It's very unlikely anyone could hear 0.018% THD+N with real program material in a room with speakers and NOT headphones. It's why I'm not a fan of the SINAD gospel of people chasing a #. I strongly believe based on my years of testing and designing highly sensitive audio communication systems for Telcom and Government defense that anything below 0.01% (80dB) THD+N is below the threshold of audiblity using test tones and with controlled listening with headphones and noise isolation. Once you play the tones in a room on loudspeakers, I think the threshold can be a bit higher. Of course with program material, all bets are off as the masking effect tends to dominate unless severe clipping is present.