An experiment with phonetic writing

[rikforto]

This seems fine to me:

I brought it to you — the word “aujourd'hui”. You and the native speaker pronounce it the same way, but you don't know how to spell it. You write it down in your own way (oujuvie), and the speaker does it the right way (aujourd'hui). 

But it does seem to differ from what you were saying before. If they are pronouncing it the same way and writing down the same pronunciation in different letters, what is the inaccuracy here:

The author clearly perceived Chinese in his own way, with a “European” ear, and we cannot know how accurately he represented what he heard (for example, the speaker might have been reading quickly, and the author might have taken two similar characters for one word).

This doesn't seem to me like you are saying the accurate pronunciation is mostly getting recorded, but if that is what you are saying, I just misunderstand you.

And to be clear because I've been repeatedly strawmanned about this in the past: I'm not troubled by the possibility of some errors, and know enough about manuscripts that a perfect transcription would be beyond remarkable. I am just trying to understand who this person who can hold a conversation and design an alphabet for distinctions in phonology but not hear them it in dictation is supposed to be.