Advice on import of large music collection

[DavidHB]

(25-02-2015 18:54)simoncn Wrote:  

(25-02-2015 18:05)DavidHB Wrote:  The trouble, as you have surmised, is that I am not sure whether or how the software can identify your files. When ripping from CD, an album is, so to speak, a known quantity, and an identifying digital 'signature' can be calculated and compared with the CD being ripped. The chances are that your files will not conform to any known pattern, and data entry cannot be automated, or is at best only partially successful.

I don't think this is how it works. I believe most rippers look at the exact lengths of all the tracks on the CD to attempt to identify which album is being ripped. This can very occasionally lead to "false positive" matches.

I was thinking of CDDB/Gracenote, which I had cause to investigate a few years ago when I was trying (unsuccessfully) to make a Brennan JB7 work the way I wanted it to. This is how the Wikipedia article on CDDB describes how it works:

"CDDB was designed around the task of identifying entire CDs, not merely single tracks. The identification process involves creating a "discid", a sort of "fingerprint" of a CD created by performing calculations on the track duration information stored in the table-of-contents of the CD ... This discid is used with the Internet database, typically either to download track names for the whole CD or to submit track names for a newly identified CD."

MusicBrainz/Picard uses an acoustic fingerprint (Wikipedia article here) to identify CDs (and possibly individual tracks).

In the latter case, the data elements used do not appear to include track lengths at all. CDDB does use track lengths, but calculates a digital signature from them. This can certainly produce multiple matches, as any user of the Brennan device will testify.

David