A system and method for regulating media volume is disclosed. The system comprises a prediction engine and a regulation module. The prediction engine collects sensor data and scene category data. The scene category data includes data describing an environment in which the media volume is regulated. The prediction engine predicts a workload for a user based at least in part on the scene category data and the sensor data and generates a predicted workload value for the predicted workload. The regulation module adjusts the media volume based at least in part on the predicted workload value.