You kinda shift the entire pitch of your voice so that the words are still, relative to each other, have the tones and it’s distinguishable to a native speaker intuitively.

Its easier in Cantonese, 6 tones, you can play around with words to make the tones go high or low, the tones don’t change much from what I’ve noticed in Cantonese songs, yo es sounds almost the same as spoken. In Mandarin, 4 tones, and the tones are much lower and deeper in the sounds, you just have to change the tones a lot to make it higher pitched, you rarely pronunce the tones as-is. Still understandable.