What determines transcription quality
Three factors, and accent isn't one of them. The first is microphone distance: a phone in the centre of the table picks up far more room noise than the same phone placed near whoever is speaking, and that noise ends up as incorrect words. The second is overlapping speech: when two people talk at once, no system can distinguish both accurately.
The third is recording-specific vocabulary. The most common errors involve company names, surnames and industry acronyms, precisely because they appear nowhere else. These are corrected in a couple of minutes by editing the text, and that's where real accuracy is gained.