Classical metadata

[PAR]

At "Q" this is no longer a problem as they are now inserting work data as a field above track data for recent releases , rather similar to printed CD booklet listings.

 

The problem remains  both at media player level ( almost all of which only work at album/artist/track levels ) and at record company level.  I still retain contacts in the record business after a large part of my life working in it ( ultimately being involved in a small way on ISO metadata projects). I am told that even now, at least here in the UK,  few practitioners personally use computer file based audio systems and thus have no idea at all of how the metadata supplied by them to accompany downloads or streams  displays in real life.

[PAR]

I can tell that none of you have worked in the record industry. Record metadata is based upon what is called "label copy".  That means what it says, the text that will appear on a record label. Record company thinking has not evolved much beyond that.

 

Each record company has its own protocols for their version of "label copy". None are willing to change it.

 

Even worse is the situation that is common now.  I have just had dinner this evening with a close friend who is an executive from a major record company .We were playing LPs ( do I get banned from this forum?) when he asked me to put on side 2 of one of his company's releases. I pointed out that I could not as the label had no "side" information. His reaction was " F*k nobody even looks at the label copy now".

 

I spent some years of my working life on behalf of the record industry's official bodies visiting the UK record companies and pressing plants and other industry bodies in London, Amsterdam and New York trying to encourage one simple subset of consistent metadata . Ever tried to herd cats?

 

The record industry ultimately only cares about money. If you can come up with a scheme that says "get the metadata correct on this album and you get $1,000"  or if there are legal sanctions for non-compliance then otherwise forget it . Unless that happens you will just have to put up with what you get.

 

 

[PAR]

I appreciate your enthusiasm for trying to advance a solution for a matter that frustrates all of us.

 

However any solution must be made on an international basis as recordings  are made and sold internationally. So having a standard that is adopted by, say, USA based record companies only is  no good if much of the classical music that you are buying is created in, say, the UK or the rest of Europe.

 

So , even before deciding what the ideal metadata should consist of you have to consider how the international adoption of it could come about.  I would strongly suggest that the only practical route is via the IFPI (International Federation of the Phonographic Industry) . However you would need to find a way of gaining access to them. Very difficult for outsiders. This is a representative trade body which would normally rely upon its member companies to promulgate new ideas to its executive. BTW, not all record companies in the world are associated with IFPI (via their local trade association) so even were they to adopt the kind of scheme you envisage there is no guarantee that everyone would comply with it.

 

Actually, looking more deeply into the problem I have concluded that the record company end is not mainly responsible for the difficulty that classical music fans have in seeing usable information about what they are listening to. Yes there are cock ups from time to time but one must treat those as exceptions.

 

Have you some downloads purchased from legitimate online vendors ? The recording file and its accompanying metadata do not originate from the vendor but from the record company. If your downloads are in FLAC format do you know how to open the ID3 file? If you do then I suggest that you look at the ID3 file for a track  that has been downloaded. You will most likely find that ALL of the metadata that you require is present.

 

The problem that we have is that the designers of the playing applications have not provided access to all of the main fields of interest but only to a selection.  They are , album title , artist ( or album artist) and  track title. They do not provide a field in the display for work or for composer even though that data is already in the ID3 file.

 

There can be a small problem to overcome with the "work"field  as that can be separate or part of the album title or the track title fields . There is a way round this which Qobuz have adopted in the past month  by showing the work as a field above the track or tracks that relate to it.  Not all player apps from Qobuz partners  will display this but I am pleased ( for my own selfish needs) that , for example, dCS have amended their version of the Qobuz player in the Mosaic control app so that the work title ( for example, Composer name, Etude in G minor) is concatenated with the track title ( e.g. Allegro) so the display for track now reads; Composer name, Etude in G Minor, Allegro. That more or less meets my expectation.

[PAR]

22 minutes ago, EdmontonCanuck said:

 

I agree. I always thought that some sort of crowd-sourcing model based on the Wikipedia model would be the only way to solve this. People would submit their tagged albums and the "crowd" would maintain and ensure integrity. It would be slow-going at first, but eventually it would build up to something useful that automated tagging engines could source to help us maintain our collections.

 

Unfortunately any Wiki solution will almost certainly be unsuccessful. They already exist e.g. Music Brainz . Discogs, Freedb, GD3 etc. Some are better than others but most are ultimately unreliable particularity for classical music. For example the number of entries that have the composer as the artist is unbelievable. I can guess why. It is because the player apps do not have a field for composer. So the person who has provided the input to the wiki database has replicated is own work around to get over the missing field and which allows his player app to display some kind of meaningful identification of the recording.The fact that his personal workaround may be inappropriate for anyone else or for anyone not using the same player application seems not to have occurred to him/her. In fact it is clear to me that not too many of the "crowd" that currently populate these databases necessarily understand how to create accurate metadata that can be applied universally. Often they will just take the easy way out e.g. Composer "Bach" , but which one? Composer Bacharach but no David etc.I could go on endlessly giving examples but will discipline myself not to do it right now 🙂

[PAR]

1 hour ago, Musicophile said:

I understand that older crowdsourcing efforts are a mess, I witnessed it myself when I ripped my CD collection. A crowdsourcing effort would only work with rigorous standards.  And the first thing one would have to set up is a unique identifier for composers, to ensure that Chostakovich Shostakovich and Schostakowitsch aren’t triple counted.  And not confuse the different Bachs etc. So ideally drop downs for most of the standard fields. Same for work names (symphony or Sinfonie etc.). That‘s how I tagged all of my thousands of albums manually. 

 

So this could be sorted with the appropriate governance. But the problem will remain who will be able to manage the overall platform?

And who is going to decide what spelling to use? It isn't just composer name but work . Symphonie Fantastique or Fantastic Symphony? Les Biches or The Deer, Does or Hinds, Les Illuminations or The Floodlights 😉. And when it comes to rendering Cyrillic or Arabic etc. which other language is correct? We English speakers don't even agree with each other in our own native tongue(s). Oh, what about artist names? Berlin Philharmonic of Berliner Philharmoniker?

 

BTW , the point of this post is to demonstrate just one issue that would cause endless argument.

 

Incidently the different spellings of Shostakovich is a completely different sort of  problem to which of the Bach family the composer is. One is linguistic the other factual. A list of values will not solve the latter.

 

I think one should stick to what the issuing record company called it.  That is the echt metadata.
 

 

[PAR]

Although there are certainly metadata issues that need attention, having re-read this entire thread  the major issue is not that the data does not exist or is incorrect but that the designers of music players have created them with only pop music in mind. It is the absence of fields for work or composer that cause the difficulties for classical music. I do not see how some  kind of wiki based metadata for classical music will impact this. The managing committee of the wiki can decide upon a standard spelling of Shostakovich but this is redundant if there is no place for it in the player software.

 

One thing that has happened though, particularly as this thread stems from the Official Qobuz Issues thread in the streaming forum, is the positive reaction from Qobuz.  They have now added a composer/work field for recent releases and promise a better search engine at the end of the summer or thereabouts.

 

So that leaves  JRiver, Audirvana, Tidal, Spotify, Apple  and, no doubt, Amazon for their forthcoming hi-res initiative etc. etc. The real meat of this issue is how to  persuade them  to invest in a revision of their software just  to please what is doubtless a minority of their customers.

[PAR]

2 hours ago, Mike48 said:

If the changes reportedly being made by Qobuz come to fruition, that will be a nice interim step.

 

Two screenshots below. The lower is of what Qobuz provided for the metadata of a new release of Malcolm Arnold works a few months ago. As you will see you have no idea of what anything is except for track1. The one above it is as Qobuz have amended it recently where now the work is added in grey above the track titles. It is a similar layout to what you would find in a CD booklet.

 

As for some kind of universally correct metadata, having been part of record industry working groups in the past on the implementation of ISO 3901 and subsequent related issues one of which concerned the creation of a lexicon of standard terms to describe every contribution to the entire recording including some of those under discussion here, I wish you all the very best of luck. It was a vast piece of work which was still underway when I left the industry some years later - and that was with industry professionals and considerable resources.  Was it ever completed? Even if you do get there  some bright spark at a record company is bound to  undermine it if they find a way of exploiting it to their advantage. As I said earlier, herding cats.

[PAR]

The tagging fields for composer and work may be available in JRiver but only for tagging. The problem is that these fields  are then not displayed in the JRiver players; Gizmo, JRemote, Panel etc. So  the fields may exist but are hidden and have no practical consequence.

 

Frankly this standards issue is IMO a red herring. The problem for classical music , as I keep saying, is that the music players are only designed for pop music and that the structure ( not the content) is incompatible with classical music. I don't care if Shostakovitch is spelt in this way or as Schostakowitsch. I am not so dim that I cannot recognise them as the same composer.

[PAR]

1 hour ago, Musicophile said:

Same for the work tag, I know that a work tagged "Mondscheinsonate", "Piano Sonata No. 14", or "Sonate pour piano op. 27 no. 2" are all the same thing. A+ doesn't. 

 

So I still see the tagging as a major issue.  

Yes, I know what you are getting at but it will be difficult to resolve the fact that the metadata is originally prepared by the record company and record companies are not all of one nation. So the metadata prepared by Harmonia Mundi, for example, will be in  French, by Hyperion in English but use British spelling and so forth. Further, many companies, including those above, may use the language that the composer used for the work e.g. the prevalence of Italian as the "international language" of music. So, Le Nozze di Figaro is used, for example,  by the French company Harmonia Mundi, not even Le Marriage de Figaro . The designations of movements in Beethoven sonatas are mostly in Italian but sometimes ( e.g. Les Adieux - BTW that is never translated into English) in German.

 

Arriving at completely standardised tagging in respect of language alone and with a standard which will be accepted worldwide I believe to be unattainable.

[PAR]

That would be a nice point if it were correct these days. Sure they have a lot of back catalogue, however their activity in the classical world these days is virtually token.  Nearly all new classical releases ( including re-releases of back catalogue) are from independents.

 

When I say nearly all I am not exaggerating. The classical music recording release schedule for this August here in the UK  includes 143 from independent labels and ( count 'em) 2 from the majors ( 1 from Decca, 1 from Warners). Although the ratio will vary month to month the overall picture has been like this for years.

[PAR]

Yes indeed especially as not all languages use the Latin alphabet.  I have mentioned in this thread that I have had some experience of working on metadata issues at an international level . I will never forget the Japanese delegation at one meeting unrolling a huge piece of paper showing dozens of characters and saying " Hey, don't forget us, we have three alphabets".

[PAR]

I listed 3 JRiver controllers two of which run respectively only on Android or iOS tablets or phones. You cannot add fields to them ( you can't right click for a start).