There exist at least two systems for looking up song details by a "disk-id":

• http://www.cddb.org
• http://www.freedb.org/

The problem is they suck.

Their details are limitted to artist + song. There is no flexible notion of genre.

They do not support:

• Lyrics
• Cover images.
• Albums which contain multiple disks

I wish to replace these systems with my own, which will correct these faults.

There are three problems:

1. How do you identify a given audio CD-ROM
2. How do you store the data effectively.
3. How do you allow clients to retrieve it.

I think the first one can be solved by:

request = sha1(audio Track 1 ) . sha1( audio Track 2 ) + .... sha1( audio track N )

(Where the requests are deliminated by some means. Perhaps the input will be XML?

The second is just a matter of database design:

• Table: Songs
• Table: Collections (has N songs)
• Etc

Querying against the database with a given collection of hashs should be trivial. The hard part is giving back useful output. Perhaps:

<collection>
<name="The best of the doors" >
<artist>
<name="The Doors">
</artist>
<artist>
<name="The foo orchastra" >
</artist>
<track>
<sha1="xxxxx">
<md5="xxx">
<title="Test">
<Genre="Foo">
<Length="22:22">
<lyrics>
Once upon a time
Far far away
there was a cd-rom


It was cold
It was shiny
It was silver
</lyrics>

</track>
</collection>

Thoughts welcome. Pointers to already recognised schemas especially so.