ResearchBuzz

I love learning about these sites that create games that allow humans to teach computers. Last year I covered Google’s game that allows players to tag images from Google. Recently I learned about a game from UC San Diego, which has players tag/describe songs.

It’s called Listengame and it’s available at http://www.listengame.org/home.php. You’ll need to have Java enabled on your browser to play. You’ll also need to register to play (free). Once you’re registered, you can jump into a game. The games consist of rounds where you pick the best and worst word to describe a song out of a set. You can pick “Emotional content”, for example, or the best use of a song (reading, exercising, etc), or which instruments of a song are most prevalent. There’s also a “freestyle” round when you pick a word to describe a song.

You’re scored based on how many of your picks match the picks of the other people who are playing. At the end of the game (I would guess a whole game takes five or six minutes) you’re shown your score but ALSO which songs you were listening to (how about links back to eMusic or iTunes? I was getting into some of those songs.) I am weirdly good at this game.

I heard about this game by seeing a press release at the UCSD News Center that describes the game as well as where the people who came up with the game are hoping to go with it. Check out the press release for some links to scientific papers that discuss the thinking behind the game. There are new games to be released in 2007 and an e-mail address you can contact to learn when the games are released.

Fun to play, interesting to discover new music.

Looking for Internet radio stations? You can browse or do keyword searches with the new search engine iheard, at http://www.iheard.com/ .

The front of the site has a directory with the usual categories (jazz, rock, folk, oldies, talk) and a couple of unexpected (ambient, eclectic.) (There’s a much more detailed category page if you want one.) There’s also a keyword search. I did a search for one of my favorite Internet radio stations, Groove Salad, and got three results. Results include brief description, genre, a play button, and (painfully small) icons that indicate the software you’ll need to play the station (Real, WinAmp, etc.)

Click on the name of the station for a little more information about the station, but it’s not clear where the ratings are coming from and sometimes the descriptions are just too minimal.

If you don’t know what you want to listen to and can’t even think of something to search, you can always check out the most popular stations or even the stations organized by language. I just wish a bit more information was offered…

I get the impression that this site is meant to be used with Last.fm, but actually it’s a pretty good tool all by itself. Sleevenotez is in alpha at http://www.sleevenotez.com .

Sleevenotez allows you to search for musical acts and get information from several different sources on one page, including from Wikipedia, Flickr, and YouTube. And though sometimes it’s hysterically wrong, it also found me things that I hadn’t found before.

The front page will encourage you to log in, use Last.FM, etc. Ignore that. Use the artist search in the upper-right corner. I started with a search for Eurythmics. I got two matches — both Eurythmics and Eurythmics/Aretha Franklin. Choosing Eurythmics, I got a page of Eurythmics information. Starting with artist information from Wikipedia, the page also contained list of albums from the band, photos from Flickr, a discography, and videos from YouTube.

The Wikipedia article of course was right-on, and the list of videos was interesting (though by no means complete.) On the other hand the Flickr photos contained a variety of images, including a picture of a slug and a skyscape. There were also shots of Eurythmics album covers and a live appearance of the Today show, but it looks like Sleevenotez is just searching for Eurythmics as a keyword.

Professor Longhair as a search term worked both better and worse. Searching for that term found both the artist name and his real name (Henry Byrd). A listing for Professor Longhair also found the Wikipedia article, Flickr photos, YouTube, etc. Interestingly the Flickr photos were more relevant (including paintings and a shot of the sign at Tipitina’s) but the YouTube videos were all over the place. Some of them were Longhair performances, some of them were other artists performing Longhair songs, and one of them was a guy playing a Longhair song and singing the theme from the Dukes of Hazzard (this worked disturbingly well.)

Sleevenotez works best when the artist name is distinct. When it isn’t the system breaks down. Searching for The Peels found the correct discography and some interesting Flickr photos (unfortunately the Peels don’t seem to have a Wikipedia entry), but the videos were all wrong. Searching for techno artist Olive was hysterically funny — the discography and a couple of the YouTube videos were right, but the Wikipedia article was for olive-the-food, and the pictures of course were similarly-oriented. If the name of the artist you’re seeking is common, you might want to use a musician name instead.

I found this a great information aggregator; it introduced me to a lot of videos and photos I might have missed otherwise. Worth a look.