How to start or re-visit tagging of a large library

[DavidL]

I've started a new thread as the contents may be of wider interest than where this started (processing power needed for BubbleUPnP):

Serge said

"Hi David, I have a question for you, off this topic. So either we start a new thread on Music Tagging in this forum, or we continue via PM if any interest.

My library is roughly the same size as yours. 4380 albums; 78k+ tracks. 240 are Dsf, the remaining in Flac. Various origins: my CD ripped, my SACD ripped, Qobuz, HDtracks, etc.

Now, my tagging is not perfect. Qobuz does propagate the tagging from the Label company, which is never very good. Some of my CDs are not well taggued. Because my tagging is not perfect, and I must apologies to Simon, I am not using all Simon's beautiful features, shame on me. Therefore I have decided that my structure would be purely 100% based on my file structure. So I have 11 top lever folders (Test - Renaissance - Baroque - Classique - XXth Century - Opera - Jazz - Pop/Rock/Folk/Electro - World Music - Chanson Française - Movie music; plus a very complex sub structure different according to the top level folder.

Of course it has drawbacks : I can't do a search easily, some parts are ambiguous (as an example, I have World Music - Oud with one sub folder for any Oud musician but some albums are really jazz. Or I have Renaissance - AliaVox then all Jordi Savall albums except the one for Bach, etc...)

Now my question : how do you manage yourself the tagging ? Have you been able, at that level to have a perfect tagging ? Have you followed some rules, such as for classical, the album tag should contain (or not) the composer. Are all your classical composer clean i.e. first name last name. Did you put all the performers in opera. Etc...

Sometimes I fear that the work of re-tagging my library would be enormous; specially that I don't allow my Syno files to be visible on the network (only upload via SFTP)

Thx for the answers !

cheers
serge"

I guess most of us start off tagging by using what we inherit (e.g. metadata from CD originator), or set up a simple folder structure. The first point to make is that it is worthwhile spending a good amount of time researching whats available before making a particular file structure / metadata choice. Definitely so if you have just started your collection.

Like Serge I was well into collecting music (on LPs and CDs) before I even thought about this issue. My first encounter with a computer-based music library was (like many others of a certain age I'm sure) via iTunes. Today many feel that this is not an ideal basis from which to start, particularly as an iTunes user one is at the mercy of Apple's whims. However I've found that adopting just the library-curation functionality of the iTunes software I can do all I need whilst benefitting from the many hundreds of man years of software development put into it.

To answer Serge's questions I use the standard (Apple and ?others) simple file structure for storage: Artist/Album/Track hierarchy. All other detail is in the metadata. I wanted to keep the file storage as simple as possible. This made transition to storage of the music library on a NAS straightforward (something I had not contemplated when I started building the library). Incidentally if you wish to put an iTunes library on a NAS I very strongly recommend putting all the iTunes files/folders on the NAS i.e. the folder 'iTunes' and all its contents. This avoids any data corruption if for some reason the NAS becomes disassociated from the curating computer running the iTunes software.
I use the Apple 'genre' property extensively as a top-level option for selecting music to play:
e.g. Jazz-traditional, Jazz-mainstream, Jazz-modern etc
and Classical-early, Classical-baroque, Classical-classical, Classical-romantic etc.

For classical music I use 'Album' in a very specific fashion:
'Composer' - 'Work': 'qualifier'
e.g. 'Handel - Hercules: Arias'
Of course this format has to be followed consistently and accurately if sorting and searching are to work quickly and correctly.

Fortunately I chose this approach early on so the initial editing to this format was not too large, and now I edit every new album when I rip or download it. I wouldn't say my setup is perfect but it very rare that it gives me any problems. I use Doug's scripts to find lost tracks and regularly check for errors in the MinimServer scan log (Simon's checks throw up a lot of errors not apparent in iTunes).

David

[lyapounov]

Hi David

thx for the answer !

my problem is not the structure of my library; on the contrary this is very clean (I could give you access via MP to my DS audio so you can see its content)

My problem is that, because of bad tagging, I can't access it via other methods. Like : I want to see all my albums conducted my Ferenc Fricsay, as an example. Or I want to see all albums where Aril Andersen is playing. Just because I am not sure of the tagging, and besides how to tag multiple interprets is still not clear to me.

Also, rules for tagging are not clear. I used to give in 1995 as exercise to my student to sketch the conceptual model (in database sense) of classical music recording. That is a very complex structure, which I know Simon has adressed. Except that, exactly like Artificial Intelligence, you can have the best algorithm, but it is worthless if the data are not here.

What I would wish is a magic wand : I define the tagging rules, and voila all my files are properly taggued :-)

I don't know how geek you are, but as I mentioned all my files are on a synology server, and I started writing a web access to those files so that I can get the ID3 content. There is a nice php library called getID3. However, this library can't write tags for flac files :-( Pity, because I would have been close to go from Rules -> retagging.

The only php routine I have done so far is that for any album I rip, I rewrite the tracks filename in a pretty format.

Anyway, I hope I am not too geek. Sometimes I dream of an exchange place where people who ripped their album could see what is the better tagging, and exchange their files. But the stupid music industry who still understands nothing to digital age (even if they pretend to) is always here to prevent customers to be smarter than them...

Cheers
Serge

[DavidL]

Hello Serge - you are certainly more geeky than I am as regards tagging!

Below I interpolate in your posting some possible help.

(06-12-2017 04:49)lyapounov Wrote:  My problem is that, because of bad tagging, I can't access it via other methods. Like : I want to see all my albums conducted my Ferenc Fricsay, as an example. Or I want to see all albums where Aril Andersen is playing. Just because I am not sure of the tagging, and besides how to tag multiple interprets is still not clear to me.

When necessary I use searches (via BubbleUPnP) on 'artist' or 'album artist' to solve this problem. In your example I would have set Ferenc Fricsay as album artist unless the recording was a concerto, when I'd set the soloist as album artist, but I would set Fricsay in the artist list. In both instances a search would display occurrences of Fricsay.

Also, rules for tagging are not clear. I used to give in 1995 as exercise to my student to sketch the conceptual model (in database sense) of classical music recording. That is a very complex structure, which I know Simon has adressed. Except that, exactly like Artificial Intelligence, you can have the best algorithm, but it is worthless if the data are not here.

I think you are trying to squeeze a quart into a pint pot here. Tagging classical music is as you say complex and ideally you'd like to have a relational database. You can set as many tags types as you wish but in the end the music is a flat 'card-file' database so you have to rely on searches. As I described in the other thread once your library becomes large these searches/ordering can take a noticeable time.

What I would wish is a magic wand : I define the tagging rules, and voila all my files are properly taggued :-)

Hope this helps

David

[paultaylor]

(06-12-2017 04:49)lyapounov Wrote:  What I would wish is a magic wand : I define the tagging rules, and voila all my files are properly taggued :-)

Hi Serge, FWIW this is exactly what we are trying to achieve with SongKong. This is an automated tagger, but one that trys gives you enough control to tag as you require.

In general in my opinion relying on the folder structure for anything more complex than album/artist is not practical. But since in theory you can add a field for anything you want and since you can configure MinimServer to use that field you can configure it so you can generally search/organize your music as required.

[Dieter Stockert]

(05-12-2017 21:56)DavidL Wrote:  Are all your classical composer clean i.e. first name last name.

You know Gabriel García Márquez? And you know Gabriel is the first and García Márquez is the last name? But I doubt that an algorithm would do that right. So I don't think "first name last name" is a good idea.