Amplitude contingent selective adaptation to speech