Most species of bats making echolocation use Doppler-shifted frequency of ultrasonic echo pulse to measure the velocity of target. The neural circuits involved in detecting the velocity feature have been well known. Neurons in the inferior colliculus (IC) of the mustached bat have been shown to respond with discharges that are tightly phase-locked to the waveform of sinusoidally amplitude-modulated (SAM) signals. However, it remains unclear how IC neurons emerge the tightly phase-locked responses. To address this issue, we developed a neural network model of bat's auditory system for processing information of Doppler-shifted frequency. The model consists of the networks of dorsal nucleus of the lateral lemniscus (DNLL), medial superior olive (MSO), and IC. Each network has the tonotopical map in which frequency information of echo sound is represented by a linear array of neurons or columns. IC neuron was modeled with the Izhikevich neuron model. It receives an excitatory input from MSO neurons and an inhibitory input from DNLL neurons. We show here that the phase locking of IC neurons evoked by SAM signals is caused by an inherent property of IC neuron and a large noise involved in the input signal. The inherent property of IC neuron generates an oscillation of subthreshold membrane potential being resonant with SAM signal and the large noise facilitates the phase locking of IC neuron in the range of higher SAM frequencies. We also show that the offset time and strength of inhibitory input are elementary factors for causing the different response property of IC neurons elicited by SAM signals.