The overall performance and power utilization of an audio device may be improved with an adaptive sidetone generation system that generates sidetones optimized for different application-specific problems. In particular, systems that include sidetone generation capabilities may be developed to include numerous microphones from which information may be received and processed to generate optimized sidetones. For example, the information from the microphones may be used to receive and/or determine the audio device's operating mode. The information from the microphones and the received and/or determined mode may then be used to generate a sidetone that is optimized for the particular mode and particular conditions in which the audio device is operating. Through the generation of optimized sidetones, the audio signal quality may be improved, thus reducing the amount of subsequent audio processing required, and resulting in improved performance and power utilization.