Due to the unknown nature of the mechanical loads, proper tuning of the speed controllers is always a serious challenge in servo motor drives. The problem becomes more complicated when the speed signal feedback is received from the conventional sensors such as incremental encoders which are prone to the quantisation erros. This could basically result in creation of unwanted noise during the closed-loop speed regulation process. To mitigate the problem, this paper proposes a superior offline tuning method for the speed controllers as compared with the conventional approaches. The bandwidth is calculated in a way that the fluctuation of the reference torque is limited to a given percentage of the motor-rated torque, which leads to lower acoustic noise. In addition, an offline detection algorithm is proposed to estimate the mechanical features of the load. In this method, two new auxiliary mechanical parameters are defined for the motor-load set and estimated using a computationally efficient approach. The estimated mechanical features as well as the bandwidth value are used to calculate optimal gains for the speed controller.