[pdb-discuss] Something concrete we could do in the next week

Fri Mar 16 10:03:03 UTC 2007

Ian Ibbotson wrote:
>> Exactly though remember that to be truly PD track must also be more than 
>> 50 years old -- which is why i don't think we will find many pd tracks 
>> in musicbrainz at present even though we mind find quite a few composers 
>> who died more than 70 years ago. For out of copyright recordings we are 
>> really going to start needing to upload/analyse the BBC 'grams' (i.e. 
>> gramaphone) data.
> 
> Don't want to push noise onto the list when everyone is busy preparing
> for saturday, but Just for fun...... 

Not at all -- this is *really* useful.

> the musicbrainz query of this form:
> http://musicbrainz.org/ws/1/release/?type=xml&query=date:[19570101%20TO%
> 2019580101]

Thanks for this tip. the musicbrainz python bindings do not currently 
have support for dates but using this and a bit of hacking I've written 
a simply bit of python to pull back sets of releases before particular 
dates, see find_releases_before(date='19570101') method in:

http://project.knowledgeforge.net/pdw/svn/trunk/src/pdw/mb.py

> Reports 
> 106 releases between 1957 -1958, 
> 103 releases between 56-57
> 59 releases between 55-56
> 32 release between 54-55
> 23 for 53-54
> 15 for 52-53
> 14 for 51-52
> 9 for 50-51
> 7 for 49-50
> 
> 35 for 1939 to 1949
> 36 for 1919 to 1939
> 7 for 1800 to 1919
> None for 1700-1800
 >
> Of course this isn't then cross correlated with composer information,
> but it gives a bit more info about the depth of the MB data I suppose. I
> guess one of the nice things about the MB data is that it's real easy to
> create a datafeed of works that might be candidates for entering the PD.

This is fantastic Ian -- as you say the next step is to start cross 
correlating against the composer dates. I've started formalizing 
copyright status calculations in:

http://project.knowledgeforge.net/pdw/svn/trunk/src/pdw/copyright.py

Regards,

Rufus