Time Division Multiplexing over Internet Protocol (TDMoIP) is a pseudowire technology for emulating TDM circuits over Internet Protocol (IP) networks. In such networks, timing and synchronization plays a key role in achieving the required jitter in terms of variance of interdeparture interval. A jitter buffer is used at the receiver, to circumvent the impairment of the packet networks: delay, jitter and loss. But, out of these, delay and loss can't be compensated for, while QoS in IP networks is used to minimize them. The jitter (or variance of packet delay) can be reduced to a tolerable level at the receiving Inter Working Function. A tradeoff between delay and jitter is required to achieve the desired jitter. This paper presents the condition under which the jitter buffer at the receiver is to be operated for minimum output variance in a TDMoIP framework, to achieve minimum slip rate and thus better voice quality. The receiver jitter buffer is modeled as a correlated M/G/1 queueing system with EARMA correlations between the interarrival and the service times. The motivation for the above correlation structure is that, given the correlations within the service intervals, the EARMA correlation results in reduction of variance in the interdeparture interval. This is a step towards achieving CBR upstream. The key advantage of using EARMA correlation is that the analysis of such a correlated queue is analytically tractable. The variance of the interdeparture times of the above queue is presented. The analysis of the departure process, the waiting times of incoming packets of this correlated queue and the relevant simulations show that if the variance of the interdeparture time process constituting output TDM stream is to be less than that of the interarrival time process of the jitter buffer, which is modeled as M/G/1 queue, then the mean waiting time of the packets in the jitter buffer would be greater than that of independent (M/M/1) case. The values of the parameters of the M/G/1 queue which minimizes the variance of interdeparture interval are identified. Our study also included a G/G/1 queue in which interarrivals are also correlated. Extensive simulations demonstrate our analytical results. Copyright © IARIA, 2012.