I am not absolutely certain what the problem is, because ‘does not work’ can mean just about anything.
Note that according to the documentation, the first argument ‘x’, the time-domain signal must be ‘specified as a real-valued vector, or a single-variable timetable with a single column. If x is a timetable, x must contain increasing, finite row times.’
MATLAB sound files characteristically contain 2 columns, corresponding to the left and right channels of a stereopohonic recording. If that is the situation with your ‘X’, choose one column or the other, not both, to present to the emd function.