MP3, MP2 and the legacy you'll still see

Why this matters

If you build, buy, or operate a video product, you will receive MP3 and MP2 files for years to come — from a radio station's playout system, a legacy archive, a broadcaster's satellite feed, or a customer who only knows the word "MP3". You need to know what these formats can and cannot do, whether you can use them without paying a license, and when to convert them to something modern before they hurt your quality budget. This article is written for a product manager or founder with no audio background; every claim is checked against the original standard or the patent holder's own statements, not a secondhand summary. Read it and you will never again confuse "the file is an MP3" with "the audio is fine".

One standard, three layers

The story starts with a single document. In August 1993 the Moving Picture Experts Group — the same MPEG that standardizes video — published ISO/IEC 11172-3, the audio part of the MPEG-1 standard. That one document defined three audio compression methods of increasing complexity, called layers. Layer I was the simplest and is effectively extinct. Layer II became MP2. Layer III became MP3. The names "MP2" and "MP3" are just shorthand for "MPEG-1 Audio Layer II" and "MPEG-1 Audio Layer III" — the number is the layer, not a version.

All three layers share the same core idea, the one explained in how audio compression works: use a model of human hearing to throw away sound the ear cannot detect, then pack what remains. The difference is how hard each layer works. Layer III (MP3) runs a more aggressive analysis with a finer frequency transform and an extra entropy-coding stage, so it squeezes the same audio into a smaller file. Layer II (MP2) does less work, which sounds like a disadvantage until you see where it is deployed.

MP3 was developed largely at the Fraunhofer Institute in Germany, with a team led by Karlheinz Brandenburg, building on perceptual-coding research from the late 1980s. MP2 grew out of a parallel European effort (the MUSICAM system) that fed into the same standard. Both were finished and frozen into ISO/IEC 11172-3 together. That shared birth certificate is why people who only know MP3 are often surprised that MP2 is a peer, not a predecessor.

Figure 1. One standard, two survivors. MP2 and MP3 were both frozen into ISO/IEC 11172-3 in 1993; MP3 conquered consumer audio, MP2 settled into broadcast, and MP3's patents finally expired in 2017.

Why MP3 won the consumer world

MP3 arrived at the exact moment three things lined up: hard drives big enough to hold an album, internet connections fast enough to move one, and a compression ratio good enough that a four-minute song fit in about four megabytes instead of forty. The arithmetic is the whole story. A CD-quality stereo track is raw pulse-code modulation — the sound wave measured 44,100 times per second, 16 bits per measurement, two channels:

44,100 samples/s × 16 bits × 2 channels = 1,411,200 bits/s ≈ 1,411 kbps

MP3 at 128 kbps cuts that by eleven times:

1,411 kbps ÷ 128 kbps ≈ 11× smaller

An eleven-fold shrink turned a 40-megabyte song into a 3.5-megabyte download, and a 1990s dial-up connection could finally move music in minutes instead of hours. That ratio, not any single feature, is why MP3 — and not MP2 — became the format of Napster, the iPod, and two decades of music libraries. MP2 at the same perceived quality needs a noticeably higher bitrate, so it lost the consumer race on file size alone.

Why MP2 never died

Here is the part that surprises people. The format that lost the consumer war is the one still embedded in professional broadcast infrastructure in 2026. MP2 survives because the things it is good at are exactly the things broadcast engineers care about, and the thing it is bad at — file size — barely matters on a dedicated transmission link.

MP2 has three properties that MP3 traded away for compression. First, lower encoder latency: MP2 analyzes audio in smaller, simpler chunks, so the delay added by encoding is shorter. On a live broadcast, where audio must stay locked to picture, every millisecond of codec delay is a millisecond of lip-sync risk to manage. Second, better error resilience: MP2 frames are more independent of one another, so a single corrupted packet on a noisy satellite or terrestrial link damages less audio and recovers faster. Third, transparent quality at higher bitrate with low complexity: at the 256–384 kbps that broadcast links can easily afford, MP2 sounds clean and costs little CPU to decode on millions of cheap receivers.

Those properties explain its deployment map. In the United Kingdom, the original DAB digital radio system still broadcasts in MP2 today; the newer DAB+ standard used elsewhere in Europe switched to HE-AAC, but the UK's installed base kept MP2. DVB digital television carries MP2 audio on many channels, often alongside Dolby Digital. And in the studio-to-transmitter and contribution world — the professional links that move live audio over satellite, ISDN, and IP — MP2 is a common default because it is robust, low-delay, and universally decodable.

Pitfall — "MP2 is just an old MP3" is wrong and will cost you. They are siblings, not generations. If a broadcaster hands you an MP2 contribution feed and you assume MP3 tooling will treat it identically, you can mis-detect the format, pick the wrong decoder, or transcode it twice and add audible artifacts. Always read the actual codec field in the container, not the file extension or the customer's description.

The 2017 patent thaw: what actually changed

For two decades, using MP3 in a commercial product meant paying royalties. Fraunhofer and Technicolor (formerly Thomson) ran a joint licensing program: a hardware maker shipping an MP3 player, or a software company shipping an encoder, took a license and paid a fee. That program ended in 2017.

The mechanism was simple — patents expire. The last patent in the program, U.S. Patent 6,009,399 held by Technicolor, expired on April 16, 2017, and the licensing program formally closed on April 23, 2017. Fraunhofer stated the program ended "due to the fact that the last patent included in the program expired." The practical result: the core MP3 patents are gone, and you can implement, ship, and use MP3 without paying anyone for those patents.

One caution from Fraunhofer's own statement, worth quoting because the popular "MP3 is free now" headlines skipped it: the end of the core program does not automatically mean every MP3-related patent has expired. There may still be implementation-specific patents on particular encoder optimizations. For the baseline format — encode, decode, play — you are clear. If you are licensing a specific high-performance commercial encoder, confirm its terms. In practice, for almost every product team, MP3 is now royalty-free.

MP2's relevant patents expired even earlier, which is part of why it stayed welcome in open broadcast standards.

So should you use MP3 or MP2 in 2026?

Almost never as a new choice — and frequently as a received one. The table below maps the realistic decisions.

The efficiency gap is the reason new projects skip MP3. A 128 kbps AAC file sounds roughly as good as a 192 kbps MP3 — about 30 to 50 percent more efficient — so choosing MP3 for a new service means paying more bandwidth for less quality. The one durable reason to encode MP3 today is reach: a baseline MP3 decoder exists in essentially every device ever made, so MP3 remains the safest format when you genuinely cannot predict what will play your file.

Where Fora Soft fits in

In our video streaming, OTT, and broadcast-adjacent projects, the legacy-audio question is routine: a client's archive or live feed arrives as MP3 or MP2, and the job is to ingest it cleanly, decode it correctly, and re-encode to a modern codec for delivery without stacking lossy generations. In e-learning and telemedicine work, we still accept MP3 uploads because users record on whatever tool they have, then normalize and transcode server-side. Knowing exactly what each legacy format guarantees — and what it costs in quality when re-compressed — is part of building an ingest pipeline that does not quietly degrade audio. Our audio processing software development team builds these normalize-and-transcode ingest stages, and our music app development work deals with legacy MP3 catalogs every day.

The book

Audio for Video: A Complete Guide to Sound in Video Products: From PCM and Codecs to Loudness, Accessibility, and Real-Time Voice Processing

This article tells you what MP3 and MP2 are and when you are allowed to keep them; the volume spends 1,229 pages on the codecs that replaced them and on the ground both generations stand on — sampling, bit depth, channels and loudness, then AAC-LC through xHE-AAC, Opus, and the speech codecs a call actually runs on. It is the book to open when a legacy file lands in your ingest and you have to decide what to transcode it into, at what bitrate, and what that re-compression costs you. Written by Nikolay Sapunov, CEO at Fora Soft.

Read it on Amazon →Kindle & paperback · Fora Soft Video Engineering Handbook

See our case studies

References

1. ISO/IEC 11172-3:1993 — Information technology — Coding of moving pictures and associated audio for digital storage media (MPEG-1), Part 3: Audio. ISO/IEC, 1993. Tier 1. https://www.iso.org/standard/22411.html — The single standard that defines Layer I, Layer II (MP2), and Layer III (MP3). Normative text paywalled; used for the layer definitions and the shared psychoacoustic basis.
2. "Alive and Kicking — mp3 Software, Patents and Licenses." Fraunhofer IIS Audio Blog, 18 May 2017. Tier 3 (first-party maintainer). https://www.audioblog.iis.fraunhofer.com/mp3-software-patents-licenses — Fraunhofer's own statement that the licensing program ended because the last included patent expired, and the caveat that implementation-specific patents may remain. Primary source for the patent-thaw section; overrides popular "MP3 is dead/free" press summaries that omitted the caveat.
3. mp3 product page. Fraunhofer IIS. Tier 3 (first-party). https://www.iis.fraunhofer.de/en/ff/amm/consumer-electronics/mp3.html — Fraunhofer's role in MP3 development and the end-of-licensing notice.
4. "30 Years of .mp3: Three Letters That Changed the World." Fraunhofer IIS, 2025. Tier 3 (first-party). https://www.iis.fraunhofer.de/en/magazin/panorama/2025/30-years-of-mp3.html — MP3 origin, Karlheinz Brandenburg, and the 1993 standardization timeline.
5. U.S. Patent 6,009,399 (Technicolor/Thomson). USPTO. Tier 1 (primary legal). Cited via Fraunhofer/Technicolor statements; the longest-running MP3 patent in the program, expired 16 April 2017. Used for the patent-expiry date.
6. MPEG-1 Audio Layer II — deployment in DAB and DVB. ETSI / DVB standards context, summarized. Tier 4. https://en.wikipedia.org/wiki/MPEG-1_Audio_Layer_II — Used for orientation on current UK DAB (MP2) vs DAB+ (HE-AAC) split and DVB audio carriage; deployment facts cross-checked against broadcast-vendor sources.
7. Fraunhofer IIS — MPEG-1 Layer II / MUSICAM heritage and broadcast use. Fraunhofer Audio Blog (Broadcast/DAB tags). Tier 4. https://www.audioblog.iis.fraunhofer.com/tag/broadcast — Context for MP2's broadcast role and the DAB → DAB+ codec transition to HE-AAC.
8. AAC vs MP3 efficiency comparison. Industry/codec comparison sources. Tier 7 (orientation only). https://www.gumlet.com/learn/aac-vs-mp3/ — Used only for the widely-reported "128 kbps AAC ≈ 192 kbps MP3" efficiency rule of thumb; the underlying claim is consistent with AAC's design goal stated in ISO/IEC 14496-3.