One thing I have noticed about all three videos so far is that the overall loudness is quite low. According to YouTube's 'Stats for nerds', the loudness of the videos is 13.3 dB, 12.4 dB and 12.8 dB below its standard loudness of -14 LUFS. However, since YouTube only reduces the audio level of content with loudness exceeding its standard, the audio level could probably be increased by more than 10 dB before YouTube would reduce it.
To see how much the audio levels could be increased without the need to introduce adverse effects, I downloaded the audio components from the first and the latest video using
https://youtube-converter.online/ and calculated their peaks, RMS levels and crest factor for both stereo left and right and also mid (mono, sum of left and right) and side (difference between left and right) while also plotting their waveforms. The unit used is dBO, where the O stands for overload, referenced to the highest sample value.
First, the first video, which was the review of the Topping D30Pro. The audio seems to be mono and fairly consistent in level and can probably be increased by 12.08 dB without introducing adverse effects. This would put its loudness only 1.22 dB below YouTube's standard.
View attachment 112289
View attachment 112291
Similarly for the latest video, which was the investigation of differences from fancy cables. The audio also seems to be mono and fairly consistent in level and can probably be increased by 11.60 dB without introducing adverse effects. This would put its loudness only 1.20 dB below YouTube's standard.
View attachment 112290
View attachment 112292
Thus, while the dynamics of the material seem to be about right for YouTube, the level is not. However, it can probably be increased to be so without the need to introduce adverse effects by use of peak normalisation. Given that YouTube uses a lossy data compression codec for the audio, the peak of the audio before compression should probably be normalised to below the highest sample value to avoid exceeding it after being encoded and decoded. How much exactly, I don't know. However, from what I recall, 1 dB may well be enough.