During everyday listening, we seldom hear sounds in isolation. Whether listening to someone speak over the hum of an air conditioner, or in the bustle of traffic, the things we hear are commonly embedded in background sounds. In the previous example, the acoustic backgrounds, or contexts, have very different statistics. The sound of an air conditioner is composed of sounds that are not very loud nor very quiet; as such, its volume does not vary much over time. On the other hand, the sound of traffic can vary greatly over time, from relatively quiet periods when not many cars are present to very loud sounds such as a car horn or truck engine. In this example, the dynamic range, or contrast, of the sound of the air conditioner is low, while the contrast of the sound of traffic is high. Changes in contrast pose a unique challenge to the auditory system, which is composed of neurons with limited dynamic range. In order to consistently encode changes in volume within these two contexts, auditory neurons adapt the sensitivity of their response to compensate for the change in contrast, a process known as contrast gain control. In this thesis, we test whether contrast gain control affects the way mice hear target sounds (Chapter 2), examine the neural substrates of this behavior (Chapter 3) and explore neural mechanisms for contrast gain control (Chapter 4). We found that stimulus contrast shapes perception and neural encoding of target sounds, that auditory cortex is necessary for this process, and identified a specific inhibitory cell-type that controls gain. Taken together, this body of work demonstrates that the things we hear are shaped by efficient neural encoding of incoming information, and points towards neural mechanisms that would allow targeted manipulation of the process of perception.