I am working on tagging my classical opera collection. To do this I have previously used the detailed info from from the CD booklet which accompanied the CD.
Unfortunately I moved house recently and all my CD booklets are in store and will not be accessible for a while. I need to refer to these and the libretto text to find the performer details for each track.
Is there any way to find this info other than digging in my store container?
Any guidance and assistance would be appreciated.
I feel your pain. I also enjoy listening to opera. I also have a number of opera releases which I entered into MusicBrainz to a basic level, but would like to get to a better level, with all the performer relationships attached to all the necessary tracks.
I’m afraid that it is not really possible to do that without the specific evidence about that specific album. That means finding their CD booklets or equivalent. I think it’s not good to rely on information from other releases. There are too many ways that information could be misleading. Recording producers can choose track divisions which differ from track divisions on other recordings, and don’t match the structural divisions in the score. Conductors and directors of recordings can do things like assign parts from one singer to another, or consolidate roles. Knowing how things were done for another release doesn’t confirm how they were done for the release you are working on.
Maaaaybe, if you are fortunate, you can find a copy of the CD booklet as a PDF file on the producer’s web site. Or, another contributor with the same exact release might have scanned the relevant pages of the CD booklet and added them to the Cover Art archive. In that case, once you have made sure that the info is really, truly about the same release, you can work from that to enter performer relationships between the artists and recordings of your release.
For operas, musicals, oratorios, and other works where there are named roles, and performers who are performing those roles, there is a structure which MusicBrainz could describe, but does not currently do so.
I have been experimenting with a possible structure for roles, using annotations. One could list the roles of a work (opera) in the Work pages. See for example the annotation of work/425f46a, the Work entry for Meyerbeer’s Le Prophète. Then, in a Release with a performance of that Work, one could make a list of the roles, and which singer performs each one. See for example the annotation of release/9a57002, the Release entry for the 1976 Le Prophète with Horne, McCracken, Lewis, and the RPO. And for each performer relationship to a track in that Release, include the role name in the relationship’s Credited As attribute. See for example, “Zacharie”, “Berthe”, “Fidès”, and “Jonas” in the vocalist relationships with recording/7c7d37f, the Recording entry for disc 1 track 3, Le Prophète: Act I: “Fidès, ma bonne mère”.
If someone entered such information about roles in a work and in a release, that might be enough information to enter performer relationships. However, role information is somewhat theoretical. The CD booklet is the most likely to have the real information.
So, my suggestion is, do what you can now, with the information you have. Later, dig the CD booklets out of storage, and improve the MusicBrainz entries based on that evidence.
The good news is, it is perfectly legitimate to do MusicBrainz data entry in stages. It is less tiring, too.
Hi Jim,
Many thanks for your reply. It is good to learn that others have been struggling with this and for sharing your own experience.
I note your posts in the other thread I had started and hope you will forgive my misuse of the terminology I used. As you will understand I am very new to all this and am at the bottom of a steep learning curve. I also am having to overcome the assertions in CSG which in the case of track performer are not what I need, but I think you know this.
Before I embarked on this I had been doing all my own tagging using EasyTag, having been put off by my experiences with SongKong some six years ago. I used to create simple text file with the list of singers, voice and part and then, using the booklet track text, cut and past the appropriate names in the Track Artist field.
With all my collection disks in storage I decided to try SongKong again and overall have been very pleased with the results which has brought me to this point. Unfortunately this reversed some of my own work but overall my modest collection is now almost fully tagged, albeit with the basic style tagging. I now need to work on my operas.
Your post is exactly what I needed and I am full of admiration for what you have done. This leaves me with so many questions on the details of how to do what you are doing I am not sure where to s so I shall just say thanks again. If you have time for more questions from this novice please let me know.
Regards,
Budge.
Hello, Budge!
No worries! Classical music metadata is complex in the real world, and therefore is also complex in MusicBrainz. All of us have been up this learning curve. Part of the learning curve is seeing other use cases than one’s own, and how that affects what the core MusicBrainz data model should be.
There is nothing stopping you for doing a similar workflow, but using the MusicBrainz database and the MusicBrainz Picard tagger. Open your music files in Picard, let Picard populate the metadata fields using the MusicBrainz data, then manually copy text from the Performer and Conductor fields into the Track Artist field. Save the music files from Picard. Result: the music files have exactly the Track Artist tags which you want.
This does require you to repeat this manual intervention every time you wan to update the metadata from MusicBrainz. But if you don’t intend to update often, this is not much of a problem.
You are welcome to keep coming back to this forum and creating new topics for new questions as you need to. There are many of us here who are ready to help. Part of our reward will be that you will no doubt contribute improved data to MusicBrainz along the way. Have fun!
Best regards,
—Jim DeLaHunt
Hi Jim,
Many thanks again. I shall have a look at Picard. I have it in my workstation but as I recall there is a dearth of Performer info.
As I mentioned earlier I have been trying SongKong (SK) to populate my entire collection and this has been successful for 99% of my collection with only the operas causing issues. SK works using the MB database so I do not expect there to be any big differences and the issues to which I refer are my wish to see who is singing in each track, a desire which it seems not so many share.
I have also started a new thread on what I sometimes refer to as housekeeping as I continue to work on curating my collection.
Regards,
Budge
Oh interesting, I am not familiar with SongKong. Looking at their website, I see that they claim to “understand classical”. They say,
Track Artist… in Classical music, usually various performers, an orchestra or choir, a conductor and a composer are all credited. The default for SongKong is to store the Performers, Ensemble/Choir/Orchestra and Conductor in that order, but we offer different masks for different uses.
If I was writing a tagging app with that description, I would have an option to add names of all performers mentioned in a Recording’s performance relationships to the Artists metadata tag in the music file.
If SongKong can read CSG-compliant MusicBrainz data, and tag your files with the Artists contents which you seek, then you don’t need to add extra entries to MusicBrainz TrackArtist fields, or lobby for changes to the CSG. From what I see on their website, the developers of SongKong seem to be the sort that already be trying to do this, and would be open to suggestions from paying customers about what issues to fix.
Best regards,
—Jim DeLaHunt
Hi, to summarize SongKong uses an algorithm to identify if a release is classical (you can also force it to consider a release classical or non-classical), then if it is classical you have options that only apply to classical releases, such as adding recording performers to the Track Artist (they are also added to Performer field with their role/instrument)
But the trouble that @Budgie2 was having is for the opera release we looked at the data was missing from MusicBrainz in two ways
There were three release performer relationships, but since the performers were only involved in some tracks they should have been recording level performer relationships.
There were two performers listed as album artists, but they had no performing relationship so it was not possible to identify that they were actually performers rather than something else.
It looks like progress has been made on editing the release https://musicbrainz.org/release/73e029ea-107f-4a81-aacb-88545faff197/edits good work
Keep in mind release level is the right place for people who perform on just some recordings, unless you know which recordings they are (people who perform on all recordings do not belong at the release level, they should just be added to all recordings instead).
So if I have understood you correctly it is acceptable to add release level performer relationships if missing the information to pin down to the exact recordings they are on. But if you do have the information then it is always incorrect to use release level performer relationships.
@ijabz Thank you for the reply. It is great to have tool developers speaking up on the forums with well-informed posts. This is a great example.
Also,
@Budgie2 Do you agree that this was the trouble? In another thread you were advocating to enter performer data into an opera Release entry which interested you, in a way different from the Classical Style Guide. Now it seems to be that your primary obstacle was that the data was absent from the Release entry, and that your tagging tool will give you the results you want, if the data is entered according to the Classical Style Guide.
That is a good outcome: you get what you want, you avoid having to advocate for an exception to the rules, and the MusicBrainz project gets a contribution of well-structured data in that entry.
Yes 
The only stuff that belongs ideally at the release level is stuff that affects the release rather than the recordings (things like cover art design). See the relationships guideline.
Hi all,
Many thanks for all who have contributed. The advice and comments are much appreciated. I understand the views concerning Artist vs Composer. As far as I am concerned the guide is just wrong for classical music artists tracks but who am I to blow against the wind.
Fortunately SongKong goes a good way to fixing this and I have been very pleased with the results from SongKong which for me is much better than when I first tried it in 2017. It is unfortunate that my choice of the René Jacobs Cosi fan tutte just happened to have some variants I could not fathom without the box and booklet.
Meanwhile I am on hold with this topic until I can get hold of my CD disks and cases which are still in a container as I do not have the data I need. Unfortunately it will be a while as I now have no space.
I do have a further thought on possible future improvements:
Would it be possible to enhance the database with a library of pdf copies of the booklets published with the CDs? I am sure others have thought about this but has anything been done?
Best wishes to all,
Budge
Cover art already supports pdf as one of the options, alongside images, so this is already possible.
Actually i had a question about that. Although cover art supports pdf in most cases users upload separate images for each page of the booklet and do not upload the booklet itself, why is that, are pdf uploads discouraged ?
IMO for most users access to the actual booket would be much more useful. There are practical reasons for this such as MinimServer UPnP Server which is commonly used for Hi-end audio systems because it is much more configurable than alternative (such as asset/Twonky) has an option to serve a pdf booklet, but this can only be used to show one item so images of each page are no use here.
Because if you scan a booklet, what you generally get are images, that you would then have to put together to get a pdf, I expect.
Additionally, images allow you to scan only pages that are relevant (only per-track credits for example) if you feel uncomfortable with uploading the entire booklet (some classical labels explicitly ask not to share the whole booklets with others since they have a lot of extra “academic” notes and background about the music which are both copyrighted and part of the added value for the CD).
And at least for me it’s a lot faster to find the credit images from the cover art page based on the thumbnails and load only those if I’m trying to use them to check whether an edit is correct than it would be to download the whole booklet in pdf and then find the right part. Since my main usage for the CAA booklets are as proof for edits, that’s what I do when I rarely scan something.
But they’re not discouraged in any other way - if I get a digital release which comes with a PDF, I’m certainly going to upload that PDF unless the label explicitly asks me not to
Thanks, also unfortunately I cannot find an example link now but I noticed that sometimes although there is no pdf listed in Musicbrainz Cover Art Archive if I go to the Internet Archive they may well have the pdf.
Do you know given a MusicBrainz Release Id is there a way to programmatically go direct to the Internet Archive item and check for a pdf there ?
I have only a basic understanding of the relationship between MusicBrainz, the Cover Art Archive, and the Internet Archive. But my impression is that the Cover Art Archive is just a small part of the Internet Archive. A lot of people upload a lot of content to the Internet Archive independent of MusicBrainz and the Cover Art Archive. My experience with those uploads is that many are poorly catalogued and have little metadata describing them.
So, if someone uploads a PDF file, that happens to describe a music release, to the general part of the Internet Archive, I expect that it would be unlikely but not impossible for that person to upload metadata with a MusicBrainz Release ID, and so it would be unlikely that an Internet Archive API could find it from the ID.
The way to upload a PDF to the Internet Archive in a way that is correlated with a Release ID, and is retrievable programmatically, is to use the MusicBrainz Cover Art user interface. As far as I understand.
Hi and many thanks. I have not actually tried what you state as the pdf’s I have in my data have been from when I have made an on line digital purchase. I only have a few of these as until now I have purchased the actual CD sets which come with the booklet. My work on this is on hold for now - too much real work!
Thanks for the reply,
Budgie2
Yes, but unfortunately most of time for whatever reason MusicBrainz editors seem to prefer scans of individual pages rather than uploading whole booklet so their is not many pdfs avaialble via MusicBrainz Cover Art UI.
Good point. That sounds like a problem with style guidelines and with editor training, rather than an API deficiency.
One style guideline issue might be: does uploading a complete booklet create a significantly greater copyright risk, compared to uploading selected pages for documenting metadata? Another issue might be: what is the value added, and to whom, by uploading a PDF file of the complete booklet.
If your tools do something good with PDF booklets when a users plays a release, it would be helpful for you to illustrate that in a blog post on your site. Then both the style guidelines and the user instructions have the option of linking to that blog post to say, “See? If you upload a PDF booklet, you enable this helpful behaviour.” And the blog post helps with your own sales efforts.
