1: \begin{abstract}

2: We discuss stochastic calculus for large classes of Gaussian processes,

3: based on rough path analysis. Our key condition is a covariance measure

4: structure combined with a classical criterion due to Jain--Monrad \cite{JM83}. 

5: This condition is verified in many examples, even in absence of explicit

6: expressions for the covariance or Volterra kernels. Of special interest are

7: random Fourier series, with covariance given as Fourier series itself, and we

8: formulate conditions directly in terms of the Fourier coefficients. We also

9: establish convergence and rates of convergence in rough path metrics of

10: approximations to such random Fourier series. An application to SPDE is

11: given. Our criterion also leads to an embedding result for Cameron--Martin paths and complementary Young regularity (CYR) of the

12: Cameron--Martin space and Gaussian sample paths. CYR is known to imply Malliavin regularity and also

13: It\^{o}-like probabilistic estimates for stochastic integrals (resp.\ stochastic differential equations) 

14: despite their (rough) pathwise construction. At last, we 

15: give an application in the context of non-Markovian H\"{o}rmander theory.

16: \end{abstract}

17: