In recent years, the study of the (ir)regularity of various stochastic processes has attracted the attention of several authors. In this work, we aim to study a large class of Gaussian fields defined via a stochastic integral. More precisely, we seek to obtain a general strategy to deduce sharp regularity and irregularity estimates using wavelet methods. This approach should allow us to recover and compare regularity results appearing in the literature within a unified framework. Key ideas will