Skip to main content ↑ ↓

Notice

Please note that most of the software linked on this forum is likely to be safe to use. If you are unsure, feel free to ask in the relevant topics, or send a private message to an administrator or moderator. To help curb the problems of false positives, or in the event that you do find actual malware, you can contribute through the article linked here.
Topic: Create a new Tagging Standard? (Read 16291 times) previous topic - next topic
0 Members and 1 Guest are viewing this topic.

Create a new Tagging Standard?

Reply #25 –
Strangely, I posted a question along these same ideas in a thread here just last week and got no real response.

As far as tagging standards go, you have MPEG-7, AAF (a partial implementation of MPEG-7), ASF tags (by MS), and ID3v2/1 (v2 is indeed a standard, just poorly implemented and not created by the MPEG group).

MPEG-7  has much to fine a resolution to be convenient for regular consumers.  License fees may also apply.  Also, I'm not sure that its been finalized yet.

AAF is less granule, but has something like 1600 tags.  (maybe more now that its been finalized)  License fees may also apply.

ASF tags are what is used in ASF/WMV to hold media information.  Information available here, here, here, and here.

ID3v2 is a pretty good standard with the one exception that more than one type of data can be held in a single field.  This means that two partial implementations can't read each other's data for the same field.  The specs are of course here.

For a synopsis of my conclusions about these formats, feel free to read here.

If anyone is interested, a compilation of almost all tags tags more major formats are available here.

Create a new Tagging Standard?

Reply #26 –
I checked out the XL file and it presents my opinion in a very complete way...
Why do we have so many formats and why some implement more fields than others to store tag info for the same purpose ?
Why do not need a complicated method to tag our files instead of a nicely defined and smart standard powerful yet easy to implement , with rigid standards yet flexible and user friendly.
So where do we go from here? stay like that or evolve?
Dimitris

Create a new Tagging Standard?

Reply #27 –
Quote
a nicely defined and smart standard powerful yet easy to implement , with rigid standards yet flexible and user friendly.

What an amazing sentence.  You must work in marketing. 

The problem is that there are always tradeoffs on either side.  By its nature, if you make something more rigid, it becomes less flexible.  For instance, in the Matroska system the decision was made to make the tags more on the rigid side to prevent people from goofing them up.  There are certainly points where it is very flexible, but it takes a long time to find a balance of where the right spot is.  Also, there will always be people that want it more on one side or the other.

For example, if you follow the link in the thread that I pointed to, you will see a discussion on redesigning the Matroska tags system.  Some coders find the current system to complex and want to simplify it.  The simplified version is still more complex than the Ogg comments version of 'Title=Whatever', but it is less complex than the current system.  It would also remove the ability to do nested attributes that the current system allowed.  I was really hoping someone from HA/fb2k/AA(AudiophilesAnonymous) would post some comments about it, but apparently no one was THAT interested.

Matroska IS an open standard that anyone can use.  The specs are being defined into an entire system, but anyone can use any part of it for their own purposes.  So if you want an open Tagging system, perhaps completing the Matroska tags is the answer.

Create a new Tagging Standard?

Reply #28 –
Quote
There might be a reference, but my original post was to extend the reference (Preferrence, CD Ident etc). Lets talk about a standard for album covers, Cue sheet implementation.

There's already a great standard for album/cover art. Stick all your audio files into an archive, add the cover art to the archive. Zero redundancy. It's what I see all the time downloading albums from P2Ps.

Really, adding cover art (which is an album level datum) to every single file on the album is ludicrous and idiotic.  Say you have 30 odd files, with 500kb of cover art total, that's 15MB of cover art over all those 30 files. How stupid is that?

If APEv2 isn't robust enough (and I fully agree that we need to specify a few more fields by standard, like ALBUM ARTIST), there are other solutions available, as Pamel listed. But there aren't many instances where APEv2 won't suffice.

My opinion: Keep cover art out of tags, that's not where they belong. There should be one instance of the cover art, at the album level.

It would be nice to see some nice, simple encapsulation file format that would let you store all  the media files in an album in a nice, orderly manner. The problem with ordinary archives are that they take a lot of time to change or update a file inside of them.

Create a new Tagging Standard?

Reply #29 –
Quote
It would be nice to see some nice, simple encapsulation file format that would let you store all  the media files in an album in a nice, orderly manner.

Matroska allows both chapters and tracks in a single file.  That means that you can have a single long audio stream with chapters to divide the songs.  Or, you could have several seperately encoded songs in different tracks.  (Even using different compression formats if you like.)  And, it supports attachments so you could attach some cover art and the CUE sheet to recreate the original CD later if you wanted.

Create a new Tagging Standard?

Reply #30 –
Since Matroska keeps popping up, I'd like to add my 2¢:

So does QuickTime

Create a new Tagging Standard?

Reply #31 –
Quote
Matroska allows both chapters and tracks in a single file.  That means that you can have a single long audio stream with chapters to divide the songs.  Or, you could have several seperately encoded songs in different tracks.  (Even using different compression formats if you like.)  And, it supports attachments so you could attach some cover art and the CUE sheet to recreate the original CD later if you wanted.

Well, then, go Matroska!

*waits for a foobar2000 Matroska implementation patiently*

Matroska devs: foobar2000 can likely utilize all the features of a matroska audio stream. I'd love to see a component made for Matroska reading. You'd be able to impress quite a wide variety of people using fb2k's back-end power.

Create a new Tagging Standard?

Reply #32 –
I havent seen it mentioned but there should be support for multiple artist fields. That way when A and B work togther you have:

Artist 1: A
Artist 2: B


NOT

Artist(s): A and B

So when you want to listent to something by A you can find even things that A worked on. Currently you would sorrt  by "A and B" because they are linked together and the dumb tags think they are one entity when they are 2 or more.

There should also be support for multiple preferences, that way when multiple ppl use the library they can have there set of preferences. Or when ppl are listening at the same time the average of all of their preferences could be used.

Create a new Tagging Standard?

Reply #33 –
@Canar: 
This has been discussed by the Matroska devs previously, that is pretty much how they all feel.  fb2k is an optimal environment to take advantage of these features.  Unfortunately none of the Matroska devs know anything about programming fb2k services.  If any fb2k devs could help out, the library is free, as is everything else about the format.  And there is always someone in the in the IRC channel to answer questions about the format.


@Eli:
This was already done in the original Matroska tags.  You could even specify an email, URL and address for each artist.  There is currently some serious discussion about reworking the tagging system in the Matroska.Devel Mailing List. But, keeping the ability to have things like multiple artists, each with their own information, is of the top importance.

Create a new Tagging Standard?

Reply #34 –
Vorbis comments & APEv2 tags support and even encourage the usage of the same field name multiple times, ARTIST is a good example.
Now the 'problem' to solve is that we need an easy way to have this information manipulated in a way that allows us to : Rename files based on ARTIST info,Filter/sort files by ARTIST,Easilly edit the tag.

About cover art inside files or links to cover art ( or any other meta info ) it really is up to the user to decide.
From what i have seen trough the years there is (I think) only one software that supports links to images for the APIC ID3v2 frame,all others support the embedding of the file, so in spite of the standard allowing links practically no one uses them. People will say that image data inside an audio file is wrong and they are right others might say that links are easily broken and they are also right. Both should be available in a tag format.
Dimitris

Create a new Tagging Standard?

Reply #35 –
Thinking realistically, as much as it is an uphill struggle just to get a Rating label standard across formats, pushing the Matroska standard is an even more uphill battle.

I like the idea of Multiple Artists, sticking to the existing multi genre standard from APEv2 (that no one uses?), multi-artists can be seperated by \r

All I need to do is to convince the Ogg and FLAC guys this is a good idea, also the Rio guys (to implement APE2) and a few other hardware people.

Create a new Tagging Standard?

Reply #36 –
Well maybe I am wrong but I think the matroska tag is more like movie oriented.
I think that audio and movie tags should remain separated and apart from each other.
Multi genre's are a thing I use a lot, it makes classification and retrieval of files a lot better.
I think a genre/sub-genre/style format is quite convenient.
Dimitris

Create a new Tagging Standard?

Reply #37 –
Quote
Thinking realistically, as much as it is an uphill struggle just to get a Rating label standard across formats, pushing the Matroska standard is an even more uphill battle.
I think you misunderstood Pamel (or maybe I misuntderstood you ), what he means (AFAIK) is that the Matroska tagging specs can be taken and used as you want with any other format, as a template to a new spec or whatever, because it's open.

Quote
Well maybe I am wrong but I think the matroska tag is more like movie oriented.
In fact they're an amalgamation of all known tags, so they have basically any tag defined elsewhere. Again, as the spec is open you can take only a part of it.

Create a new Tagging Standard?

Reply #38 –
For my $0.02 I think the problem here is that we are talking distributed tagging systems, in essence 1 person - 1 tagging system, which is perhaps why we have a dozen or so formats.
The problems I see is that
1) People don't tag - Probably the worst bunch because if that file is shared then any details about that track are lost
Reason: many don't know about tags, or don't care
Result: If you try and tag the file you could mis-tag it and share it again - then you end up with places like kazaa
2) People who do Partial tagging - Although better than non-taggers - it seems many people tag in a particular manner - ie only ID3v1 tags and hense sometimes important track information is lost, some people like putting artist - track title in the track title field if it is a Various Artists album - why should artist information be in the track title box? I understand that it is important information but that is one of the limits of current ID3 formats.
3) Hardware vendors are not going to support any tagging system until the community demonstarates that it is a standard that has been accepted for some time

The solution I think is in developping a FreeDB-Tag system, where along with a regulated format for tag information, where a database could be kept on each track and perhaps integrate that with one of the more developped tagging systems or create a new one. It would also be nice to integrate technology like musicbrainz uses to identify tracks.

Some ideas 

Create a new Tagging Standard?

Reply #39 –
Quote
Well maybe I am wrong but I think the matroska tag is more like movie oriented.
I think that audio and movie tags should remain separated and apart from each other.
Multi genre's are a thing I use a lot, it makes classification and retrieval of files a lot better.
I think a genre/sub-genre/style format is quite convenient.

Outlyer hit most of the points on this, but I thought I would just add a little more, so if you're content with what he said, then skip it.

First, if you look at the excel spreadsheet I linked to you will notice that there are hardly any movie specific tags.  This is because there just isn't much in the way of normal movie tagging systems out there.  You have the AVI extended set, and that is it.

Second, other than a handful of movie specific tags, they share all other tags.  There just isn't a reason to needlessly increase the number of tags everyone needs to support.

Third, same as the artists, having multiple genres is very important.

Fourth, the head developer of Matroska (Steve) is a DJ, and one of the main focuses for him was to make a very powerful audio container.  So if anything, many of the features are actually audio oriented.

Create a new Tagging Standard?

Reply #40 –
@ spoon :

sorry for kind of hi-jacking this thread, but this is exactly the discussion we were seeking recently, as you may have noticed we are in the process of redesigning our tagging system.

Is it ok to go on reporting about our ideas and progress on this matter here, or should we use another thread ? BTW, jcsston has now started working on a fb2k plugin, based on DEATH's MP4 reader class ...

Create a new Tagging Standard?

Reply #41 –
Sure here is fine.

If I can speak freely about the existing Matroska tags, I didn't like the use of non strings for storage of numbers.

Create a new Tagging Standard?

Reply #42 –
Here is an idea for mp3 only,

how about a smart ape2 tag? that is for compatibility and APE2 tag is written followed by a ID3v1 tag, then if a player does not support the APE2 it will atleast read the APE2 tag. The code would be updating say artist in ape2 will update artist (as best as can) in the id3v1.

Create a new Tagging Standard?

Reply #43 –
Quote
I havent seen it mentioned but there should be support for multiple artist fields. That way when A and B work togther you have:

Artist 1: A
Artist 2: B

NOT

Artist(s): A and B

So when you want to listent to something by A you can find even things that A worked on. Currently you would sorrt  by "A and B" because they are linked together and the dumb tags think they are one entity when they are 2 or more.

There should also be support for multiple preferences, that way when multiple ppl use the library they can have there set of preferences. Or when ppl are listening at the same time the average of all of their preferences could be used.

I agree with this whole heartedly...

Categorizing music is often a fine distinction and having the fields organized like XML nodes with "unboounded" in their schema would rock.

I could then place a music file in several categories and carry around that meta data.
"Droplets of Yes and No, in an ocean of Maybe"

Create a new Tagging Standard?

Reply #44 –
Quote
Here is an idea for mp3 only,

how about a smart ape2 tag? that is for compatibility and APE2 tag is written followed by a ID3v1 tag, then if a player does not support the APE2 it will atleast read the APE2 tag. The code would be updating say artist in ape2 will update artist (as best as can) in the id3v1.

Sorry that I have to say this but what you say is at least partially the way a LYRICS3 v2 tag behaves, so this is old news. As I said in previous posts please read carefully the tag standards that exist right now before trying to evolve/optimize anything otherwise we will be re discovering the wheel.
Again sorry for sounding like that but I feel that these things must be said.
Dimitris

Create a new Tagging Standard?

Reply #45 –
The first draft of the new Matroska tagging system has just been completed.  This new version supports nesting of attributes for any tag.  There is also a proposed XML authoring format.  Also, all tags are either UTF-8 or a binary dump of other data, so they should be readable no matter what writes them.

Working draft

Remember that this is a very rough draft so there are lots of little mistakes.  But it should be very representative of the final product.

[span style='font-size:8pt;line-height:100%']Edit:  Changed URL to point to Matroska website specs as CVS has now been updated.[/span]

Create a new Tagging Standard?

Reply #46 –
I did not see a LYRICS field, also it would be nice to include the ID3v2 APIC frame functionality up to some extend... comment is listed under image attributes is that ok ?, I would also like to see more 'classification' fields. A small note about DATES .. every tagging standard lists date fields but every software on the planet writes year values in them .. just because it is a lot more practical! so this is something I might wish for as well. explicit YEAR fields instead of DATES.

Nested tags are nice but they are complex and might scare developers away! most tagging software out there supports only basics fields from various formats even if those formats are simply flat data. I would be interested to listen to some idea of theoretical implementation in software of an XML nested tag field and how the user might interact with it in batch modes or use it in queries and cataloguing software.

Finally , where will this structure be located inside the physical file?
Dimitris

Create a new Tagging Standard?

Reply #47 –
Quote
I did not see a LYRICS field
This was actually already in there, but commented out for some reason.  It shows now.  Note that this is only UNSYNCHRONISEDTEXT as synchronized text in Matroska would be contained in a track.
Quote
also it would be nice to include the ID3v2 APIC frame functionality up to some extend... comment is listed under image attributes is that ok ?, I would also like to see more 'classification' fields.
In Matroska, any pictures are attachments.  The part for attachments in the Tags isn't actually shown yet, but there is not much change.  I would refer you to the main attachment specs for Matroska.  Beyond that the current tags allow you to Tag the image with all sorts of attributes.
Quote
A small note about DATES .. every tagging standard lists date fields but every software on the planet writes year values in them .. just because it is a lot more practical! so this is something I might wish for as well. explicit YEAR fields instead of DATES.
The date fields will be in the format "YYYY-MM-DD HH:MM:SS.MSS" where the timezone is GMT and the space between DD and HH is 0x20.  If you want to write the year, then only write "YYYY". 
Quote
Nested tags are nice but they are complex and might scare developers away! most tagging software out there supports only basics fields from various formats even if those formats are simply flat data.
Nested tags are used to produce detailed data, and so that in the future we might be able to use them to implement AAF support.  We all expect that most applications will only deal with the first level of tags, skipping all of the rest.
Quote
I would be interested to listen to some idea of theoretical implementation in software of an XML nested tag field and how the user might interact with it in batch modes or use it in queries and cataloguing software.
I would also be interested in peoples ideas. 

For displaying two levels of tags, you could just have the different parts of the second level right after the first.  For example:
[ArtistName]->[Email][Phone] ; [BandName]->[URL]

Displaying anything more than two layers though would pretty much require the use of a type of tree view.
Quote
Finally , where will this structure be located inside the physical file?
Matroska has no real requirements about the location of items within a file.  Other than the SeekHead (gives locations of other items) which should always be the first item, other items can be spread out anywhere in the file.  In most cases, the Tags will likely be located at the end of the file so that it is easier to edit them. (room to shrink and grow)  But if you wanted to, you could put them at the beginning (right after the SeekHead). 

There is no worry about taking a long time to load the Tags though because the SeekHead will tell an application right where the tags are so that you can just jump strait to them.

Create a new Tagging Standard?

Reply #48 –
The new Matroska tagging system has been implemented in an fb2k Matroska input plugin now.  One thing about it though is that the fb2k framework doesn't allow for nested tags so you can only add on the base level.  Also, there is not a way to use the system to tag attachments.  And finally, because this is just a container plugin, only MP3, AAC, and FLAC are supported so far with packet_decoder plugins.  (You may notice some similarity to APEv2 design)

If you would like to see an example of this, then HERE is an example file.  Its a full album in one file with the audio encoded using a single audio stream encoded to MP3, and Chapters to define song boundries.  If you have the Matroska Shell Extension installed you will see the CD cover as the thumbnail. (Distribution permissions obtained)

What we really need from the HA community right now is help with the tag names.  We are currently using all caps, with an underscore for a space.  But for compatability with existing systems, we will be adopting terms such as "TRACKNUMBER".

So please, look at the Matroska Tags Specs and tell us if there is a different standard naming convention already being used for a specific tag.

Create a new Tagging Standard?

Reply #49 –
Quote
So please, look at the Matroska Tags Specs and tell us if there is a different standard naming convention already being used for a specific tag.

LOL .... no feedback, as usual. And afterwards they all come whining because they dont like what we did  .....