Moodagent: Going iTunes Playlists one better

It certainly is annoying how Apple limits their hardware. There is a certain aspect of control which is a positive for achieving stability, but beyond that it is just a limiting of potential. I think, for Apple, the main reason they limit control excessively is because they really want to sell you something else, in addition to what you have, as much and as often as possible. As for me, I'm not the type of person you could sell a 4GB RAM upgrade for the price of 24GB of RAM that's actually 50% faster. I tried many days in the online store, but I just couldn't stomach it. However, I like the stability, and can at least use the respective going AppleCare prices to gauge the relative quality of a specific Apple product since Apple has $35 billion in the bank and the price for coverage seems to mysteriously go up for a particular item when there are widely publicized quality issues with it. They should seriously considering making products in Japan where excellence in quality is taken seriously and personally instead of cheap China which secretly loves less principled capitalism.

Have you considered writing an AI component for the player? Then at least you wouldn't need to maintain a server to access a record for each song.

Seems the Moodagent servers are under pressure, or there's something seriously wrong with my music collection.
The iPhone application crashes on my iPhone after syncing about 100-odd songs. I've reloaded my whole library playlist by playlist, and I've gotten about 800 songs synced, with over 100 being unrecognised. The profiled songs generate playlists that are...unusual.

On to the Moodagent profiler. I've spent the whole day trying to profile my 35k-song library, and the application complains about a server error after every thousand or so.I've only managed to upload about 3500 profiles so far

I'm always excited at the potential these type of (movie/music/usage-based) profiling applications, and offer my submissions to the hive-mind willingly, but I'm inevitably disappointed with the results. The most success I've had with one of these types of systems is the MusicBrainz tagging platform.

When using the sliders all positions are taken equally into account, setting one slider to a low value does not indicate a low priority, it is just as important as turning another slider up. Thus turning four sliders down will actually influence the result more than turning a single slider up. The result you get is always the closest match to the combination of all five slider values, in any given collection.

After using the sliders you can always click the Moodagent icon next to the seed to get the actual values of the seed track, skipping from the slider positions set by the user.

Why is Sledgehammer hard to understand under "Sensual"? The song is a not very subtle reference to sex.

Info: 308 of 874 tracks have no matching Mood Profile.

Seems a bit incomplete...

Music dictionaries generally give ranges of meaning for tempo terminology but define tempo itself as a quantifiable parameter measured in beats per minute. (That's how metronomes are calibrated.)

In the data set examined at this site

http://www.grunin.com/eroica/index.htm

all the data points represent performances of a piece whose tempo is marked (by Beethoven) "Allegro con brio," yet the measured tempo ranges between 110 and 176. If "tempo" doesn't mean "beats per minute," doesn't it just mean "how I feel about this piece"--which may not be the way anyone else feels?

The Moodagent application uses 4 mood sliders and 1 tempo slider, however there's much more behind the scenes.

The technology behind Moodagent creates profiles for every song it encounters and stores these profiles in the cloud. Over 7 million profiles have been created so far and 10 million profiles should exist soon. To create a profile we analyze the digital signal of a song, similar to Pandora, we then take a segment (or segments) of the song, amplify them and run it through a series of artificial intelligence expert systems, constantly trained by musicologists. We are able to create a song profile which measures the degree of each of the song's characteristics, including moods, genres, sub-genres, styles, tempo/beat, vocals, instrumentation and production features. Even Beethoven may have some degree of "rock".

If you select a song as the "seed" track it will look at all of the characteristics of the song to build the playlist, not just the mood profiles.

To get the most out of Moodagent for your iPhone you can use the Moodagent profiler which is available at http://www.moodagent.com/profiler. It will look at your iTunes collection on the desktop and create profiles for any songs where the profile doesn't exist. This way, when you sync Moodagent on the iPhone, you'll get all of your non-DRM tracks and have a better experience creating playlists.

Aren't four of those categories are too subjective to be useful? The only one with an unambiguous meaning is "tempo," if it's expressed in beats per minute. (If it's "fast" or "medium" or "Allegro moderato" we're back in subjective-land again.)

I would love to have a mood-based music categorization system built into iTunes. That's how I listen to music after all.