PDA

View Full Version : Idle curiosity about online metadata - how does it work with vinyl copies??



donmacn
2014-01-10, 11:26
Hi folks,

Just a quick non-technical question for a Friday eve.

I've started re-ripping CDs to FLAC on a VBA, and stuck in two of my 'vinyl copy' CDs - i.e. .wav files created from whatever software I was using at the time to capture vinyl.

I expected to have to enter the album data manually, but the online sources popped up with the right information. Colour me amazed! How does this work, given that the CD is my copy of a vinyl record?? is it simply based on the combination of number of tracks and their respective lengths? Does this generate a significantly unique combination that the data sources can draw on??

Not a problem (for a change!) just curious about the wonders of technology.

Thanks

D

adhawkins
2014-01-10, 13:04
Hi,

In article <donmacn.67qpon (AT) no-mx (DOT) forums.slimdevices.com>,
> I expected to have to enter the album data manually, but the online
> sources popped up with the right information. Colour me amazed! How does
> this work, given that the CD is my copy of a vinyl record?? is it simply
> based on the combination of number of tracks and their respective
> lengths? Does this generate a significantly unique combination that the
> data sources can draw on??

It could be matching against a 'fingerprint' of the music itself.

Andy

aubuti
2014-01-10, 13:06
That surprises me, too. What ripping software are you using, and which online database(s) does it access? If multiple databases, do you have any idea which PID'd (positively identified) the album?

JJZolx
2014-01-10, 14:08
Yes, they match albums by the number and length of the tracks. Probably with some leeway for track length, so it's not surprising that you're getting matches.

pippin
2014-01-10, 15:48
Yep, I've seen the same thing, even years ago when the lookup libraries where not as good as today and definitely did not do fingerprinting.
Turns out a sequence of durations is quite characteristic. Sometimes it offers two choices but most of the time it just works.

donmacn
2014-01-10, 16:59
That surprises me, too. What ripping software are you using, and which online database(s) does it access? If multiple databases, do you have any idea which PID'd (positively identified) the album?

Hi,

Ref the above - I don't actually know the answer to all of these questions. I'm using dbpoweramp and, while it pops up very quickly when the CD is entered, it seems to be saying that it's communicating to accurateRip database, then it usually lists three or four metadata sources: freedb, musicbrainz, GD3 & AMG (standard stuff, I imagine). I don't know which of those is offering the positive ID, but something is.

The two albums concerned so far were both by Jethro Tull: 'Rock Island', and 'Too old to Rock 'n Roll; too young to die'. But there's about 350-400 more vinyl copy CDs where they came from... so the least amount of typing I have to do, the better!

I guess that I figured it must be something like: if album has x tracks of y duration then it's probably z, and that the assorted matches might be offered, but on the other hand with so many albums out there, I thought there would be too many similar matches?

I was surprised because I used (I think) a programme called 'CleanPlus'. This allowed me to put in my own 'fade up' and 'fade down' top and tail to the individual songs/tracks, so the track durations might be further distanced from the originals on the album, making a match less likely.

As I say - delighted it works, I was just curious about how.

Stig Nygaard
2014-01-11, 02:53
I'm using dbpoweramp and, while it pops up very quickly when the CD is entered, it seems to be saying that it's communicating to accurateRip database, then it usually lists three or four metadata sources: freedb, musicbrainz, GD3 & AMG (standard stuff, I imagine). I don't know which of those is offering the positive ID, but something is.

Musicbrainz uses track lengths/order to identify, the other sources probably the same. You should be able to see which of the four metadata sources finds a match by pressing the "Review Metadata" button on bottom right panel (You might have to press "Meta" button in top panel first to refresh data).

The accurateRip database is to verify correct ripping against other rips of same CD (some kind of fingerprint matching, don't know details).

jimbobvfr400
2014-01-11, 06:33
I had a similar thing when I ripped some home made discs. Thing is mine were actually compilations of various songs I'd done. Very confusing at first but I later realised it wasn't guessing from online sources but was using CD text data on the disc. I only figured it out when the same CD showed track details on my car stereo.

Sent from my HUAWEI Y300-0100 using Tapatalk

Stig Nygaard
2014-01-11, 07:46
I had a similar thing when I ripped some home made discs. Thing is mine were actually compilations of various songs I'd done. Very confusing at first but I later realised it wasn't guessing from online sources but was using CD text data on the disc. I only figured it out when the same CD showed track details on my car stereo.

Sent from my HUAWEI Y300-0100 using Tapatalk

Oh, yes. That is of course a very likely possibility :-)

Also dbPoweramp actually caches previously entered data, so if you rip a CD a second time, it shows what you might have manually entered the first time - even if long time ago.

aubuti
2014-01-11, 18:16
Ref the above - I don't actually know the answer to all of these questions. I'm using dbpoweramp and, while it pops up very quickly when the CD is entered, it seems to be saying that it's communicating to accurateRip database, then it usually lists three or four metadata sources: freedb, musicbrainz, GD3 & AMG (standard stuff, I imagine). I don't know which of those is offering the positive ID, but something is.
On the lower right of the dBpoweramp CD ripper screen is a button that says something like "Review metadata." Click that and it will show you what each of the databases offers for metadata entries.

donmacn
2014-01-12, 04:19
Ref the CD text idea: if I'd converted a vinyl LP to CD, and possibly entered that track names into the capture software before writing the CD, would that leave a text file somehwere? That would certainly be one explanation.

It's definitely not a caching thing, because all of the first rips were done using EAC. Using DBPoweramp is new in the last three or four weeks.

Ref reviewing metadata - I do this with every disc, just to try and select options that will make my tags more consistent and check spellings etc. I just meant that it doesn't seem to specifically identify which one of those sources it's choosing as it's preferred source.

As I've said with some 300 of these vinyl copies to rip I'm just delighted it works. I can see that some things will likely cause problems - long, continuous track LPs for example (war of the worlds; Floyd stuff) but at least if the bulk of them go through easily it'll save a lot of time and typing.

Cheers

D

aubuti
2014-01-12, 13:07
Ref reviewing metadata - I do this with every disc, just to try and select options that will make my tags more consistent and check spellings etc. I just meant that it doesn't seem to specifically identify which one of those sources it's choosing as it's preferred source.
Ah, ok. I was thinking of the possibility that one or two databases would consistently be PID'ing the tracks, and the others would come up with nothing. But if you're getting metadata suggestions in all four columns, then I suppose they are all looking at the number and length of tracks.

cliveb
2014-01-13, 01:59
I've done lots of vinyl transfers, and back in the days when I wrote them to CD, I did NOT include CD text data. But upon ripping, a few of them did retrieve the correct metadata. So that's another piece of evidence to suggest that the track length fingerprint is what is used. (Think about it - lots of commercial CD releases don't have CD text on them, and they still get found).

In my experience only a small proportion of vinyl transfers got a metadata match, but I tend to only transfer obscure albums that typically aren't available on CD, so it's no surprise they aren't in any databases.