec. The Opus codec is patent-royalty-free, and vastly superior to MPEG's offering. Since then, Opus has been adopted by all major browsers, is used by all standard voice chat applications, and is widely supported by operating systems and audio software. Opus is the current international standard, by any definition of “international standard” that doesn't involve suckling at the teat of the Misanthropic Patent Extortion Gang. The industry has moved from MPEG standards to Opus, making MPEG-1 audio's successor's successor's successor's successor's successor obsolete.

Also in 2012, MPEG-1 audio fell out of patent (1992 + 20 years). Thus, for a few years, there was one valid reason to use MPEG-1 audio: you wanted to avoid patents, but operated under the misapprehension that “MPEG” and “industry standard” were synonyms. In 2017, MPEG-2 AAC (baseline AAC) fell out of patent, and in 2019, MPEG-4 AAC fell out of patent, so the brief interlude during which patents were a reason is now over.

Why the history lesson? To make this point: MPEG-1 audio is hilariously obsolete. It's a terrible format that wasn't even good when it was new, it's been obsoleted so many times that I've had to carefully count the number of times to make sure that I was saying it right, and the industry had moved on by 2005. Imagine if you wanted to watch a vodcast, and their only option was to download a .mpeg file. That is exactly what you're doing when you release your podcast—or anything else—as MPEG-1 audio (MP3). It's the same standard.

So, if MPEG-1 audio is so hilariously obsolete, why do people keep using it? Let me answer that question with a comparison.

What's the name for a short, silent animation in a web page? If you said “GIF” (regardless of how you pronounce it), congratulations, you're wrong. Well, sort of.

GIF, the Graphics Interchange Format, is a hilariously obsolete format. But, it continued to roll along for a long time because nothing quite replaced it; no other image format was also a light animation format in the way that GIF was. So, even though GIF is terrible in a lot of ways, it continued to be used for decades.

Note, however, “used”. Past tense. Very little software actually cares what a “GIF” is—basically only web browsers—and so the entire web ecosystem pulled the rug out from under you in the 2010s. GIFs... aren't GIFs! Sometimes they're MPEG-4 video, but more often, they're WebP. WebP is a vastly superior format to GIF, so thank goodness.

But, if they're not GIF files, then why do we still say “GIF”? Because “GIF” has become a synonym for “short, silent animation in a web page”. If you're using that definition for “GIF”, then it's fine and correct to call WebP animations “GIFs”, even if they're not, technically, GIFs.

With audio software, the landscape is far larger and more diverse. We can't simply say “MP3 file” and use an Opus file instead, because there are too many pieces of software that would all have to agree on that change to naming convention. So, people continue to use MPEG-1 audio because the name is, in essence, a meme, even though it's ludicrously obsolete.

What can you do about it? Luckily, it's really simple: although audio users and producers tend to be ignorant, producers of audio software and hardware are not. So, just don't use MP3! For the widest compatibility, make MPEG-4 AAC files, and... that's it. They'll work everywhere, and people won't actually notice the difference (in terms of compatibility). Hell, you can even pull a GIF if you want to, and label it “MP3” as a synonym for “audio file”. If you use MPEG-4 AAC, that name isn't even a lie; MP3 is an abbreviation of MPEG-1 Layer 3, but it could just as easily be an abbreviation of MPEG-4 Part 3, which is AAC. Or, better yet, use Ogg Vorbis—the support is far better than you probably think that it is—or Opus!


¹ Technically this isn't the expansion that MPEG themselves would like to be known by, but it's the expansion that fits their purpose.

Responses to some likely complaints

“But my podcast host uses MP3!” Firstly, if they do, you shouldn't use them. Exactly how much incompetence should we tolerate from these services that ought to know better? Secondly, they almost certainly convert the audio themselves, so you should upload with a lossless format to avoid generational loss. Let them ruin your audio, instead of pre-ruining it for them.

“But everyone uses MP3!” Unfortunately, this is true, but it's because of no reason better than momentum. Everyone shouldn't be using MP3.

“But my listeners will be confused if it's not MP3!” All hardware and software supports MPEG-4 AAC, so your listeners won't even notice the difference if your “download” button downloads that.

Return to home page