Data rate performance of the modern multicarrier techniques like generalized frequency division multiplexing (GFDM) has been significantlly limited by vital impairments like symbol time offset (STO) and carrier frequency offset (CFO). Traditional data aided methods of synchronization avoid the effect of egress noise in pilot preamble and hence destroying the spectral advantage of GFDM system. To safeguard this advantage, preamble needs to be pulse shaped. The main contribution of this paper is the derivation of generalized maximum likelihood (ML) estimation of frequency and time offsets for receiver synchronization in GFDM systems, using the modified preamble by the application of matrix inversion lemma. The dependency of the choice of the filter on Cramer-Rao lower bound (CRLB) of frequency offset estimation is also emphasized. The performance of the system is analysed over additive white Gaussian noise (AWGN) and multipath channel environments. The authors carried out real time implementation of GFDM system using IEEE 802.11 short preamble in indoor environments by employing national instruments (NI) universal software radio peripheral (USRP) 2953R boards as hardware platform which is interfaced with LABVIEW for practical validations of the results.