but there is no mention of it on their official site, the artwork is only on last.fm, and no release events
2010-03-01 06049, 2010
dankine
'it was an independent release. It leaked onto the internet via blogs and Bit Torrent in 2007, but there was such minimal online success, the band took no further action. There is still no word on whether or not actual copies of the EP were ever made due to the minimal demand for the band's music.
2010-03-01 06009, 2010
dankine
so it kinda sounds like it was going to a release, then leaked, then flopped
2010-03-01 06010, 2010
dankine
so they didnt bother
2010-03-01 06021, 2010
Munger
So that is basically a bootleg
2010-03-01 06033, 2010
dankine
well, i'd say not but i don't really know
2010-03-01 06001, 2010
Munger
Well, it's currently labelled as official, which doesn't sound like the case
2010-03-01 06050, 2010
navap
If the artist was going to release it, but didn't because there was a leaked copy going round, then I think the release in MB should be marked as bootleg.
2010-03-01 06031, 2010
Munger
I'll change it and add that wiki link
2010-03-01 06033, 2010
murdos joined the channel
2010-03-01 06023, 2010
Munger
I guess I *could* add a 2007 worldwide digital media release event :-)
2010-03-01 06059, 2010
pronik_tp joined the channel
2010-03-01 06020, 2010
Munger
...or are REs strictly for official releases?
2010-03-01 06006, 2010
navap
It can be used for bootlegs as well. If you have evidence to support it being released/leaked in 2007 then go for it.
2010-03-01 06036, 2010
Munger
I'm just going by the wiki page, which I think is as substantial as we will ever get for this
alastairp, Wow! Thanks. That sounds eerily like what I was talking about the other night :-)
2010-03-01 06037, 2010
luks
you really haven't seen that before?
2010-03-01 06001, 2010
Munger
No. That is not analyzed by computer
2010-03-01 06003, 2010
luks
(I'm surprised because I though Pandora is pretty well-known)
2010-03-01 06012, 2010
luks
*thought
2010-03-01 06045, 2010
Munger
I heard of Pandora, but never looked into it. Never heard of the Genome project per-se
2010-03-01 06025, 2010
alastairp
Munger: yes, I did notice they classify manually, which sucks
2010-03-01 06003, 2010
Munger
It's probably more accurate than year 2000 technology
2010-03-01 06010, 2010
alastairp
there's been some talk recently on the Music-IR list about automatically detecting live recordings
2010-03-01 06034, 2010
alastairp
which is one vector that could be considered
2010-03-01 06012, 2010
Munger
I generated some fake pseudo-random data for 10000 'songs', weighting some imaginary characteristics to simulate 10 different musical styles. After shuffling them all into a random mess I ran them through my comparison algorithms. I did see the kind of clustering I expected
2010-03-01 06048, 2010
v6lur joined the channel
2010-03-01 06032, 2010
alastairp
nice
2010-03-01 06022, 2010
Munger
Of course. because I generated the data myself it's not surprising, but it was nce to see a visualisation of the sort in progress
2010-03-01 06042, 2010
Munger
Determining the second pass key songs from the centre of each cluster proved trivial, but I'm sure there is a more efficient way than my approach
2010-03-01 06043, 2010
Munger
The really hard bit will be finding sufficient deterministic methods by which to make real-life comparisons possible
2010-03-01 06040, 2010
luks
you are working on something to automatically classify music genres?
2010-03-01 06001, 2010
Munger
alastairp, I wonder how many of these attributes could be analyzed
2010-03-01 06004, 2010
brianfreud_ joined the channel
2010-03-01 06008, 2010
Munger
luks, Not genre, but certainly use relationships between songs to generate proper smart playlists
2010-03-01 06032, 2010
luks
so basically "unnamed genres"
2010-03-01 06042, 2010
Munger
luks, Genre is such a subjective term.
2010-03-01 06009, 2010
luks
anyway, it sounds interesting
2010-03-01 06017, 2010
Munger
Sort of, yes. Like a kind of musical star map
2010-03-01 06028, 2010
alastairp
Munger: sounds like many people are working on it at the moment :)
2010-03-01 06032, 2010
alastairp
but I'm unsure
2010-03-01 06047, 2010
Munger
My idea was to take three random key songs as a starting pont. Distribute them arond the circumference of a circle and position everything else at the centre. Then compare each song in turn to each of the key songs and move them on a vector towrds that key depending on how similar they are in each characteristic. Hopefully, you will get clusters of songs forming at various locations inside your circle. Take the centre of each cluster and make the
2010-03-01 06048, 2010
Munger
se the new key songs, returning the rest to the centre, Rinse and repeat
2010-03-01 06025, 2010
Munger
luks, I never take into account data such as performer, album or anything not directly determined from the music itself. Similar songs should gravitate towards each other. Your playlists can be derived from starting at any song and meandering off in an arbitrary direction
2010-03-01 06017, 2010
Munger
The results of course will vary from collection to collection, which IMHO is what a true smart playlist should do
2010-03-01 06059, 2010
Munger
It's still very much an embryonic concept (for me at least) so I welcome any bombshells :-)
2010-03-01 06050, 2010
luks
I haven't done anything in that area, so I can't really comment
2010-03-01 06018, 2010
Munger
I need to do a lot of research in the analysis part tyo see what is really possible, and how much is just a pipe dream. I'm reading up on speech/language recognition to see if that field has anything to offer. Not so much for lyrical analysis (although that has not been lost on me), but for detection of certain musical phrases
