Measurement of the delay is an important and common problem in communication over packet networks. The endto-end and the round trip delay are among the factors directly impacting the quality of service as well as the user satisfaction. Multimedia gateways or base stations that perform echo cancellation or suppression often rely on the round trip delay to enhance their performance or to reduce the computational complexity of echo processing logics. In this work, we present two none intrusive methods for delay estimation and tracking. Both methods find the delay using the actual audio signal that is sent through the network. The first approach uses the MDCT transformed domain coefficients of the signal while the second operates in a perceptual domain. Experiments illustrate that both schemes can track the end-to-end and the round trip delay under various network and signal conditions.