Furthering the previous article on music metadata, today I’d like to describe my recent observations about ISRCs.
ISRC is an acronym for the “International Standard Recording Code”.
It’s a 12-character, alphanumeric code that looks something like USS1Z9800123 (hyphens are often added for readability), and it aims to uniquely identify a sound recording — irrespective whether it’s a song, spoken word, or a music video.
[It] helps to avoid ambiguity among recordings and simplifies the management of rights when recordings are used across different formats, distribution channels or products. The ISRC for a recording remains a fixed point of reference when the recording is used across different services, across borders, or under different licensing deals. — IFPI
An ISRC does not identify compositions/musical works, music products or performers; that’s the function of the ISWC: The International Standard Musical Work Code.
According to the official handbook, a song’s album version will have a different ISRC to, for instance, a radio edit or a dance remix. By extension, each live performance (if recorded for “commercial exploitation”) must be assigned its own ISRC, as would a re-recording or certain remasters. It is also largely media-agnostic: the codes don’t differentiate between vinyl, cassette, CD or download (there are caveats such as with format-specific mono/stereo/multi-channel mixes).
It’s a rather dry topic overall but this 2017 video by DDEX makes it more palatable.
That being said, knowing the ISRC and understanding its implementation can be useful in identifying and differentiating between versions of a song recording. This was important to me last year when I set out to “archive” a batch of CDs (mostly singles and promos).
The tool of choice was Exact Audio Copy (EAC) due to the robust and accurate rips it produces. EAC can also generate a so-called “cue sheet” as well as a log file detailing the success rate and a checksum. These plain text files contain information such as the starting sectors of each track, its exact duration (down to frames), ISRCs (if present), and CD-Text (if present).
If you’ve ever noticed track title and artist information being shown on (most commonly) car CD players — that’s CD-Text.
It’s embedded in the subchannels of the compact disc itself (though not always well-formed) and is totally independent of ISRCs — they’re a different kettle of fish entirely: Originally devised by Philips, the ISO 3901 standard that covers them was ratified in 1986 (or 1989, depending on sources) and adopted by the IFPI who recommended that the recording industry implement it as of January 1992. Some authoritative sources even describe ISRCs as a form of “watermarking” — which is a load of bollocks.
Having previously determined that the 1993 disc shown below has/lists the wrong track/version in position 6 and because CD-singles typically contain multiple versions/remixes of a song, I had high expectations that my current batch of discs would be an interesting selection of samples to see what would turn up.
Well, so much for that! It turns out that this CD has no embedded ISRCs at all.
In fact, of the 30 or so discs I processed, barely a quarter had ISRC numbers encoded — despite the majority having been released during the nineties and beyond. A search for them on the international ISRC database yielded some rather hit-and-miss results.
Clearly not everyone is adhering to the IFPI’s recommendations — nor have I found a particular pattern in the culprits by way of label, period, manufacturer, or market territory. I checked additional CDs and found that the albums as recent as 2020 and 2021 I own contain no ISRCs whatsoever while others include only an MCN and/or CD-Text. It’s rather scattershot.
On a more positive note, I’ve not come across any duplicates yet. Yes, they exist.
Then there was this German CD:
The release year of 1988 can be regarded as 99.9% accurate considering the “Made in W. Germany” and “Manufactured by PDO Hanover, West Germany” print as well as styling/layout for the period and other factors. As would be expected, it has no SID codes (which were introduced after 1992). It even feels like a PDO/Polydor disc to the touch.
EAC detected the MCN as “0042288758624” (which matches the product’s EAN) as well as four ISRCs:
Ahem… but a CD from 1988 isn’t supposed to have embedded ISRCs. Or can it?
According to various forum discussions, drives occasionally misread ISRC and CD-Text information, so the test was repeated via ImgBurn as well as icedax on a totally different Linux box/drive. Another owner of the CD, too, confirmed that his values are identical.
So now that we know with a high degree of certainty that the ISRCs are extracted correctly from the disc’s subchannels, they are still not the correct values: neither matches with those found at the IFPI’s database. The current standards (last revised in 2019) do, however, allow the opportunity to update or correct previous errors (which makes sense).
This leads me to declare my German Visage single from 1988 to be one of the earliest such CDs out there, with embedded ISRCs of an early or prototype nature.
I’m also under the impression that for all the hype, formalities and regulations it seems that ISRCs weren’t taken seriously by the music industry until streaming came along.
Someday I’ll revise my tests with a larger selection of discs and arsenal of tools.
All scans by hmvhDOTnet.