The specs posted on CNet for MP3 players are sketchy, incomplete, and do not mention key characteristics and therefore are no help at all in choosing a player.
A significant portion of users are hard of hearing in one ear or the other. As far as I know only iRiver and Cowon MP3 players have the vital, must-have, balance/pan control for these users. And I believe iRivfer sells them only in Korea. You should know this and mention this in reviews.
Simlarly, many people suffer from hearing loss in just a portion of the audio spectrum due to damage to some, but not all, cilia. The ridiculaous EQs with names like "hall," "disco," and "concert" are worthless in general but especially so for these folks. They need a graphic equalizer. And they need a graphic equalizer that can be adjusted while music is playing. (Good luck on finding one with this basic fuctctinality in an MP3 player!).
When it comes to searching, the MP3 metatdata "tags" allow Unicode so people can, and do, type anthing in any language into "artist," "title," "album," and "genre." There is no spell checker for tags. You would not believe how many ways, for example, they spell "Alanis Morissette" - that's the right number of l's, n's, s's, t's and theres an e on the end). Ditto for no copy editor enforcing consistancey. "Genres"? What is the differece, pray tell, between "rock" and "rock & roll"? And waht in god's name is "neo-punk prog"? So good luck searching the 1000s of MP3s a player can hold by tags.
Except for iPods, most users simply paste MP3s and MP3 folders into tier player's Flash memory. And most people of normal intelligence have their MP3s sorted into folders with the group's or artist's names with maybe a subfolder for each CD. Well those folders are in the players, but only a few allow you to search by folder. This should be mentioned for those that do.
The signal-to-noise ratio (SNR) specs quoted are suspicios. I think they are calculated rather than measured. Bit resolution and SNR are really the same spec: both describe the range from the softest to loudest sample. So there is a formula for going from bit resolution to SNR. This is theoretical and not achieveable in practice. And the bel is a dimensionaless ration so dB have to be "referenced" to something as in dBv or dBm. CNet's SNR specs are worthless. And speaking of specs, where are the usual audio-amplifier speca? THD+N, gain flatness, SPL, etc.?
Chuck

Chowhound
Comic Vine
GameFAQs
GameSpot
Giant Bomb
TechRepublic