[Request] Improve (Di/Tri)phthong and Tone Contour Support in trans_list for Vietnamese

[jade-bejide]

I've noticed that trans_list and xsampa_list tends to break up tone contours from their associated phoneme and that diphthongs and triphthongs are split into monophthongs.

Set Up pip install git+https://github.com/dmort27/epitran.git

Issue

>>> import epitran
>>> epi=epitran.Epitran('vie-Latn', tones=True)
>>> epi.transliterate("iã")
'ʔiə˧˥'
>>> epi.trans_list("iã")
['ʔ', 'i', 'ə', '˧', '˥'] # a desirable output is something like ['ʔ', 'iə˧˥'] or at least ['ʔ', 'i', 'ə˧˥']
>>> epi.xsampa_list("iã")
['?', 'i', '@', '_M', '_T'] # a desirable output is something like ['?', 'i@_M_T'] or at least ['?', 'i', '@_M_T']

Is it possible to add support for multiphthongs and tone contours?

[jade-bejide]

Just to add to this, I also think the tone contours are a little wrong for example

epitran/epitran/data/map/vie-Latn.csv

Line 145 in 65a4ca2

uyề,wiə˧

uses a mid tone when it should be low rising (source: https://en.wikipedia.org/wiki/Vietnamese_phonology#Tone). I'm not a Vietnamese speaker so I can't exactly tell whether there are any other errors, but this issue links to:

• Incorrect tone outputs for vie-Latn #222
• Support for tonal langauges #180 (comment)