So it's a common issue that words transliterated from a non-latin alphabet towards latin alphabet will depend on which language using the latin alphabet they're translated into.
- Arabic example : The profet's name is Muhammad in English and Mahomet in French.
- Cyrillic (Russian) example : Tshaikovsky in English, Tschajkowski in German, Czajkowski in Polish, Tchaïkowsky in French.
- Chinese example : Mao Zedong was known as Mao Tsé-Toung in French.
For japanese however, it seems extremely rare that transliteration does not follow stricly the Hepburn system, designed towards english pronunciation.
For example, when transliterating from Japanese to French
- The city Hiroshima should be romanized "Hirochima"
- The island of Hokkaido should be romanized "Hokkaïdo"
- "Wasabi" should be romanized "Wassabi"
But nevertheless, the English romanizations are used everywhere, leading to wrong pronunciation. If this was the case for all non-latin scripted languages I'd understand that the global impact of English language is predominant over correct French pronunciation. However what I don't understand is why this is the case only for Japanese and not for Arabic nor Russian.