I agree that level matching is important when the goal is to make accurate subjective quality judgments. This is very well understood in the audio industry. R

I agree that level matching is important when the goal is to make accurate subjective quality judgments. This is very well understood in the audio industry. Rather than microphone quality, the differences between the samples are primarily attributable to microphone position and technique: specifically, being 10 cm from the mic produces something people like better than being 60 cm away. The main issue with the MacBook mic is that it's inconvenient to get close to it. The advice to get a boom to make it easier to bring an outboard mic into optimal position without strain is excellent and highly practical, IMO. The one big quality outlier is the Airpod Max sample, which exhibits compression artifacts and rolled off high end (and low end, but that matters less). It is substantially worse than all the other options. (It probably has worse latency, too.) The iPad Pro sample isn't great either, probably because the mic is off-axis when speaking into it from a natural position. Again, the degradation is primarily a result of suboptimal microphone position and technique rather than component quality.