That is correct, however most if not all audio resamplers have some type of interpolation, where rather than stuffing zeros between samples it figures out an average value of the two surrounding samples and inserts that instead.

Say you have two adjacent samples, one is at -12db and the next is at -6db. If for example you're going from 48khz to 192khz, a resampler would be quadrupling the amount of samples, but rather than inserting zeros (which in the case of 24-bit audio would be -144db) it would average the values in between -12db and -6db, and your samples would be at -12, -10.5, -9, -7.5, and finally -6. This obviously would sound much better than -12, -144, -144, -144, -6.

Of course, resampling with interpolation does have some drawbacks, in that the interpolated samples do not actually represent the original audio. This can sometimes add some sort of distortion, though this is usually at frequencies we can't hear and is usually at a very low level anyway.

The upshot of upsampling audio is that there's usually less filtering required on playback, since the Nyquist frequency is much higher than before.

I work here