I'm going to try and romanize this with as few letters as possible. The first time will be as if the entire inventory is separate phonemes, the second time will also assume many are allophones
Independent phonemes:
/m n/ np n
/b d g/ pp nn nk
/p t̪ c k q/ p t ke g kk
/v z/ ph nh
/f s̪ x h/ ho th kh h(h)
/l j w/ r e o
/r/ rr
If /t̪/ and /s̪/ weren't explicitly dental, they would have been <nh> and <rh>, respectively
/i y u/ e eo o
/e o/ ae ao
/a/ a
I don't know the phonotactics of this language (nor most of the other ones in this thread), but <hh> is used for /h/ next to a letter with which it would normally form a multigraph with. Otherwise an apostrophe is used for this purpose.
Allophones:
/m n/ m n
/b d g/ m n k
/p t̪ c k q/ p t k k k
/v z/ m n
/f s̪ x h/ p t k h
/l j w/ r i u
/r/ r
underlying consonant inventory: /m~b~v~w n~d~z~j p~f t̪~s̪ k~c~q~g~x~z~j h s r~l w j/. the archiphoneme is listed first, which further shrinks the inventory to /m n p t̪ k h r j w/.
* All consonants can lengthen, tho some of them change into other phonemes. Long consonants do not occur word-initially.
* Generally, /m n/ always occur next to another consonant, /p b t̪ d c k/ always occur when lengthened, and /f v s̪ z x/ always occur between vowels or word-finally. Word-initially, /m/ and /b/, /n/ and /d/, and /k/ and /g/ are in free variation; however, /f v s̪ z x/ can occur word-initally in rapid speech after a word ending in a vowel.
* /v/ often further weakens to /w/ or simply lengthens a following /u/ (including /uj/ becoming /y/).
* /p t̪ k~c~q/ become /f s̪ x/ if next to a consonant that is not a fricative or approximant. Otherwise, they stay the same. If the next consonant is voiced, both /k~c~q/ and /x/ become /g/.
* /i u/ become /j w/ before or (sometimes) after a vowel. When /i/ palatalizes the preceding consonant and it follows a vowel, the underlying /j/ merged with the aforementioned palatalized consonant.
* /t̪ k x/ always become /s̪ c s̪/ before /i~j/ (including /iw/ becoming /y/), tho only /k/ to /c/ occurs when lengthened. Likewise, /n~d g/ before /i~j/ or /y/ are usually /z/, but this is likely to weaken to /j/ or merely lengthen a following /i/ or /y/.
* /k/ always becomes /q/ before /a/, including /aj aw/ becoming /e o/.
* /m~b~v p~f/ take an epithentic /w/ before a vowel other than /u/ (again including /y/ from /uj/) and may lengthen said vowel. Sometimes this offglide is dropped before /aw~o/, or after /m/. This is assuming /v/ does not lenit to /w/.
* /r/ always becomes /l/ word-finally, before /j/ (but not necessarily /i/), or after fricatives. /r/ is always between vowels or after a consonant that is not a fricative. Otherwise, /r/ and /l/ are in free variation.
* /m n g/ nasalize the following vowels and semivowels. This includes /m~b n~d g/ weakening to /v~w z~j z~j/.
/i y u/ i ui u
/e o/ ai au
/a/ a
underlying vowel inventory: /a i u/. there are also the contours /aj aw iw uj/, but these often collapse into /e o y y/.
* All vowels lengthen, however this only occurs in stressed open syllables and it is not phonemic.
* Contours /aj aw iw uj/ become /e o y y/ when not in lengthening environments or before another vowel.
* Impossible consonant clusters add an epithentic /e/ (next to coronals) or /o/ (next to other consonants). This vowel is not written.
* The exact quality of /a/ is rooted in vowel harmony. /a/ is perfectly central normally or if the next syllable has another /a/ (including /aj aw/ becoming /e o/). If /i~j/ or /iw~y/ follows, it becomes front /a/, and if /u~w/ or /uj~y/ follows it becomes back rounded /ɒ/.
* Unstressed closed-syllable /i u/ often become /e o/.
