Metadata lookup tools

How many of you would be interested in a metadata look up app that only gets its data from the major online sources and their APIs. It would not use any user curated databases like MusicBrainz. I have started using APIs to get my metadata instead of MusicBrainz and other open databases. So instead of having someone else buy a digital album, enter some of the info (often incorrectly) to a database, and then get my data from that database, I just go straight to the source. So I have decided to build an app to lookup and retrieve album metadata. Since I will be doing the work anyways, I may as well give it out/sell it if enough people are interested. Thanks

7 Likes

I would be interested, having tried software like ā€œSongkongā€ (which made a bit of a mess of my metadata) there is an obvious need for a more accurate method/system of checking & updating. Roon does a pretty good job but still displays the problems ā€œrubbish in = rubbish out.ā€
I am curious what or who a reliable ā€œmajor online sourceā€ would be as I have seen data errors on the sites of some of the major Music labels?

1 Like

Having been bitten by the inaccurate meta bug, Iā€™d also be interested. Iā€™m like PixelPopper Paul too - curious as to what the sources are and their controls on accuracy.

@PixelPopper you bring up a good point. I am sure there are mistakes on some of those websites, but that does not necessarily mean that those mistakes occur in their APIs and when they are giving their data to content providers. They very well could be, but these mistakes could also have been made by the web page designers or for some other reason. Also, what are you comparing against? The albumā€™s liner notes? I will of course look into this further, as I always am looking for the most accurate sources for myself. But currently the most accurate sources of album metadata and album credits are the publishers and labels since they are the ones creating the data in the first place. So if it is wrong from creation, it will probably be wrong most other places too. As far as sources go I never said there was a ā€œmajor online sourceā€, but there are multiple sources that get their data straight from the publishers. When comparing the data from these sources, discrepancies between each source are pretty rare.
@Chuck_Hawks there are actually a lot of sources of accurate metadata, but they are are just not easily accessible by the average user. on my end I pull data from each resource, organize it, and then compare it. when there is a discrepancy between a piece of data, like the song writers, I display both options for you to choose from.

This makes lots of sense, as not only can the end user make a decision but, presumably, there would also be a third choice, manual edit when thereā€™s an unfound option 3. For many the ā€œreferenceā€ would be based on data in printed firm i.e. cover notes.

yeah, I figured you meant liner notes but maybe you had even better sources that I could use. :laughing:
Hopefully I am on the right path here. And as more resources become available, then I will test those and include them if they will be of benefit. I am also now looking at sources for classical music and other genres that get little attention.

My plan is to give my Roon pals a copy of the app to evaluate and use when its done. Maybe it can help Roon users fill in some of the missing metadata in Roon and in their own files, and get the best/largest cover art available. In return, hopefully I can get some good feedback as well, along with some ideas that could make the app more useful. I am hoping that people will also use this when uploading/editing album data to MusicBrainz and Discogs so that it can benefit the most people.

Also, @PixelPopper thanks for taking the time to make some comments! it helps get the gears in my head turning.

2 Likes

Iā€™d definitely be interested too, having in the past resorted to SQL to standardise and ā€˜fill in the blanksā€™ to ensure all instances of a composition are correctly identified, correct and standardise performer names etc.

All the better if it becomes a front-end for Musicbrainz submission - Iā€™m currently using a website that pulls metadata from Apple Music, Deezer & Spotify to make submissions which ultimately reflect in Roon is their preferred sources donā€™t have a hit.

What would be the main sources? I donā€™t know of a reliable database where the information is entered directly by the label or the artist. In my experience, the only reliable source remains the booklet (and 99% of the time the booklet is present only in the physical edition and not in the digital one). The truth is that there is no interest from the music provider to use and make this information available except in a very superficial way.
We are the only ones who are really interested. We ā€œnerdsā€ compile the ā€œopensourceā€ databases with the most detailed information using, if possible, the booklet. Is it the best solution? Obviously not. Is it the best solution in these conditions? Unfortunately.
Discogs has a difficult API structure to use (users have a lot of freedom to use non-standard, non-coded relationships), Musicbraiz is better structured although some choices on how to handle data in relationships are questionable

1 Like

yes, I am very familiar with that website. I really like it as it cuts down the time to add an album, and all the information is accurate. My app is similar except you dont need to go to Spotify or Deezer or Apple Music to to get an album id to use in the search. Instead you just enter the album name and artist(if known) and it returns all possible matches. You click on the correct match and all of that info is displayed. I then go a step further to try and find credits. You can also search the different endpoints of these APIs like searching for a specific track, or an artistā€™s discography. I have also included some things like current most popular songs that can be filtered by genre and country etc. Not sure if this is useful or not, but it was easy to add, so I figured why not? ha ha ha

Definitely something Iā€™d find useful, even more so if it was able to automate ingestion and enrichment of existing albums on disk, and perhaps even store its data in a SQLite database.

I would say the Spotify API is the largest or main API I am using. It certainly has the most albums and tracks.

A few years ago I would agree with you 100% about the labels not caring enough to put the data out there, but there are a couple of new databases that have popped up in the last two-three years that are only fed info from the labels and publishers and musicians. Jaxsta is a great example. They have a webpage that anyone can search and they have very detailed and accurate info. Definitely a step in the right direction

Both are great ideas! Doing the Sql database is much more likely of the two and it makes a lot of sense to do that. Thanks for the idea!

I have explored Jaxsta. Many releases are missing, there are many release date and label errors, and the credits are very minimal
This is just one example:
https://musicbrainz.org/release/c4b487c9-7fce-47db-a36d-b28e6fe04e33

https://jaxsta.com/release/a594be87-5b37-4e85-aaca-36a46cbea1c7/7a57be87-714b-5926-a0c4-a0326b461768/overview

Jaxta only has 2 versions against 11 of musicbrainz. Neither version of Jaxsta is the limited edition which has 2 extra bonus tracks.
Credits are also limited to composers, lyricists, producers and performers.
Also, unlike musicbrainz or discogs, you donā€™t have the option to add or edit here

:rofl: :rofl: The one example of an album on MusicBrainz that has complete and accurate data. A diamond in the rough.

Also, I never said Jaxsta was perfect. I simply mentioned them as an example of a database fed exclusively by labels and publishers. Their credits are also not limited like you are saying they are.

You can like MusicBrainz all you want. I am glad it works for you. But it doesnā€™t for me and many others.

No, I simply wrote relationship following the booklet (which I do with all the albums I buy). I donā€™t know any ā€œofficialā€ databases compiled from the labels where I can find the same information.
Can you provide some database examples with some release examples where I can find all the credits written as in the booklet?

Right, I understand. I always do the same thing when I would upload albums to MusicBrainz, but most people do not. I think if MusicBrainz has a better system of entering credits, more people would do it.

Jaxsta is the only public official database that I am aware of. Other databases are mostly for supplying data to apps such as Roon and they donā€™t have public websites.

The following link is an example where Jaxsta has credits that are as good as the booklet. In this case, the credits on Jaxsta are actually more complete than what is in the booklet. Jaxsta

Itā€™s not too often that Musicbrainz has that level of metadata. Iā€™d happily take what Jaxsta provides over the usual track title, performer, album & release date, especially with the sheer volume of albums flooding the market every week. The end objective is to combine metadata from various sources and present something that is richer than but no worse than any single source - that can only be a good thing.

1 Like

Some Screenshots. Clicking on the rows in the table opens that albumā€™s details, track list, etcā€¦

I should mention that the results of my search are only the results from Spotify. If there are different releases available digitally, they will be shown as well.

(do not mind the missing album cover, it is there now after a quick change)

1 Like

Looks great, happy to try as soon as youā€™re ready.

I donā€™t quite understand the issue here, nor why your proposal would fix it. But I do see a need for a better source of information. What we need is an IMDb for music. I would love to be able to search albums/tracks by the session musicians involved in the recording, for instance, or the producer. That is something Roon should do better, but I guess that info is not easy come by.

I think just getting musicbrainz to allow additions to their database would be a better solution.