A vehicle audio system such as a vehicle radio is integrated with a vehicle speech system so that outputted speech by the vehicle speech system is played through the audio system. When a speech session begins, the audio system mutes any entertainment audio being played and outputs audio speech received from the vehicle speech system at an initial volume that is constrained to be between minimum and maximum initial volume settings. These initial volume settings are automatically adjusted by the audio system if it detects that the user has a tendency to repeatedly adjust the volume past one of these settings each time a speech session is carried out.