The "phase" is just the name we give to the argument of the sinusoidal wave function.
For example if one wave is described by:
$$
f(x,t) = A_1\sin(k_1 x - \omega_1 t + \delta_1)\tag 1
$$
the "phase" of this wave is:
$$
\eta_1 = k_1 x - \omega_1 t + \delta_1
$$
For example if another wave is described by:
$$
g(x,t) = A_2\sin(k_2 x - \omega_2 t + \delta_2) \tag 2
$$
the "phase" of this wave is:
$$
\eta_2 = k_2 x - \omega_2 t + \delta_2
$$
The "phase difference" at the same point in space and time between the two above-mentioned waves of this example is:
$$
\Delta\eta_{1,2} = (k_2 - k_1)x + (\omega_2 - \omega_1)t + (\delta_2 - \delta_1)
$$
The phase difference doesn't have to refer to different wave forms. It could instead refer to a single wave form, but at different points in space, or for different path lengths travelled by the wave.
For example, if the wave $f(x,t)$ of Eq. (1) is split (e.g., with a prism) and one component of the wave travels along a path of length $L$ and another component of the wave travels along a path of length $\ell$ and then the waves recombine, we are interested in the phase difference:
$$
\Delta \eta = k_1(L - \ell)
$$