An audio preamplifier (10) based on an operational transconductance amplifier, in combination with a class D audio output amplifier (12) is disclosed. The input signal is coupled to the preamplifier (10) through a capacitor (14) in series with a resistor (17) that sets the transconductance of the preamplifier. The preamplifier (10) includes a differential operational amplifier (20) that drives output MOS transistors (22a, 22b), which are biased by current sources (24a, 26a; 24b, 26b). Feedback from the drain nodes of the output MOS transistors (22a, 22b) to the inputs of the differential operational amplifier (20), along with the series capacitor (14) and resistor (17) input coupling, ensures minimum offset voltage and current at the output of the preamplifier (10). Common mode feedback control amplifiers (25, 29) ensure proper bias of the components into the saturation region.