The Library of Congress maintains romanization tables for transliteration of non-roman scripts, which are regularly added to.

https://www.loc.gov/catdir/cpso/roman.html

African languages having romanization tables:

Amharic

Arabic

Coptic

Mande languages (in N’ko script)

Moroccan Tamazight

Tamashek

Tigrinya

Vai

Enter all diacritics as found.  Special characters are represented by a standard character that is closest to it, with a double underscore.   In a few cases, special characters are part of approved romanization tables; in that case both versions may be used, e.g. special characters transcribed in field 245 and an added field 246 for the title with underscored characters.

The N’ko romanization table, for example, provides this note:

The use of extended Latin characters in this document leverages use of the Unicode

standard, as approved by the NDMSO office of the Library of Congress in December,

2007, while remaining usable for libraries using MARC-8 as well. Within a strictly MARC-8

environment, the double underscore may continue to be applied to a base character,

following LC-PCC PS 1.4 for the characters ɓ, ɗ, ɛ, ɣ, ŋ, ɲ, and ɔ: b̳, d̳, e̳, g, n̳, n̳ and o̳.

LC-PCC PS 1.4 also allows for the character ə (schwa) to be recorded as ä without

double underscore.

​

Google Translate includes an "Africa Latin" virtual keyboard for African languages.  (Amharic also has its own virtual keyboard.)  Click on the small keyboard icon at lower left of the input window; click on Ctrl + Alt to access the special characters.  OCLC will accept these characters in a bibliographic record; but, at least at present, it will not permit saving of a name authority record that includes them.  (Note in the illustrated example that the transcribed word is in a West African language not yet incorporated into Google Translate, but selecting a Southern African language permits transcribing that word.)