Status

This documents does not yet contain Kåre's notes and needs general editing.

Event

A two hour tour at the CD-ripping department for Kåre and Toke, performed by Christian Jensen (cj) and ???, jan. 3, 2006.

Results

Workflow

The files from the rips are stored on clio and transfered to permanent storage during the night. The files are named by the unique number assigned to each entry.

When a CD has passed the rippers, they have validated the metadata for it.

DOMS specific

Tracklists are not strictly standardized. There are variations like "3 blind mice" vs. "Three blind mice" and the list might be out of order. The MARC-record aren't always correct. About 25% of the ripped CDs are fully matched with metadata.

Miscellaneous

DOBBIN from Quadriga / Cube-Tec is a complete system for ripping, transcoding and presentation of CDs. The department consider buying a license.

A rip should be bit-perfect. In case of errors, a new rip is performed. In case of three errors, the CD is marked as defective and the best rip is stored. There are bit-errors on about 10% of all rips.

Sound older than 50 years is sold by Statsbiblioteket on CDs.

All music- and sound-CDs with relation to Denmark is ripped. That includes CDs with danish artists, compilations with a single danish artist on one of the songs, CDs pressed in Denmark and so on, CDs from danish magazines and so on. DBC is responsible for keeping track of this and maintains the DMO-samling.

Monografi at Statsbiblioteket handles collection (?) of CDs and creation of meta-data for the CDs. They create entries in Horizon and ships CDs down for ripping in batches of 50. There is 1500-2000 entries each year.

Horizon-entries for new CDs are copied to the CD-ripping system once a week. Coordination is somewhat lacking with regular minor errors. It if unknown if corrections to the data are merged back into Horizon or ultimately DanBib.

The ripping process unearths a fair amount of error or omissions to the metadata. Hidden tracks and the number of CDs compromizing an entry are examples of this.

Entries are identified by a number, which is typically represented by a barcode on the cover. Some CDs do not have covers.

There is no real backlog of CDs.

Some of the CDs received are home-burned. They do represent a challenge, as they are often hard to read.

After ripping, the CDs are shipped for scanning. Only the cover is scanned. This job is typically handled by a student worker.

The delay from the creation of a CD to rip is about ½ year. Worst-case is about 1½ year.

It is possible for users of Statsbiblioteket to borrow a CD for local listening, as soon as it is registered in Horizon. This intrudes on the ripping process, as the current system isn't that good at postponing ripping. It is handled by ripping a Creedence Clearwater Revival CD, which is without error. The ripped tracks are then removed from the system, prompting a re-rip. The borrowing of a CD might take up to a week.

This disturbance happens 5-10 times a week and is a lot of work, since the CD is placed at different physical locations, during registratipn, ripping and scanning.

CDs with scratches or copy-prevention systems takes up to 45 minutes to rip.

Errorreporting and handling is generally done using human contact, instead of a formalized computerbased system.

Small CDs aren't ripped p.t., but the next ripper wil be capale of such.

The current system has a 105 CD changer (7*15), but will be replaced by 4 standard CD-ROM drives.

Datatracks on the CDs aren't stored.

Bootlegs aren't ripped.

CDRipping 20060103 (last edited 2010-03-17 13:08:52 by localhost)