The increased demand for bandwidth and power efficient timing recovery in cellular systems has triggered the need for new timing recovery methods. Classical timing recovery methods rely on transmission of training symbol sequence for the timing information. The resulting timing information synchronizes the receiver through traditional phase-locked loops. Such timing recovery methods are however, bandwidth and power inefficient. In extremely lossy channels, the exploitation of code properties to derive soft timing signals is very crucial. In this paper we present a fast converging timing recovery algorithm for cellular mobile receivers in low signal to noise ratios. In the proposed method, the receiver exploits the soft decisions computed at each turbo decoding iteration to provide reliable estimates of a soft timing signal which in turn improves the decoding time. The derived method based on sequential minimization techniques, approaches the theoretical Cramer-Rao bound with unbiased estimates within a few iterations. Though an 8-PSK baseband equivalent communication system is simulated, it was found that this model worked well with most modulation schemes. The proposed scheme is also insensitive to carrier offsets recovery. Simulation showed that the proposed method outperforms conventional timing extraction methods with respect to jitter performance.