The noise can affect the listening experience in many real-life situations involving loudspeakers as a playback device. A solution to reduce the effect of the noise is to employ headphones, but they can be annoying and not allowed on some occasions. In this context, a system for improving the audio perception and the intelligibility of sounds in a domestic noisy environment is introduced and a real-time implementation is proposed. The system comprises three main blocks: a noise estimation procedure based on an adaptive algorithm, an auditory spectral masking algorithm that estimates the music threshold capable of masking the noise source, and an FFT equalizer that is used to apply the estimated level. It has been developed on an embedded DSP board considering one microphone for the ambient noise analysis and two vibrating sound transducers for sound reproduction. Several experiments on simulated and real-world scenarios have been realized to prove the effectiveness of the proposed approach.