This paper presents the design and realization of a gain-programmable microphone audio preamplifier that does not need external input-decoupling capacitors. Three possible preamplifier implementations with on-chip decoupling capacitors have been presented and compared in terms of audio quality performance, power consumption, and silicon area occupation. The chosen solution has been integrated in a 40-nm digital CMOS process with seven metal layers and a double gate-oxide option. Process fringe capacitors have been used to implement the internal decoupling, amplifying, and dc biasing functions. The active area is 0.19 mm2, whereas the power consumption is 330 μ W at a 1.5-V power supply. Measured dynamic range (DR) and maximum signal-to-noise-and-distortion ratio (SNDR) of approximately 105 dB A-weighted and 101 dB A-weighted, respectively, have been achieved for 0-dB gain and balanced inputs. The preamplifier can process both balanced and single-ended inputs, and its gain can be programmed from 0 to 19.5 dB in 1.5-dB/steps with acceptable performance losses.