Published: 2026-06-05 · Reading time: 21 min read · Author: Nikolay Sapunov, CEO at Fora Soft

Why this matters

If you build, buy, or operate a video product, the codec names in your spec sheet are not interchangeable trivia — each one carries a licensing bill, a device-support map, and a quality ceiling that was set decades ago and still constrains you today. A product manager who knows that AAC is the safe default, that Opus is free and owns real-time, and that LC3 lives in the earbuds your users wear can make faster, cheaper decisions and ask engineers sharper questions. This article is the orientation map for the whole codec block: read it once, and the deep-dive articles on AAC, Opus, and the rest will slot into a timeline you already hold. We wrote it for a smart reader with zero audio background; every date and standard number is checked against the primary source.

First, what a codec actually is — and why audio needed one

Before any history, fix the word. A codec is a matched pair of programs: a coder that compresses sound into a small stream of bits, and a decoder that expands those bits back into sound you can play. The name is just "coder-decoder" mashed together. Every codec in this article is one of these pairs; what changes across fifty years is how cleverly the coder throws bits away.

Why throw anything away at all? Because raw, uncompressed digital audio is enormous. Recall from What is digital audio that raw sound is stored as pulse-code modulation, or PCM — the wave's height measured tens of thousands of times a second and written down as plain numbers. Do the arithmetic for CD-quality stereo:

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

That is about ten megabytes per minute, for one stereo track. Stream that to a thousand viewers, or carry it on a 1990s phone line, and the cost is impossible. A codec's job is to get that 1,411 kbps down to something like 128 kbps — an eleven-fold shrink — while the ear hears almost no difference. The history below is the history of doing that shrink better and better.

There are two broad ways to shrink. Lossless compression packs the audio without discarding a single sample, the way a ZIP file packs a document; FLAC does this. Lossy compression permanently deletes detail the ear is unlikely to notice, which buys far smaller files; MP3, AAC, and Opus all do this. Most of this story is about lossy codecs, because that is where the hard, clever work happened.

Timeline of major audio codecs from G.711 in 1972 to LC3 in 2020, with speech codecs on a lower track and music or general codecs on an upper track, color-coded by standards body Figure 1. Fifty years of audio codecs on one timeline. The speech lineage (lower track) and the music or general-purpose lineage (upper track) ran in parallel for decades before Opus and xHE-AAC merged them.

The two tribes: speech codecs and music codecs

The single most useful idea for understanding codec history is that audio compression grew up as two separate families that solved two different problems, and barely talked to each other for forty years.

The speech tribe came from the telephone network. Its problem was narrow: carry one human voice, in real time, over a thin and unreliable link, using as little bandwidth as possible. Voice is predictable — it comes from a single throat with known physics — so speech codecs model the voice itself, predicting the next moment of sound from the last. This approach is called linear prediction, and it makes voice astonishingly compact, but it falls apart on music.

The music tribe came from broadcast and consumer media — CDs, digital radio, later the internet. Its problem was the opposite: carry any sound, including a full orchestra, at high fidelity, where a delay of a second is fine because nobody is talking back. Music codecs do not try to model the source; they transform the sound into frequencies and then exploit a quirk of human hearing called masking — a loud sound hides nearby quiet ones, so the quiet ones can be deleted. This works beautifully for music and is wasteful for a single voice.

Hold these two tribes in mind. Almost every codec below belongs to one of them, and the climax of the story — Opus in 2012, xHE-AAC the same year — is the moment the two tribes finally became one codec.

The four ideas behind every codec

Underneath the two tribes sit four core techniques. Every modern codec is some blend of these, and naming them now makes the rest of the history readable. They get their own full article in How audio compression works; here is the one-paragraph version of each.

Psychoacoustic masking is the music tribe's master trick: human hearing cannot detect a quiet sound played right next to a loud one of similar pitch, so the codec measures what the ear will miss and deletes exactly that. Frequency-domain transforms are the tool that makes masking usable — the codec converts a slice of the wave into a list of frequencies using a maths operation called the modified discrete cosine transform, or MDCT, which is the engine inside MP3, AAC, and the music half of Opus. Linear prediction is the speech tribe's master trick: model the voice as a buzzing source shaped by the mouth, and send only the small corrections, which is how CELP-style codecs squeeze voice into a few kilobits. Entropy coding is the cleanup crew shared by both tribes — a final lossless pass that packs the remaining numbers as tightly as mathematics allows, the same family of trick a ZIP file uses.

You do not need the maths to follow the history. You only need to recognise that a codec built on transforms and masking is a music codec, a codec built on linear prediction is a speech codec, and a codec that uses both is one of the modern unifiers.

1972–1988: the telephone era — G.711 and G.722

The first audio codecs were not for music at all. They were for telephones, and they came from the International Telecommunication Union, the ITU, the body that standardises global telecoms.

The grandfather of them all is ITU-T G.711, standardised in 1972. G.711 is barely a "codec" in the modern sense — it is companded PCM, a lightly clever way of storing 8 kHz, 8-bit telephone audio at 64 kbps using either the μ-law (North America, Japan) or A-law (Europe) curve to give quiet sounds more precision. It throws away almost nothing and compresses almost nothing. Its importance is its immortality: G.711 is still the universal fallback codec in every SIP trunk and PSTN gateway in 2026. When a modern video call cannot agree on anything fancier, it drops to G.711, and both ends just work. We cover this lineage in Speech codecs you'll still meet.

Sixteen years later, in 1988, the ITU approved G.722, the first wideband voice codec — it sampled at 16 kHz instead of 8 kHz, roughly doubling the audible range and giving telephone audio that "HD Voice" clarity where speech sounds present instead of muffled. G.722 still runs at 48, 56, or 64 kbps. It is the moment the speech tribe first reached past the muddy 3.4 kHz telephone band toward something pleasant.

These two set the template for the speech tribe: low delay, low bitrate, voice only, run by the ITU. For two decades that tribe and the coming music tribe developed in complete isolation.

1991–1993: MPEG arrives and the music tribe is born — MP2 and MP3

In the late 1980s a new committee formed under ISO and IEC, the international standards bodies: the Moving Picture Experts Group, or MPEG. Its job was to standardise digital video and its audio, for CD-ROM, digital broadcast, and digital television. MPEG would dominate the music tribe for the next thirty years.

MPEG's first standard, ISO/IEC 11172 — MPEG-1 — bundled three audio "layers" of increasing sophistication. All three layer algorithms were approved as a committee draft in 1991, finalised in 1992, and published as ISO/IEC 11172-3 in 1993. This is where our title's "1991" comes from.

Layer II, known as MP2, was the practical workhorse. It was simpler to encode in real time, which mattered on 1990s hardware, and it became the audio of digital broadcasting. Three decades later MP2 is still the audio codec on much of DAB digital radio and DVB-T digital terrestrial television in Europe — a striking example of a codec surviving by owning broadcast infrastructure that nobody wants to re-engineer. We dig into this in MP3, MP2 and the legacy you'll still see.

Layer III, the famous MP3, was the most sophisticated of the three and the hardest to encode. The work came from a consortium led by Fraunhofer IIS in Germany, with the University of Hannover, AT&T-Bell Labs, Thomson, and CCETT. MP3 was the layer that swept the consumer world — not because it was technically dominant over MP2 in broadcast, but because it hit a quality sweet spot at low bitrates just as internet bandwidth and the Napster era arrived. MP3 became a cultural noun. It also became the codec everyone had to license, and Fraunhofer's licensing program — and its eventual end in 2017, when the core patents expired and Fraunhofer terminated the program — is its own chapter of the story.

MP3's real significance for this history is that it proved psychoacoustic masking worked at scale. The music tribe now had a hit, and a business model.

1997–1999: AAC, the codec that quietly won — and still wins

MP3 had limits baked into its 1990 design. So MPEG built a clean-sheet successor: Advanced Audio Coding, AAC, first standardised in 1997 as MPEG-2 Part 7 (ISO/IEC 13818-7). AAC was deliberately not backward-compatible with MP3 — the committee even labelled it "NBC", Non-Backward-Compatible — because compatibility was the chain holding MP3's quality back. Freed from it, AAC delivered better sound than MP3 at the same bitrate, and much better at low bitrates.

In 1999 AAC was folded into the broader MPEG-4 family as MPEG-4 Part 3 (ISO/IEC 14496-3), where it has lived and grown ever since. That single standard number, 14496-3, is worth remembering: it is the home of the entire AAC family.

AAC's quiet victory is the most important fact in modern audio. The base profile, AAC-LC (Low Complexity), became — and remains in 2026 — the default audio codec of online video. YouTube, Netflix, Apple, Disney+, and most of the streaming world ship AAC-LC audio inside their video. It is the safe, universal, hardware-accelerated choice. We give it a full treatment in The AAC family.

The family then grew to cover lower bitrates. HE-AAC (High-Efficiency AAC, around 2003) added a trick called spectral band replication that synthesises the high frequencies instead of coding them, halving the bitrate for acceptable quality. HE-AAC v2 added parametric stereo, squeezing listenable stereo into about 32 kbps for mobile radio. These extensions kept AAC competitive as bandwidth got cheap and then, on mobile, expensive again.

2000–2002: the open revolt — Vorbis

The music tribe had a problem: it was patented. Every MP3 and AAC encoder owed license fees, and in September 1998 Fraunhofer announced it would charge them more aggressively. The reaction was an open-source revolt.

Chris Montgomery's Xiph.Org project answered with Vorbis, an MDCT-based music codec deliberately built to be patent-free and royalty-free. The bitstream was frozen in May 2000 and the 1.0 reference software shipped in July 2002. Vorbis, carried in the Ogg container, was technically excellent — competitive with AAC at many bitrates — and it found a durable home in video games and Wikipedia media, but it never dislodged MP3 or AAC from the mainstream. Its true importance is as a rehearsal: Xiph learned how to ship a serious open codec, and the same people would later build the codec that did win. We place Vorbis in context in How audio compression works.

1991–2015: the cinema and broadcast tribe — Dolby AC-3, E-AC-3, AC-4

Running alongside MPEG was a third force that belongs to neither pure tribe: Dolby, the company that owned cinema and home-theatre sound.

AC-3, marketed as Dolby Digital, was released in 1991 and reached cinemas with Batman Returns in 1992. Its breakthrough was multichannel: AC-3 was the first practical coding system to carry 5.1 surround — five full channels plus a low-frequency effects channel — in a single compressed stream. It became the audio of the DVD, of ATSC digital television in the US, and of home theatre. The full spec lives in ATSC A/52 and ETSI TS 102 366.

Around 2000, Dolby extended it to E-AC-3, marketed as Dolby Digital Plus (DD+), widening the bitrate range up to several megabits, raising the channel count, and adding the substream structure that streaming services need. DD+ is what carries surround and the base of Atmos on Netflix, Disney+, and Apple TV+ today. We cover both in AC-3 and E-AC-3.

Then Dolby started fresh. AC-4, developed from late 2011, was standardised by ETSI as TS 103 190 in 2014 and approved for commercial use by Dolby in December 2014. AC-4 is a next-generation broadcast and streaming codec — far more efficient than AC-3, built for object-based audio and immersive sound, and adopted in the ATSC 3.0 "NextGen TV" standard. It is the future-broadcast bet, covered in AC-4 explained.

2010–2012: the merger — Opus

Here is the climax of the whole story. For forty years speech codecs and music codecs were separate because they solved opposite problems. In 2012 that ended.

Two teams were each building half of the answer. Xiph.Org had been developing CELT, a very-low-delay MDCT music codec, since 2007. Skype had been building SILK, a modern linear-prediction speech codec for its calls. At the IETF — the internet's standards body, a different world from ISO and the ITU — the two efforts were merged into one codec that could switch between, or combine, a speech mode and a music mode inside a single stream. The result was Opus, standardised as IETF RFC 6716 in September 2012 (authors Valin, Vos, and Terriberry; later updated by RFC 8251).

Opus is the first codec that genuinely served both tribes. It scales from 6 kbps narrowband speech to 510 kbps stereo music, switches modes automatically as the content changes, and runs at delays low enough for live conversation. Critically, it is open and royalty-free, standardised through the IETF with a BSD-licensed reference implementation, libopus.

Opus did not displace AAC in stored video — AAC's hardware support was too entrenched — but it did something arguably bigger: it ate WebRTC. Every browser-based video call, every modern conferencing app, every real-time audio link defaults to Opus, because it is free, excellent, and built for exactly that job. It is the single highest-traffic codec in real-time media, and it gets its own deep dive in Opus: the open codec that ate WebRTC.

The same year, 2012, MPEG finalised its own unifier: MPEG-D Unified Speech and Audio Coding (USAC), standardised as ISO/IEC 23003-3 and marketed as xHE-AAC (Extended HE-AAC). xHE-AAC also crosses the speech-music divide in one codec, scales from 12 kbps to 300 kbps, and carries mandatory loudness control built in. It deploys more slowly than Opus because it is licensed, but it now powers Apple Music's loudness-managed delivery, Netflix mobile, and Amazon Prime Video. Two unifiers, same year, from two different worlds.

2015–2020: the new immersive and low-power codecs — MPEG-H, LC3

The merger solved the speech-versus-music problem. The last decade of this history is about two new frontiers: immersive audio and ultra-low-power audio.

MPEG-H 3D Audio, standardised as ISO/IEC 23008-3 in 2015, is the ISO answer to object-based, immersive sound — audio described as movable objects in a room rather than fixed channels, rendered to whatever speakers or headphones the listener has. It is the codec behind South Korea's ATSC 3.0 broadcasts and a competitor to Dolby's AC-4 in the next-generation-TV race. We treat it in MPEG-H 3D Audio explained.

The last milestone in our title is the quietest and, for most users, the most personal. LC3 — the Low Complexity Communication Codec — was adopted by the Bluetooth SIG on 15 September 2020 as the mandatory codec of Bluetooth LE Audio, with the full LE Audio specification completed on 12 July 2022. Developed jointly by Fraunhofer IIS and Ericsson, LC3 replaces SBC, the baseline Bluetooth codec that had shipped since version 1.0, and delivers equal or better quality at roughly half the bitrate. It runs on tiny frames of 7.5 or 10 ms to keep wireless-earbud latency low, and an LC3plus extension reaches high-resolution and broadcast use. LC3 is why your customer's earbuds matter to your video product: call quality, mouth-to-ear latency, and even hearing-aid support now flow through it. Full coverage in LC3 and LC3plus.

A worked comparison: how much smarter did codecs get?

Numbers make the progress concrete. "Transparent" below means most listeners cannot reliably tell the compressed audio from the original in a blind test, for typical stereo music. The bitrate needed for transparency is the cleanest single measure of a codec's efficiency.

MP3 (1993):     ~256 kbps for transparency
AAC-LC (1997):  ~192 kbps for transparency  (≈ 25% smaller than MP3)
Opus (2012):    ~128 kbps for transparency  (≈ 33% smaller than AAC-LC)

Walk the last step out loud. Going from AAC-LC's 192 kbps to Opus's 128 kbps for the same perceived quality is a saving of:

(192 − 128) ÷ 192 = 64 ÷ 192 = 0.333… ≈ 33% fewer bits

A third fewer bits, for sound the ear treats as identical, across fifteen years of research. Stretch the line back to the telephone era and the contrast is starker still: G.711 spends 64 kbps to carry one muffled voice, while Opus can carry clear wideband speech in under 24 kbps and full music in 128 kbps. That is the fifty-year dividend.

The geopolitics: who runs the standards

Codecs are not only technology; they are turf. Four kinds of body produced everything above, and knowing which one made a codec tells you a lot about its licensing and reach.

The ITU (ITU-T for telecoms, ITU-R for broadcast) made the speech codecs — G.711, G.722 — and the loudness and lip-sync standards. ISO/IEC MPEG made the music-tribe heavyweights: MP2, MP3, the AAC family, USAC/xHE-AAC, and MPEG-H. These are powerful, widely deployed, and licensed — you pay to ship an encoder. Dolby, a single company, made AC-3, E-AC-3, and AC-4, and licenses them as proprietary products with strong cinema and broadcast lock-in. The IETF — the internet's standards body — made Opus, and its culture of royalty-free, openly specified standards is why Opus is free where AAC is not. A newer fifth force, AOMedia (the consortium behind the AV1 video codec), pushes the same royalty-free philosophy into media broadly.

The practical upshot for a 2026 product: if you want free, reach for the IETF and Xiph lineage (Opus, Vorbis, FLAC). If you want universal device support, reach for the MPEG lineage (AAC). If you ship to broadcast or premium home theatre, you will meet Dolby whether you like the license or not.

Codec Year Tribe Standards body Licensing Still relevant in 2026?
G.711 1972 Speech ITU-T Free Yes — universal PSTN/SIP fallback
G.722 1988 Speech ITU-T Free Yes — HD Voice
MP2 1991/93 Music ISO/IEC MPEG Expired Yes — DAB radio, DVB-T
MP3 1993 Music ISO/IEC MPEG Expired (2017) Legacy — still everywhere, rarely the right new choice
AC-3 (Dolby Digital) 1991 Cinema Dolby / ATSC / ETSI Proprietary Yes — DVD, broadcast, home theatre
AAC-LC 1997 Music ISO/IEC MPEG Licensed Yes — default audio of online video
Vorbis 2002 Music Xiph.Org Free Niche — games, legacy web
HE-AAC 2003 Music ISO/IEC MPEG Licensed Yes — low-bitrate streaming
E-AC-3 (DD+) ~2000s Cinema Dolby Proprietary Yes — streaming surround, Atmos base
Opus 2012 Both IETF Free Yes — owns WebRTC and real-time
xHE-AAC (USAC) 2012 Both ISO/IEC MPEG Licensed Growing — Apple, Netflix mobile, Amazon
AC-4 2014/15 Cinema Dolby / ETSI Proprietary Yes — ATSC 3.0 NextGen TV
MPEG-H 3D Audio 2015 Immersive ISO/IEC MPEG Licensed Yes — ATSC 3.0, immersive
LC3 2020 Both Bluetooth SIG Licensed Yes — every Bluetooth LE earbud

Table 1. The major audio codecs in checklist order, with tribe, standards body, licensing posture, and 2026 relevance. Dates are first standardisation; see the References for the controlling document of each.

A common mistake: "newer codec means I should switch"

The most expensive misreading of this history is treating it as a ladder you must keep climbing. It is not. Each codec survives because it owns an ecosystem, and the ecosystem, not the compression maths, decides what you should ship.

AAC-LC is twenty-nine years old and is still the correct default for stored video, because every phone, TV, and browser decodes it in hardware. Opus is newer and more efficient, but shipping Opus audio in a movie file would break playback on devices that expect AAC — so you use Opus where it owns the ground (real-time, WebRTC) and AAC where it owns the ground (stored streaming). MP3 is patent-free and universally supported, yet it is rarely the right new choice because AAC and Opus beat it at every bitrate. The lesson of fifty years is not "always pick the newest codec" — it is "pick the codec that owns the job you are doing." The decision framework lives in How to choose an audio codec for your service in 2026, and the head-to-head numbers in The 2026 audio codec comparison table.

Where Fora Soft fits in

We meet this whole timeline in production. In our WebRTC conferencing, telemedicine, and e-learning work, Opus is the default — free, low-latency, and built for live voice — and we lean on its speech mode for clear talk and its forward-error-correction for lossy networks. In OTT and streaming projects we ship AAC-LC for universal playback and reach for E-AC-3 or AC-4 when a client needs surround or Atmos. On the edges we still handle the survivors: a SIP gateway that falls back to G.711, a broadcast feed that arrives as MP2, a customer whose earbuds run LC3 and change the latency budget. Knowing which codec owns which job — and which old codec is about to surface from some legacy corner — is part of scoping any audio pipeline correctly.

What to read next

Call to action

References

  1. ISO/IEC 11172-3:1993 — Information technology — Coding of moving pictures and associated audio for digital storage media at up to about 1,5 Mbit/s — Part 3: Audio. ISO/IEC (MPEG). Tier 1 (standard). Defines MPEG-1 Audio Layers I, II (MP2) and III (MP3); algorithms approved 1991, finalised 1992, published 1993. https://www.iso.org/standard/22412.html (accessed 2026-06-05).
  2. ISO/IEC 13818-7:1997 — Information technology — Generic coding of moving pictures and associated audio information — Part 7: Advanced Audio Coding (AAC). ISO/IEC (MPEG). Tier 1 (standard). First AAC standardisation as MPEG-2 Part 7. https://www.iso.org/standard/25040.html (accessed 2026-06-05).
  3. ISO/IEC 14496-3 — Information technology — Coding of audio-visual objects — Part 3: Audio. ISO/IEC (MPEG). Tier 1 (standard). The MPEG-4 home of the AAC family (AAC-LC, HE-AAC, HE-AAC v2) from the 1999 edition onward. https://www.iso.org/standard/76383.html (accessed 2026-06-05).
  4. ISO/IEC 23003-3 — Information technology — MPEG audio technologies — Part 3: Unified speech and audio coding (USAC / xHE-AAC). ISO/IEC (MPEG). Tier 1 (standard). USAC finalised early 2012; basis of xHE-AAC. https://www.iso.org/standard/85268.html (accessed 2026-06-05).
  5. IETF RFC 6716 (September 2012) — Definition of the Opus Audio Codec. Valin, Vos, Terriberry. IETF, status Proposed Standard, updated by RFC 8251. Tier 1 (standard). Opus combines SILK (linear prediction) and CELT (MDCT); scales 6–510 kbps. https://www.rfc-editor.org/info/rfc6716/ (accessed 2026-06-05).
  6. IETF RFC 9639 (December 2024) — Free Lossless Audio Codec (FLAC). IETF (CELLAR WG), Standards Track. Tier 1 (standard). Formal specification of FLAC, in use since 2000. https://www.rfc-editor.org/info/rfc9639/ (accessed 2026-06-05).
  7. ITU-T Recommendation G.711 (1972, in force) — Pulse code modulation (PCM) of voice frequencies. ITU. Tier 1 (standard). The 8 kHz / 8-bit / 64 kbps telephone codec; μ-law and A-law. https://www.itu.int/rec/T-REC-G.711 (accessed 2026-06-05).
  8. ITU-T Recommendation G.722 (1988) — 7 kHz audio-coding within 64 kbit/s. ITU. Tier 1 (standard). First wideband (16 kHz) voice codec; HD Voice. https://www.itu.int/rec/T-REC-G.722 (accessed 2026-06-05).
  9. ISO/IEC 23008-3:2015 — Information technology — High efficiency coding and media delivery in heterogeneous environments — Part 3: 3D audio (MPEG-H 3D Audio). ISO/IEC (MPEG). Tier 1 (standard). The ISO object-based immersive audio standard. https://www.iso.org/standard/63878.html (accessed 2026-06-05).
  10. ETSI TS 103 190 (2014/2015) — Digital Audio Compression (AC-4) Standard. ETSI. Tier 1 (standard). Dolby AC-4, approved for commercial use December 2014. https://www.etsi.org/deliver/etsi_ts/103100_103199/103190/ (accessed 2026-06-05).
  11. ETSI TS 102 366 — Digital Audio Compression (AC-3, Enhanced AC-3) Standard. ETSI. Tier 1 (standard). Normative text for AC-3 (Dolby Digital) and E-AC-3 (Dolby Digital Plus). https://www.etsi.org/deliver/etsi_ts/102300_102399/102366/ (accessed 2026-06-05).
  12. Bluetooth SIG — LE Audio and the LC3 codec. Bluetooth Special Interest Group. Tier 4 (vendor/standards body blog). LC3 adopted 15 September 2020 (Bluetooth 5.2); full LE Audio spec completed 12 July 2022; jointly developed by Fraunhofer IIS and Ericsson. https://www.bluetooth.com/blog/a-technical-overview-of-lc3/ (accessed 2026-06-05).
  13. Xiph.Org — "Opus audio codec is now RFC 6716" (2012) and Vorbis project pages. Xiph.Org Foundation. Tier 3 (first-party, spec authors). Opus release announcement and Vorbis history (bitstream frozen May 2000, 1.0 software July 2002, built as a patent-free MP3 alternative). https://xiph.org/press/2012/rfc-6716/ (accessed 2026-06-05).
  14. MP3 patent expiry (2017). Fraunhofer IIS terminated the MP3 licensing program in April 2017 as the core patents expired. Tier 4 (vendor/press corroboration). Used to date the licensing thaw, not as a standards source. https://www.iis.fraunhofer.de/en/ff/amm/consumer-electronics/mp3.html (accessed 2026-06-05).

Per §4.3.2 source hierarchy: codec dates and standard numbers above follow the controlling standards documents (ISO/IEC, IETF, ITU-T, ETSI, Bluetooth SIG). Where popular articles loosely date "MP3 = 1991", this article distinguishes the 1991 committee-draft approval of the MPEG-1 audio layers from the 1993 publication of ISO/IEC 11172-3, and dates MP3 the codec to the 1993 standard, per the ISO record.