Print

Please login or register.
1 Hour 1 Day 1 Week 1 Month Forever
Login with username, password and session length

[ADDiCT]

During the last few weeks, i was fiddling a lot with various aspects of metadata in MC's library and media files - ID3 tags in general, Musicbrainz tags, Last.FM tags, etc. . This led me to some ideas, which i'd like to share and discuss here.

Let's define the term "metadata". Metadata is "data about data". This means that all information in the MC database, as well as the various types of tags in media files (JPEG header data, ID3 tags, APE tags, etc.), are metadata by definition.

In my opinion, metadata handling is one of the strongest features in MC. I'd even go so far as to say it's the most important aspect of MC's functionality. For a long time, MC was one of very few "Media Players" with a database-like backend, which allows for efficient management of media files and metadata, and nifty ways of selecting the media to play (Smartlists, etc.). It's nearly impossible to manage large collections of media without this kind of "Media Management Software". If you look at modern media players, you'll notice that many of them are now offering some kind of media management/metadata management functionality. Good examples would be WinAmp (which used to be a simple media player without any management functionality at all), foobar2k, or iTunes, of course.

So how do you get the metadata in your media database, or your tags? For a long time, there was only one solution: type it in manually. This is a tedious process, and very error-prone, of course. In the last few years, however, new ways of importing metadata were made available. I think it started with FreeDB-aware media players, and continued with software like MP3Tag (which is able to import metadata from various sources) or MusicBrainz Picard. In my opinion, Picard is "state of the art" when it comes to getting metadata into music files. And then there's Last.FM with it's "scrobbling", which relies heavily on correct metadata.

So, in a nutshell: media management and metadata management is getting more and more important - not only for "hardcore media collectors" with large libraries, but also for "casual media users" using tools like Last.FM.

If you look at ways of getting metadata into MC, you're basically looking at the "do it manually"-approach. This is quite interesting, considering the fact that MC is positioned as a "all-in-one"-solution. The truth is, however, that you'll need to use various tools, and quite some thinking and manual work to efficiently manage metadata in MC.

As i'm using MC for audio mostly, i can't say much about video or image files, but i'm quite sure the situation there is the same. You'll have to input metadata manually, or with complicated workarounds like MPL import.

I think MC needs more ways of importing and managing metadata. MPL import is a step in the right direction, but the functionality is not very user-friendly or "noob-proof", and has some other problems.

I have developed some small tools for my own usage to do a lot of cool stuff with my media files (music files, mainly) and MC's data/database. This has opened my eyes for the possibilites, which are amazing, in my opinion. To whet your appetite a little, here's some of the stuff that's already working:
- Album Rating (for "completely rated" albums)
- Import of MusicBrainz Metadata into MC (either into standard tags like "Artist" or "Album", or into dedicated library fields for manual or automatical comparison and correction)
- Import of Last.FM data into MC (Similar Artists, Last.FM Tags, artist data, etc.)
- Various "metadata checking" tools (check for common spelling errors or tracks with identical names, comparison of two MC libraries to identify changes, etc.)
- Various statistical functions (average rating/artist, top artists, top albums, etc.)

All this stuff enhances my library and user experience greatly. For example, with the Last.FM "Similar Artists" data, i simply type in an artist name in a certain way, and get a list of all similar artists in my MC library. With some more keystrokes, i can look at the list of similar artists that are _not_ in my database, and browse the web for more info on them.

So what could possibly improve the user experience in terms of metadata management? I have some tangible ideas for that, but describing these will take a little time. I'd like to have some feedback first. (;

[AoXoMoXoA]

I would love to see MC's tagging become more automated, especially the ability to lookup and import tag data from outside sources as seeking this info manually is extremely cumbersome and tedious.
Hopefully this will become part of MC's functionality in the near future, or perhaps your tools may at some point become plugins.

[zxsix]

The similar artist functionality I would use quite a bit.
I also wish we could fetch coverart and track names from amazon.com and musicbrainz in addition to yadb.
Poll all three and show me the results in a window similar to how the get coverart from internet feature works now.
The display cover art feature works great now, you'd just be adding 2 new sources to it.
Add to that the track names, listing what each source showed.
Let us click checkbox next to one coverart and one track list and update the info.

[zirum]

Your are touching some very good ideas that MC hopefully will focus more on in the future. I find it difficult to elaborate on exactly where I would like this to go, but more interaction and tighter communication with other web-services would have been really cool. I love the information last.fm, rateyourmusic.com, pandora etc provide, but just do not have the energy to start looking into how to use the plugins etc. to make it work (if possible) with MC.

The other aspect is the metadata replication. I would love a data exploration db, instead of the continious low level typing. Something like a song has a set of artists, which was at that point this band, which has provided these albums an so on. A way to link the information, instead of replicating so extensively. Was there not a vote on this for MC13 features?

I also would like to be able to set custom rules on import, like if some validators match, fetch subfolder name and put at some column and etc. That is a diffrent subject...

[leezer3]

In a nutshell, I believe the whole problem with this sort of approach at the moment is that MC uses a simple 'flat' database, not a one with a more relational structure.
I've been advocating adding a relational layer to the DB for a while now, but this is a massive task, albeit one with many benifits.
At the moment, I'm mostly using smartlists/ expressions and some input from the scripting plugin to automate some of my metadata collection, but an automatic approach would be nice in many cases. Small things like the ability to link one field to the content of another would make my life and that of many others a whole lot easier.

Data exploration is quite possible, just requires a little tuning to your viewschemes and the way you populate them as such. The biggest way I'd change this though, would be a way to shift between viewschemes. A short example for you:

1. I'm browsing my audiobooks list with a basic Genre ==> Author ==> Book viewscheme.
2. I find an author, but I'd like to look at the complete works of this author in the author based variant of this viewscheme- Author ==> Genre ==> Book
3. At present I need to manually open this viewscheme, and browse to the author. I can nest it in the viewscheme tree, but this is messy, and doesn't open the viewscheme to the author I was looking at

On the other hand however, the more complex items of metadata are one of those things which becomes highly individualised This means that its far harder to implement something like YADB for this, simply because different people look for different things in it (For example, how much if at all should be put in a band's biography/ history?)
No real way around this that I can see, but I'm sure other people will have some ideas.
I wouldn't really use a feature of this nature, but I'm sure many would

Cheers

-Leezer-

[Listener]

> The biggest way I'd change this though, would be a way to shift between viewschemes.
> A short example for you:

> 1. I'm browsing my audiobooks list with a basic Genre ==> Author ==> Book viewscheme.
> 2. I find an author, but I'd like to look at the complete works of this author in the author based variant of this
>     viewscheme- Author ==> Genre ==> Book
> 3. At present I need to manually open this viewscheme, and browse to the author. I can nest it in the viewscheme tree,
> but this is messy, and doesn't open the viewscheme to the author I was looking at

If you check "filter in both directions", you can use a single view scheme to browse in either order (with panes.)  I don't don't know how that feature works if you browse by images.

Bill

[leezer3]

> The biggest way I'd change this though, would be a way to shift between viewschemes.
> A short example for you:

> 1. I'm browsing my audiobooks list with a basic Genre ==> Author ==> Book viewscheme.
> 2. I find an author, but I'd like to look at the complete works of this author in the author based variant of this
>     viewscheme- Author ==> Genre ==> Book
> 3. At present I need to manually open this viewscheme, and browse to the author. I can nest it in the viewscheme tree,
> but this is messy, and doesn't open the viewscheme to the author I was looking at

If you check "filter in both directions", you can use a single view scheme to browse in either order (with panes.)  I don't don't know how that feature works if you browse by images.

Bill

Been there, done that- IMHO it confuses things completely
Several issues with that approach-
1. When there are 4 or 5 panes in a view (Certain of my video views are a prime example here), it becomes more and more irritating.
2. There is no easy way to reset the view to its starting point without navigating away from it and returning if multiple filtering selections have been made.
3. The panes are then in the 'wrong' order if I start filtering. Little thing I know, but I work from left to right, and again I just find it that bit easier to shift viewscheme.
4. This doesn't work at all with viewschemes with disparate elements in them If I'm moving from a films to a TV viewscheme or vice-versa, the panes in each of these are completely different.

Cheers

-Leezer-

-Leezer-

[Listener]

Suit yourself.  I provided an easy way to address the original problem you stated using MC right now.

> 1. When there are 4 or 5 panes in a view (Certain of my video views are a prime example here), it becomes more and more irritating.

You didn't say why it becomes more irritating.  I use 5 panes in a music view all the time.  Changing from browsing by sub-genre-->composer-->work-->Artist-->Version to browsing by Artist-->Composer-->Work is much easier for me than than changing views.

> 2. There is no easy way to reset the view to its starting point without navigating away from it and returning
> if multiple filtering selections have been made.

If you mean selections in more than one browser pane, click on "Reset Selection" or click on "All" at the top of a pane's list.  Seems easy and very understandable to me.  Maybe things have changed after version 13.0.50.

> 3. The panes are then in the 'wrong' order if I start filtering. Little thing I know, but I work from left to right,
> and again I just find it that bit easier to shift viewscheme.

Asking JRiver to do some development work because you prefer to work from left to right seems misguided.

> 4. This doesn't work at all with viewschemes with disparate elements in them If I'm moving from a films to a TV
> viewscheme or vice-versa, the panes in each of these are completely different.

This seems completely unrelated to the original point.  Of course, you can make different view schemes to suit different purposes.

Bill

[leezer3]

> 4. This doesn't work at all with viewschemes with disparate elements in them Smiley If I'm moving from a films to a TV
> viewscheme or vice-versa, the panes in each of these are completely different.

This seems completely unrelated to the original point.  Of course, you can make different view schemes to suit different purposes.

Bill

Totally related 
Take the example of Star Trek. This has various films & several TV-Series within it. If I find one of the films from my films viewscheme, then it would then be nice to be able to simply shift to the related TV series in the TV viewscheme, rather than having to manually shift viewscheme.

> 2. There is no easy way to reset the view to its starting point without navigating away from it and returning
> if multiple filtering selections have been made.

If you mean selections in more than one browser pane, click on "Reset Selection" or click on "All" at the top of a pane's list.  Seems easy and very understandable to me.  Maybe things have changed after version 13.0.50.

Fair play, hadn't noticed that one I'd move it to a clearer position (The tab header IMHO is entirely the wrong place for something of this nature, needs to be in the panes structure itself), but otherwise the functionality is there.


The initial viewscheme was only a simple example to try and illustrate the problem that I percieve as being there.
I'm not asking anyone to do development work off the bat of a single thread- The whole point of this thread was discussion of alternate ideas, and I'm just presenting things as I see them!

Cheers

-Leezer-

[darichman]

Excellent post. It was the "database" behind the "player" which drew me to MC in the first place - it was truly the first program, that I'm aware of anyway, that did this properly. I agree that metadata is one of the most important parts of any digital media solution. At the most basic level, it really comes down to "I need to know how to find that song (or movie etc)" and "What song am I listening to". As Addict pointed out though, it's become a lot more than that as well. People want to know more about what they're listening to / watching / looking at. While "Links" and internet searches can certainly find this information, it's not as easy as an automatic lookup and storage in the file's metadata IMHO.

As far as its implementation in the program goes... there are three important aspects. Some have been touched on above.

1. Getting the Metadata
At some point, SOMEONE has to do a manual submission. That's unavoidable. Lookup services like YADB, CDDB, gracenote get us the basic info. As YADB has grown (and expanded to include movies) I believe MC is doing okay here. As Addict pointed out though, there are a lot of other (untapped?) sources of info out there with some really cool stuff. A lot of users, myself included, would really love to have this stored in the database. It keeps it local, and it affords you some element of control over the data displayed (let's face it, "links" and web searches can be hit and miss). Personally, I'd be interested in much of what Addict said he had to offer... I'd also like to see more "plugins" to hook into online sources - I like YADB (and use it regularly) but it's really only useful for basic info (name, album, artist, genre, year)

2. Storing the Metadata
At the user level this really comes down to choice - do I want to store everything in embedded tags, or just in the database? Do I want to keep MPL backups? If I, personally, have any request here, it would be some ability to store information in "sidecar" files, next to the actual file - for files which don't support embedded metadata (or which MC can't write tags to).

In terms of the database itself,  Like Leezer, I'm a big supporter of a relational approach to storing data for all the reasons he mentioned above (and a few more!). It has been requested in the past if you do a search. My name might come up once or twice

3. Making use of the Metadata
Okay, so we have all this awesome info now. But what do we do with it? Navigation is a big factor, and perhaps the most obvious one - we need to know a little bit of information about a file if we ever hope to find it in a search or by drilling down through lists.

But one area I think MC is really lacking in here, is display of metadata. I have a movie. I have details about the director, year, a synopsis. I have a review. How do I display it? Much the same can be said for albums as well. The Tag AW is really meant for, well... tagging - not a visually friendly representation of the data. If I'm in theatre view, I have less luck still. As some recent posts have described, users often feel the need to learn a little bit more about that thumbnail before clicking the play button (or indeed, once they discover they really like that song and would love to know more about it).

The new "Links" feature covers this to an extent. This is, however, not the same as metadata - it launches a browser, removing the user from their current view Don't get me wrong, I love this feature and use it a lot - it does what it does well - but I think that if the user has managed to carefully build up a database with useful metadata, surely there needs to be a way to display it?

If MC could
 - get my main music info from allmusic.com (AMG), with other stats and interesting tidbits from MusicBrainz and LastFM
 - get my movie and TV info from allmovie.com (AMG) and IMDb
 - organise it all in a way that People (Artists, Composers, Performers, Actors, Directors, Producers), Albums and Movies/Shows were navigable in a relational way
 - display relevant metadata in a meaningful way, both in standard view and theatre view

... I, for one, would be one very happy media-hoarder. And I think you'd really have someone that no one else does.

[JimH]

3. Making use of the Metadata
But one area I think MC is really lacking in here, is display of metadata. I have a movie. I have details about the director, year, a synopsis. I have a review. How do I display it?

We expect to do more in this area in the next few months.

[darichman]

We expect to do more in this area in the next few months.

Can't wait to see what you come up with! And let us know if you'd like some suggestions

[flac.rules]

We expect to do more in this area in the next few months.

Very nice, I am looking forward to it. I really like how the the devolopers listen to the users when it comes to MC

[Tiger100]

Very nice, I am looking forward to it. I really like how the the devolopers listen to the users when it comes to MC

I agree - this would be a great enhancement.  Meedio has been doing this for a while now.  It would be cool to browse movie covers with cover flow (i.e. iTunes).  When a movie is selected the screen in Theater View would jump to a screen that showed movie synopsis, actors, rating, .....  The movie information would have to be able to grab information from IMBD or Amazon.

Regards,

David

[hit_ny]

We expect to do more in this area in the next few months.

Hope this will be more than a just a way of displaying video specific metadata 

ADDiCT is alluding to increasing options for data entry.

So how do you get the metadata in your media database, or your tags? For a long time, there was only one solution: type it in manually. This is a tedious process, and very error-prone, of course. In the last few years, however, new ways of importing metadata were made available. I think it started with FreeDB-aware media players, and continued with software like MP3Tag (which is able to import metadata from various sources) or MusicBrainz Picard. In my opinion, Picard is "state of the art" when it comes to getting metadata into music files. And then there's Last.FM with it's "scrobbling", which relies heavily on correct metadata.

This is one area where competitors beat out MC easily i would say. But what happens after is they can't catch up with what MC can do ?

So if you want the best then you need an extra app (or custom plugin) in the workflow to do the job.

I've done most of my entry manually, its a long tedious process, but in the end i'd like to think its more accurate than automating it off other sources. The downside of course is if you want to add more metadata later, then you have to do it all over again at which point most would not have the courage to continue past the basic stage. So you are left with what i think is the most rudimentary of metadata, basic tags artist, album, genre, year and never proceed further. Maybe thats enough for some but in a way it limits the possibilites of how you could view your media had you been given a chance to get more info faster.

For example, with the Last.FM "Similar Artists" data, i simply type in an artist name in a certain way, and get a list of all similar artists in my MC library. With some more keystrokes, i can look at the list of similar artists that are _not_ in my database, and browse the web for more info on them.

'Similar Artists' would be very helpful in creating a map of what to play next and may make random play or radio that much better.

I've been advocating adding a relational layer to the DB for a while now, but this is a massive task, albeit one with many benifits.

Cpl of  limitations with MC spring to mind, the inability to automate
-  comparison of fields of one file with that of another.
-  operations on a user-defined group as everything is only file level.

Why go relational for these or could JRiver figure out a way with the current system ?

The biggest advantage of a flat file system is speed, over the years they manage to squeeze out more & more 

[newsposter]

A few months ago I pointed JR to a commercial embedded RDBMS that could be licensed and distributed for very small $$.  Very Small.