Electric vehicles offer cleaner transportation with lower emissions, thus their increased popularity. Although, electric powertrains contribute to quieter vehicles, the shift from internal combustion engines to electric powertrains presents new Noise, Vibration, and Harshness challenges. Unlike traditional engines, electric powertrains produce distinctive tonal noise, notably from motor whistles and gear whine. These tonal components have frequency content, sometimes above 10 kHz. Furthermore, the housing of the powertrain is the interface between the excitation from the driveline via the bearings and the radiated noise (NVH). Acoustic features of the radiated noise can be predicted by utilising the transmitted forces from the bearings. Due to tonal components at higher frequencies and dense modal content, full flexible multibody dynamics simulations are computationally expensive. Based on previously developed metrics for sound quality, a methodology is proposed with the requirements to design a reduced order model, coupling analytical solutions for the electrical motor along with lumped parameter powertrain modelling. Namely, for an adequate representation of the tonal content, the proposed model considers the harmonic excitations of the electric motor and torque ripple, time-varying stiffness and backlash for the meshing gears and flexibility of the housing at the bearing connection points. The reduced order model predicts the transmitted bearing forces, enabling an overview of the sound quality of the radiated noise with lower computational time and complexity.