Speckle noise is an undesirable granular pattern generated by coherent multiple-scattered photons in Optical Coherent Tomography (OCT) images. It impedes accurate OCT image analysis in different clinical diagnostic interpretation tasks. To solve, numerous speckle removal filters have been introduced in the literature. However, many popular despeckling filters for enhancing OCT images often result in degrading image quality and introducing artifacts and blurring details. Also, the conventional OCT despecking filters require manual adjustments. In contrast, the modern data-driven approaches have a superior performance in improving image quality through automatic feature learning and preserving details of the natural image. Recently, deep learning (DL) has emerged as a promising data-driven solution to many inverse problems such as image enhancement and restoration. In this paper, we propose a DL-based OCT despeckling method that adapts the single-stage blind real image denoising network (RIDNet). Our experiments utilize Shepp-Logan phantom images and images from publicly available datasets such as BSDS 500 and Open ICPSR dataset. To validate, we compare the results obtained using the proposed method against those of existing approaches using both qualitative and quantitative evaluations.