Bandcamp as source of AcousticBrainz data

AcousticBrainz
, , , ,
41

Hierarchy is complicated, merge all rock-n-rolls to one, find other tags like this, slit taglines is easy and can be done in semiautomatic way

1 Like
42

This is the reason why it seems time to make a decision about moving from folksonomy tags and genres to a structured taxonomy.

1 Like
43

700106 tags submitted!

7 Likes
44

if someone begins to do this for spotify let us know. Maybe I will try.

3 Likes
45

I thought about it. I think Spotify not so tolerant to downloading as Bandcamp, so it can be harder and slower. But maybe subset of releases without AB data, without BC link and with Spotify link not so big, then speed not a big problem.
Will be awesome if you try to do it

1 Like
46

great project, i also thought about how useful the 128kbit streams could be for AB!

did not see it being brought up, but especially if you start to download the lossless files constantly, Bandcamp might notice a regular increase in traffic. maybe you want to contact them before. or maybe i am too cautious.

1 Like
47

Spotify should use Tagtraum genre annotations, which should already be in the AcousticBrainz Genre Dataset: I don’t know who is working on it now.

1 Like
48

If you don’t ask permission you can’t be refused :wink:

Anyway average speed i can get from BC is 50 mbit/s, i don’t think it is notable «increase in traffic» for BC. And i constantly download FLACs for last week (and MP3 for two months) or so using only two (static) IP addresses. If BC don’t like my activity they can easily block me

1 Like
49

One million tags submitted. 150k Bandcamp pages downloaded and parsed, about 200k pages wait to be downloaded, it will took at least 30 days

6 Likes
50

Since last post i finished big Bandcamp downloading, now i have on disk 240k Bandcamp pages (10 Gb). Tags from downloaded pages (+700k tags) submitted to MB. There is another useful information in these pages (see above), but i really don’t want write bot to submit it to MB. If someone want use this information i can share pages, metainformation and some parsed information from these pages.

Also submit to AcousticBrainz and AcoustID 5k releases, another 10k releases in progress.

Hope to the and of year number of unique recording on AB will exceed 7 millions

12 Likes
51

My CPU busy now, so i start submitting to AcoustID releases that have AB data but don’t have AcoustID fingerprints. It’s much easier to CPU i can do it much faster then AcoustID+AB submitting. There is 142k such releases

4 Likes
52

138k releases was submitted

8 Likes
53

Done. There is 7,000,025 unique records in AcousticBrainz

15 Likes
54

so, I already asked elsewhere, but I figured I’d ask here too


is there any chance of re-running the Bandcamp tag import portion of this project on newly added Bandcamp releases?

1 Like