Photoplethysmogram (PPG) obtained from wearable devices encounters persistent challenges due to significant motion artifacts, necessitating waveform denoising and restoration. The literature extensively documents that ineffective procedures often induce alterations or degradation of the underlying cardiovascular information. Contemporary trends in biosignal processing prioritize advanced data-driven artificial intelligence tools, particularly end-to-end deep learning, to encode features, exploiting them to reconstruct high-quality signals. However, such methods often struggle with denoising inherent bodily signals due to various random and systematic distortions. To address these challenges, this paper introduced a style transfer–assisted cycle-consistent generative adversarial network (stccGAN) to denoise a 3-channel PPG (red, green, and infrared wavelengths) signal acquired by the Soundi chest sensor (patented and certified). The study employed two identical devices: one gathering the chest PPG signal (to be denoised), and the other synchronously acquiring the finger PPG signal (reference signal). The acquisition protocol, adhering to Ethical Committee approval Opinion 3/2019 Politecnico di Milano University, involved 30 healthy subjects, with each undergoing a 5-minute rest period in a sitting posture and a 5-minute session in a standing posture, repeated 5 times per participant. The acquired signals undergo uniform sampling at 400 Hz, filtering (0.5 to 25 Hz pass-band Bessel model), amplitude normalization, subdivision into 5-second chunks, and quality checking (removing undue finger PPG chunks along with corresponding chest signals). The developed stccGAN model employed two identical generators, each comprising a deep network based on the UNet model, and two discriminators. The network architecture processed a tridimensional array (red, green, and infrared) of 2000 samples, the chest PPG to be cleaned, generating denoised chest PPG signals. The training optimized concurrently adversarial and cycle losses. Results demonstrated a signal correlation exceeding 95% on average on the independent test set, indicating the method's efficacy in restoring the quality of PPG signal recorded at the chest.