« Webjay Closing June 2007 | Main | Peel Makes Managing Music Blogs Easy »

Audiobaba: Trait-Based Recommendations

audiobaba_logo.gifAudiobaba is a simple and elegant music search and recommendation engine. Most recommendation engines provide song matches based on related artist data (artist X is similar to artist Y) and collaborative filtering (people who listen to X listen to Y). Audiobaba takes a different approach. Every song in the Audiobaba database is “fingerprinted” with acoustic characteristics and other special-sauce qualities. When you search for a song on Audiobaba, not only do you get the song you are looking for (provided it’s in their database), but also songs that sound similar based on a “fingerprint” match. This is great for both discovering new music and helping you create playlists that flow together based on similar qualities.

Just type the name of a song or artist into the search box and Audiobaba returns an alphabetized list of individual songs along with their respective 30-second samples (if available). When you find the song you are looking for, click the “match” button and Audiobaba returns a list of similar songs.  You can filter the results on a sliding scale from mainstream to independents .

audiobaba_artist.jpg

To test the system I tried to find matches for Charles Mingus’ very unique sounding “Boogie Stop Shuffle”.  Audiobaba returned many songs from the jazz genre that indeed matched many qualities such as instrumentation, tempo, style and mood. A few other searches yielded similar successes while others were mixed.

audiobaba_matches.jpg

If you feel a song isn’t related, you can click the “Thumbs Down” button to help Audiobaba refine their results. They should really change the name of this button since “thumbs down” really implies that I don’t like a song. Perhaps allowing user to use a sliding scale from Great Match to Poor Match would be better as it allows for positive as well as negative user feedback.

A cool feature is the RSS Songfeed. When you find a song you like, you can subscribe to it’s RSS feed. When new songs show up that match your song, they will automatically appear in the feed. Audiobaba also allows you to bookmark tracks. However, since there is no login for the site, if you switch computers or browsers, you wont be able to access your bookmarks. Many of the independent tracks are provided as free downloads and they also provide links to purchase songs on Amazon, Buy.com and iTunes.

Audiobaba is still in Beta but needs to make some improvements to the search functionality and UI. Simple misspellings will return no results and the system does not provide a Google-style “Did you mean X?” feature. While I do like the idea of universal search, they need to provide a better way of filtering the results by songs and artists. For instance, if I search for “The Doors”, I need skip ahead many pages since results are provided in alphabetic song format. While this results format makes sense for all songs by a selected artist, its a real pain when just searching for a specific artist. There are also some funky AJAX problems such as strange linking and back button behaviors.

Despite its faults, I found myself having lots of fun playing around with Audiobaba. Their unique approach to music discovery, coupled with a simple UI creates an addicting user experience.

PrintView Printer Friendly Version

EmailEmail Article to Friend

Reader Comments (2)

Personally, I'm not a big fan of trait-based recommendations - MusicIP/Predixis also plays in this space. Even if I do subsconciously like songs that sound-alike... I don't want to admit to myself that I am that predictable or shallow. :-)

05.23.2007 | Unregistered CommenterJ Herskwoitz

Usually some of the music members use to download the latest mp3 and pop albums always,might it would help for those kinds of people.
*************
Gomez
Social Bookmarking

11.27.2008 | Unregistered Commentergomez

PostPost a New Comment

Enter your information below to add a new comment.

My response is on my own website »
Author Email (optional):
Author URL (optional):
Post:
 
Some HTML allowed: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <code> <em> <i> <strike> <strong>