I tried your method and it seemed to be working well. I made some changes which I thought you or other readers may find useful.

I often leave my music playing over night or walk away from the computer forgetting to pause it, so I found that with the default script, most of my songs had high scores, whether I liked them or not. While your method would have resiliance to this effect, it’d still happen over time. To counter it, I added the following:

sleepmax=10.to_f
# count how many tracks haven’t been skipped
sleepcount = sleepcount + 1

#see if track was skipped early
if( percentage < 99 )
sleepcount = 0
end

#average of previous score and new score, weighted according to the
#sleepcount. The longer it is since i skipped a track, the more likely
#i’m asleep so the less we’ll change the score.
guess4 = (guess3 * (sleepmax-sleepcount) + prevscore*sleepcount) / sleepmax

#only change the score if sleepcount is within limits
if( sleepcount <= sleepmax )
system( “dcop”, “amarok”, “player”, “setScoreByPath”, URI::decode( url ), guess4.to_s )
end

Of course it doesn’t stop the playcount incrementing, but i’m hoping it’s close enough.

1) close amarok
2) sqlite3 ~/.kde/share/apps/amarok/collection.db
3) type “UPDATE statistics SET percentage=0;”
4) quit and launch amaroK

This worked for me.

Setting the score to 1 isn’t what I want (I know, I’m being finicky, but indulge me!). Also, with more than 4000 songs in my collection presently, that’s a lot of waiting!

So, since all my tag info is saved with the file itself, the best thing for me to do was to nuke the db file that Amarok uses to store its data. Then, rescanning the collection effectively reinitialized the library.

But, yes, for those with smaller collections who don’t mind a small, but non-zero score, this method is perfectly okay.

One final concern, though. Messing with the track information applet often writes an ID3v1 tag in addition to an ID3v2 tag. Since I obsessively delete my v1 tags, I am scared to change anything, lest it writes in new v1 tags with their limited support for long song titles etc. Yeah, I’m that finicky about my music collection.

1. insert your songs into playlist
2. right click “edit track information”
3. go to statistics tabs, set score (don’t change anything else or you’ll break your tags!). I had to use score 1, strangely zero didn’t work for me.
4. click on save
5. wait a lot

I’ll just listen to whatever I’m in the mood for!

Rediscover your music, as Amarok says, and let the script worry about the score of the songs!

Quite a lot of tracks have a silent part at the end. Being rather impatient, I tend to just skip to the next song. Practically, the song has been played in its entirety but Amarok doesn’t know this and will assign a lower score to it on account of me having skipped those last silent 5, 7 or 10 seconds.

So I did the same thing I’ve done for the incipient section of the song, but “in reverse”, so to speak. If one has played 97% of the song, the percentage variable is artificially upped to 100 and the rest of the calculations are done with this value.

Just thought I’d mention it, so that this minuscule tweak doesn’t forever rot in the bowels of my PC forever.