Just where does CD metadata come from...and how does PerfectMeta work???

[d2b]

How does dBpoweramp "decide" which metadata is best? I realize that there are simple ways to "vote" on the available metadata from on-line sources such as discogs,sonata, freedb and so on. But there's likely a wealth of information already built into a commercially-manufactured CD from a responsible record label. This possibly-most-accurate information is presumably listed in 'containers' on the CD called CD-Text, CD-ISRC and CD-UPC. Of course not everyone includes this information on new CDs perhaps out of laziness, ignorance and maybe a lack of knowledge on their part or the company that manufactures the actual discs. Too bad, because we often have to rely on crowd-sourced metadata which could be wrong..and often is not accurate even if nearly so.

To further complicate the gathering of metadata, there are options within CD Ripper to write tags such as Conductor, Soloist, Style, ..., etc but with NO obvious way to enter those tags manually that I've been able to discover. Not only that, a lot of these optional tags NEVER show up from ANY on-line metadata provider.

Then, as I previously reported, sometimes the built-in metadata on the CD seems to be either ignored or even deleted from the screen. I believe Spoon is looking in to this issue.

I've also noted that some CDs I've ripped and tagged within dBpoweramp, when reinserted later that day or the next, do not retrieve all of the as-ripped tags from the dBpoweramp cache. Now that's bizarre.

Is there a source document that describes specifically what is going on when a new CD is inserted and the CD Ripper screen finally settles down and displays something?

Does that document explain how such tags as ORCHESTRA are found or at least added manually by the operator? Does the on-CD text information have priority over web-harvested PerfectMeta results? Who decides? How does this work? What metadata fields are supported by the on-line metadata providers? Do "they" ever provide ISRC data? Do they ever populate fields such as CONDUCTOR, STYLE, LYRICIST and so on?

Sorry...I'm just frustrated trying to figure out if CD Ripper is really working well when it comes to optimizing the collection of metadata while at the same time wondering if the optional tags in the metadata configuration really mean anything.

[garym]

Re: Just where does CD metadata come from...and how does PerfectMeta work???

Very few of my over 5000 CDs have CD-text or other built in metadata. The metadata must come from The cloud (mostly crowdsourced with plenty of errors). One can manually enter any metadata field. For sure after rip. Not sure about doing them all at ripping time as I havent tried.

[d2b]

Re: Just where does CD metadata come from...and how does PerfectMeta work???

I normally use mp3tag to tag CDs after ripping securely with dBpoweramp. But one of the questions was (reworded for clarity): How does one use dBpoweramp to add tags other than what appears at the top of the screen when ripping a CD?

And how do you know that there's no CD-Text on many of those CDs? Once dBpoweramp settles down after inserting and searching for on-line metadata, there's no indication that what you see might have come from CD-Text.

Are the ISRC codes that occasionally appear from the CD or from an on-line source? It's not at all obvious what is provided on-line...except for the very basic tags displayed in PerfectMeta. Looking at the PerfectMeta page, it looks like only a few very basic tags are covered by the on-line providers: Lead Artist, Album, Album Artist, Genre, Year, Disc Number/Total Discs, Title, Artist and Composer.

So why are there all of those check boxes in the CD Ripper configuration options?

[garym]

Re: Just where does CD metadata come from...and how does PerfectMeta work???

For settings set option on retrieve metadata from menu to only CD Text. And nothing will show up when you try to rip. So there is no CD text. There may be other ways to tell. I dont bother checking this on every cd obviously. But conventional wisdom from many people that rip lots of CDs also informs me that very few have cd text.

I know that if I select all those many tag fields in the settings, many will be filled in upon ripping for *some* CDs (eg style). But I cant see how to make these fields show up in the ripper window for manual editing. I only use the basic tags myself so this is not an issue for me. I also use mp3tag for post rip editing. Extended tags there shows me all the tags.

[schmidj]

Re: Just where does CD metadata come from...and how does PerfectMeta work???

Metadata , the bane of my existence.

You can enter whatever metadata you want when (before) ripping. Titles, artists, etc. that appear (or don't) in the album listing can all be clicked on after you highlight a track and edited in the same way you can change file names in explorer. Or you can type the entry at the top of the page. If you highlight the blank line under the bottom track, any entry typed in the spaces at the top of the page will be entered for all the tracks, useful if the same artist is on all the tracks, for instance.

If you right click on the headings at the top, you'll get a long dropdown list of tags you can add as columns on the track list.

Another way to see/edit tags is to select the tags tab in the bottom center. Select a track from the tracks above, you'll see (and can edit) all the tags. Select the blank line under the last track, you'll see the tags common to all tracks.

Be aware that tags are generally stored only in the actual file with the music, i.e. the FLAC or mp3 or whatever file, not in a directory that contains the files. You can really create a mess if you for instance change the album name in some but not all of the tracks of an album. The players that use tags will think it is two different albums.

Assuming you are using the current version (16.whatever) you can click on the thing that looks like a luggage tag at the top which shows you what metadata was recovered and where it came from. Stuff from cd-text just ends up in the chosen field, there is no separate column for it. When you have that "manual metadata review" page displayed, the red entries are the ones not chosen, click on one and it will replace the chosen one. Note the check box for view track artist, that also displays the track composer metadata. Also note the "compilation" check box, used for indicating compilation CDs. Check or uncheck if you think its wrong for the CD you are dealing with.

If you re-rip a CD before the cache overflows, it will pull the metadata from when you last ripped it from the cache, for better or worse. But again, you can edit it all you want.

It's really messed up, and you want to start the metadata over? Click on the box at the top with the circular arrows. That will requery the metadata sources. Use with caution, it will clear any manual entries you have made.

Click on the box at the top with the green lines. The drop down entry "active providers", if you click on it, will tell you what sources for your metadata are "turned on". The ones with the ticks are on. Want to turn one on or off, click on it, the tick will change to the opposite state.

The drop down entry on the green lines box that says "clear track metadata" will do just that. This is useful if some of the entries are for the totally wrong CD, as there is quite a bit of "hidden" metadata (not really hidden, the "Tags" tab in the middle bottom will show all, I believe) and some of it, like incorrect "sort" tags can cause all kinds of confusion when you use a player to select an artist or such.

Do take a never ripped CD and play with the tagging options. See how they work. Rip the CD and then right click on the tracks and click on "select ID tag" and see what is there. Select all the tracks then click on select ID tag and you can edit all the tracks at once (for instance to fix the album name), or you can scroll through the tracks one-by-one and edit the tags individually. You can also download a free program, "mp3tag" which is a multi-featured tag editor (not just for mp3's, it works for most any audio file now).

If you really messed up that CD you just ripped, then select it with windows explorer, delete it and rip it again if you want. Just remember that the cache is bringing back the probably messed up metadata from the last entry, so either edit it to fix it or click on the box with the curved arrows to download the metadata again.

A little more on metadata recovery: Most software, including dBpoweramp use the track numbers and lengths recovered from the CD to search the databases for matching metadata. If there are more than a few tracks, the probability of a mismatch is fairly low, although it does happen. If you have a CD single or only a couple of track, there in all likelihood have been several CDs pressed with the same length tracks and the probability of the database returning the wrong data becomes very high. So you'll probably have to manually enter data if you go to rip CD singles, or others with few tracks.

Now there are a couple of services which use "fingerprints" to match CDs (or other audio) to metadata. The program "Music Brainz Picard is a free one. But the database is still limited, not too many albums have had fingerprints made yet. In the long term, I'd suspect fingerprint databases to become more common. That will greatly improve the accuracy of the metadata recovery. But the fingerprint algorithms have limitations. PerfectTunes uses fingerprints to find duplicates in your collection. But in addition to finding the duplicates in "best of" albums, which I'd expect, it also flagged cover versions by a different artist as duplicates to my surprise.

Now where does all that on-line metadata come from? Originally crowd sourced. From people who ripped CDs and entered the data and then submitted it to the database. Now new CDs mostly are added to the metadata by "members" who are usually the record company or artist. As anyone who has looked at what is displayed can see, it is strewn with errors. And though the metadata source businesses think they get better quality and consistency, from what I see, the "crowd source" users were as good or better than the "professionals" at entering accurate metadata.

Also be aware that there are many metadata sources beyond the 5 currently used by dBpoweramp, some much more complete for unusual genres. But because of contractual limitations or business issues (the source wants to be paid too much for access) they aren't available on dBpoweramp. Some are available to individuals for free or small subscription fees. I do wish that spoon would write an API that would let computer literate people subscribe to other databases if they wish (and can arrange a subscription) or even to recover metadata they may have in their own personal (or business, as many radio station automation system) databases and interface that with dBpoweramp to improve the metadata situation. Even a formatted select and track or album paste would be useful.

Anyway, that's at least part of the situation. Experiment and learn.