Looking at mp3HD, the “Lossless” MP3 Format Nobody Ever Used.
I came across the Wikipedia article for mp3HD again and tried to clean it up somewhat.
In doing it, I actually looked at what a horrible format this was. As another harebrained scheme to “extend” MP3 and keep extracting royalties on it somehow, Thomson (Technicolor) (Now bankrupt.) and Fraunhofer Society (amusingly, my spell correct wanted to call them the Fraudster Society) collaborated to create a “lossless MP3 file” format about a decade after FLAC and WavPack already existed.
FLAC is proposed as an IETF standard as of 2019, but whether it becomes one or not, even Microsoft and Apple support them, which means it’s not only mainstream, but it supplanted their attempts at a proprietary lossless audio codec too. (Windows Media Audio Lossless and Apple Lossless).
In the end, Apple gave up and made Apple Lossless open source, after it had been reverse engineered anyway. Apple Lossless takes 400% more CPU time to decode than FLAC, compresses the files less, and has no official way to do error checking.
Although I suppose you could hack it in by running md5sum on each file, then adding it as a comment on the tags. It still wouldn’t be as good as FLAC or Wavpack’s because you couldn’t just ask the playback software to check and compare, and you’d need to store two values. One for the source file and one for the ALAC file (to make sure you could verify the source if you were to unpack it later).
Windows Media Audio Lossless has even more problems. I’ve actually only encountered that one once and ended up reading the data out into WavPack with the help of FFMpeg. I did a checksum verification on both ends and they matched. Then I never looked at WMAL again.
But mp3HD was a terrible codec. I’ve never actually used it, but I have read the specification. I think I played around with the encoder once to see if it could easily make standard MP3 files like the “MP3 Surround” encoder could, but they took that feature out.
The marketing tagline was “It’s time to preserve your music forever.”, and apparently “forever” was the two years it took Thomson/Technicolor and their Patent Troll Pals at FhG to give up on the format, forever.
mp3HD took advantage of the fact that you “can” shove up to 256 MB of arbitrary data into an ID3v2 tag, but just because you can doesn’t mean you should. They also claimed that the part of the file that was a lossy MP3 would play on anything that supported regular MP3 files. Which was a lie.
Why was shoving arbitrary data into an ID3 file a bad idea?
Well, tagging software.
If the user edited the actual comments in the tag, or the tag itself, the lossless data could become corrupt or lost entirely since MP3 with a tag is essentially two files in one.
The tag piggybacks the end or the front of the file (depending on the version) in an otherwise empty frame or series of frames.
So the mp3HD codec simply packed non-MP3 data into empty MP3 frames of an arbitrary length misusing a possible field in the ID3 tag (not to exceed the 256 MB tag size limit).
The actual MP3 part of the file was now just a standard MP3 using whatever encoding settings the person who made the “mp3HD” file selected.
So there was a huge potential for data loss. Edit the wrong field or accidentally rewrite the tag, bye bye lossless data correction.
Further, since this is undefined behavior by the ID3 specification (the closest thing to data you are supposed to put into one, officially, is album art), player software is free to interpret the data in the tag pretty much however it wants to. It’s not standard MP3 data, so it won’t play audio, but what it did do, in mplayer at least, was play the lossy MP3 followed by an additional 40 minutes of silence, according to a person on Hydrogen Audio.
The 256 MB limit in the ID3 tag means that your limit for the file is roughly 38 minutes in CD quality (16/44.1), which immediately means no “one huge file with .cue” like you can do with WavPack and FLAC. It also was limited to 16-bit / 48 kHz source, like MP3 was, so no native ability to directly deal with “High Res” sources. Also, if you used 16/48 then your time limit per file would probably drop to half an hour.
(Coincidentally, 38 minutes is the amount of time a Stargate can be connected without a ZPM plugged in, or a black hole on the other side, or the overload weapon used by Anubis, or a planet full of Naquadah melting down around it.)
So given the potential for data loss (due to the abuse of the ID3 tag), the larger size than FLAC or WavPack, the inability to use Hybrid Lossless (like WavPack’s lossy .wv with a lossless correction .wvc in the same folder) meant that you couldn’t break them apart and just put the lossy section on your portable device, there’s already no point in even trying to use something like this.
The fact that most devices only played MP3 in standard quality meant you’d waste at least three times the storage space vs. the quality of playback you received.
When you could have just put a FLAC file on the device and listened to it in lossless in exchange for all the space it uses.
That’s also if the thing even worked at all and didn’t attempt to read the non-MP3 data in the tag and play 40 minutes of static per file between every track.
Looking at this format, it’s hard to even think how they thought this would work or what tortured mind this even came out of. But we can be glad that it didn’t succeed.
As usual, proprietary software companies and patent trolls like to come to the party a decade late and re-implement something that already exists, badly.
My father used to work for RCA as an electrical engineer.
After GE purchased the company, it really went to Hell. No longer were they an innovative company that at least stood a chance against the Japanese.
GE sold the RCA consumer electronics division to the French (Thomson/Technicolor) who did not take good care of it. They ceded a decade of potential innovation (the 90s) to the Japanese, outsourced product manufacturing to Mexico, hired my dad back as a contractor who lived in Mexico at more than his previous salary had been, and then finally fired the Mexicans and sold the “brands” to the Chinese.
After this, they became a patent troll that was living large on MP3 royalties, and after the patents expired in the US they were sunk, and declared bankruptcy in France, followed by Chapter 15 Bankruptcy in the United States. I think all that’s left of them now is some motion picture stuff.
It’s just simply unbelievable how incompetent the French were with RCA. They destroyed an American icon and managed to blow up the French company a little later on.
Going back to MP3, the standard didn’t actually specify DRM.
Actually, it didn’t even specify a tagging format, so the only official use of the ID3 tag from Thomson/Technicolor and FhG was to violate the specification with mp3HD.
(You can also use APEv2 tags in an MP3.)
There were “potential helpers” for some DRM to be added to MP3s later, like the Copyright Bit (which pirated ones always said “No”, of course) and the people who did the ID3 specification left a comment field to indicate what DRM scheme was in use.
But by the time anyone may have wanted to add this, Microsoft and Apple already had their preferred formats anyway and they came with DRM. Apple extended MPEG-4 Audio (AAC) with a digital restrictions scheme called “FairPlay” and Microsoft had one for Windows Media Audio (WMA) called “Janus”.
Since I despise DRM, I always called them FoulPlay and Anus DRM when I was talking about them. In my last post, I mentioned my reaction to the iTunes Store in 2004, and it was to delete it immediately and refuse to ever touch it again.
Apparently, “FairPlay” means you allegedly bought something but when you try to play it in your preferred software it doesn’t work, then you find out you have to buy an iPod and “manage your licenses” with iTunes, and that’s totally not something I was ever interested in doing.
The MP3 format, for all its many technical and legal flaws, nobody ever bothered to restrict it like this.
You can still purchase them at Amazon without DRM and there’s none of this “FoulPlay and Anus” stuff controlling what you do and eventually taking the file back from you without a refund.
But Amazon is not your pal. Their Kindle Store works exactly like “FoulPlay and Anus”, and the Free Software Foundation took to calling it the Amazon “Swindle”.
BaronHK's Rants 