I'm wondering just how confident I can be when running ENMFP data through song identify. I run a 60-second sample using codegen.exe, 30 seconds into the song, and send the resulting data to http://developer.echonest.com/api/v4/song/identify
Here's one of the more extreme examples. For this particular song, my database shows: Augustus Pablo / The Definitive Augustus Pablo / Track 14 / Unity Dub (feat. Rockers All Stars) But Echo Nest shows the artist as "Hugh Mandell" and the song title as "Levi Dub"
The results from the Echo Nest query shows this as a Match Type 8 - "There were multiple matches and the first one was so strong we didn't need to look closer"
Another, more mainstream example: I have David Bowie's album Sound + Vision II with track 11 being "Fascination (Ryko Mix)" yet Echo Nest returns "03 David Bowie - Day In Day Out" This is also a match type 8.
Is this sort of thing expected with Match Type 8s? Is there anything I can do to get more accurate results?