Modern technologies such as binaural rendering rely on providing specificallyprocessed audio signals to the listener. However, most of the time theseprecisely calculated signals are played over headphones, under the assumptionthat the transmission through the ear canal would not significantly modifythem. While this is true for low frequencies, it most certainly is not for higherones, especially when using closed in-ear earphones. Although it is impracticalto include the ear canal transmission in a binaural model or measured HRTFs,as it depends on the model and fit of the headphones among other things,modelling it at the time of rendering is possible. The model would have to bespecific to the individual as well as the headphones/earphones used, and itsparameters would have to be determined using data that can be realisticallyobtained just before or during a listening session, without invasive and/orcomplicated procedures.A method to build such a model is proposed, using measurements fromsensors that could be built into an earphone. The method is tested on an earsimulator in laboratory conditions using a custom setup.