sounds kinda like what routee 66 has in its MySql database... Not a bad idea for windows users
Hey, i was sitting around working on my AI research project when I came up with a really cool although perhaps useless plugin idea. The jist of the idea is that there is a program that remembers what songs you have told it you liked in the past, and builds a personal preference for certain songs, and instead of just picking random songs to play, the plugin would tell winamp to pick songs that you like that havent been played in a while. Then, there could be several different saved profiles if you have multiple people driving the car, or if you have different passengers in the car (e.g. the girlfriend profile, the guy down the streets profile, your parents profile, etc). I'm not sure how someone would change profiles, but that couldnt be too hard. What do people think?
- Experience varies directly with equipment broken ---------- Kevmo
sounds kinda like what routee 66 has in its MySql database... Not a bad idea for windows users
Car
1964 Ford Falcon powered by a 1993 Mustang 5.0 EFI, AOD transmission, with disc brakes up front and more mechanical goodies
The Player
Gone! Soon goign to repalce the screen based system i HAD with an iPod. http://mp3falcon.crazyshaxs.com
Whats route 66?
- Experience varies directly with equipment broken ---------- Kevmo
I've been thinking about something like that as well. Basically, each song would get a rating between 1 and 4. This rating would depend on how many times, you specifically selected that song, how many times it was skipped, how many times it was played, etc. Then, in "Favourites Random" mode, the player would show favor to those songs with a higher ranking.
It's really not that hard to program, but I want to avoid database writes on the road (bad thing if it gets corrupted).
Player: Pentium 166MMX, Amptron 598LMR MB w/onboard Sound, Video, LAN, 10.2 Gig Fujitsu Laptop HD, Arise 865 DC-DC Converter, Lexan Case, Custom Software w/Voice Interface, MS Access Based Playlists
Car: 1986 Mazda RX-7 Turbo (highly modded), 1978 RX-7 Beater (Dead, parting out), 2001 Honda Insight
"If one more body-kitted, cut-spring-lowered, farty-exhausted Civic revs on me at an intersection, I swear I'm going to get out of my car and cram their ridiculous double-decker aluminium wing firmly up their rump."
I've written this already. I used a neural net and mapped it to the archive of songs on the computer. Some behavior causes a song to become "better" (like individually selecting it or listening to it all the way through) and some behavior causes a song to become worse (like skipping it).
I've run into a few asthetic issues that I want to resolve before I release the software:
1) Should you have a seperate command to skip tracks that does not update the song's rating? This would be useful for skipping thought long playlists, but might be abused and throw off the learning algorithm.
2) Currently, I sort songs by their rating when you add them to the playlist. This way, for favorite songs get played first. However, should I do this for pre-defined playlists? I might implement this as an option, but I don't want to add an option that nobody will use.
3) I also have a Build Playlist feature that builds a playlist based on what songs you like. Right now it just adds all the songs you have to the playlist in order of best to worst. Should there be a cutoff (Only songs you really like are added?) or is one long playlist acceptable?
Any input would be appreciated.
Bent
for the plylist put only songs that have a higher ranking.
Car
1964 Ford Falcon powered by a 1993 Mustang 5.0 EFI, AOD transmission, with disc brakes up front and more mechanical goodies
The Player
Gone! Soon goign to repalce the screen based system i HAD with an iPod. http://mp3falcon.crazyshaxs.com
Higher than what? A song that has no ranking (0.0)* or higher than some user defined value?Originally posted by crazyshaxs:
<STRONG>for the plylist put only songs that have a higher ranking.</STRONG>
*(the song ranking is a floating point value from -1 to 1)
isnt a neural net kind of overkill for this application? how fast does it run?
- Experience varies directly with equipment broken ---------- Kevmo
Bookmarks