There are (generically) two player types for web audio, whether embedded or otherwise:
- Single stream – Players which feature a single song or a continous sequence of songs which can’t be control.
- Multi stream – Players which feature multiple songs, which can be flicked through in some way.
The former is not a particularly interesting case; it’s essentially just data (music) down a wire (connection). A reasonably solved problem, but useful for radio and simple repetitive sounds (whether using continual loops is good is not being questioned here).
The latter, however, is not as clear cut as far too many players get it wrong. All mature media players should have a feature set including playlist creation (and manipulation), selection directly from playlist, randomisation (“play a song that has not been played recently, using a random but possibly weighted function”), and continous play (“start the next song when this one finishes, regardless of how next is picked”).
This is one reason why Real player fails. It does allow for playlist playback but not selection or creation, making the feature largely impossible to use seriously.
If players for web audio want to compete seriously then they should take a lesson from established players. Or do at least one usability study. I can’t imagine anyone who thinks many players do half of what they want.