The advent of ubiquitous mobile communication has posed a lot of challenges, one of the prominent being suppression of background noise, especially in non-stationary noisy environments. In literature, several speech enhancement techniques have been proposed to tackle this problem of noise reduction. Codebook-based speech enhancement (CBSE) employing trained speech and noise codebooks, is one of the most effective noise reduction technique for handling non-stationary noise. However, the high compute intensive nature of this technique renders it inapplicable in real-time speech enhancement scenarios by introducing a significant delay in speech transmission. In this paper, this problem is addressed by providing an efficient, parallel CBSE algorithm. The proposed parallel CBSE algorithm achieves significant speedup and reduced execution time, resulting in a speech transmission delay which is well within the limits of realizing real-time speech enhancement. The proposed parallel CBSE algorithm is then used as a basis to provide a novel cloud based framework to achieve real-time speech enhancement in mobile communication as a proof-ofconcept. The proposed parallel implementation can also be used in a variety of applications which demand real-time speech enhancement such as teleconferencing systems, digital hearing aid devices and speech recognition systems.