The linguists at Chásur reconstruct Proto-Taol something like this:

 p   t   kʲ   k
ˀp  ˀt  ˀkʲ  ˀk
 pʰ  tʰ  kʲʰ  kʰ
 f?  s   ʃ    h

n l r ʁ

i  e  a  ɑ  o  u
ī  ē  ā  ɑ̄  ō  ū
î? ê  â  ɑ̂? ô  û?

looks neat!

The existence of *f is uncertain; it is largely reconstructed to explain Maotic words which have /h/ (instead of Ø), however, it has been noted that most such instances precede /a/, e.g. *fālu > hálu "female human".

kʲ ʃ will be spelled *c *š.

The second vowel series is the long counterpart of the first. The third vowel series, marked with a circumflex, is infrequent; its properties are uncertain, but it was preserved into the earliest stages of all Taol languages (including Eastern Taol, which lost length) long enough to be invariably marked in the standard orthographies. In Maotic both the second and third series collapsed into the first; in Dálx the second and third series are both pronounced as long; additionally, the circumflex marks vowels compensatorily lengthened by the loss of a consonant (e.g. *lek-ši > lês. In Eastern Taol the third series is pronounced long (usually analyzed as doubling of the vowel, e.g. âcasi /a.a.tʃa.si/, etc.) The origin of the third series in Proto-* shows several apparent contractions in *VyV, among scattered other sources.

do you think irl (the constructed rl), the difference is tonal? or with voicing modality?

...
A rough sketch of changes from Proto-Taol to the earliest literary period looks like so:

short *ɑ merges with /o/, long *ɑ̄ breaks to /ao/

long *ū breaks to /ue/

preglottalized consonants become voiced and lose preglottalization

interesting. i'd expect them to lose preglottalization and become voiceless plain, while the old voiceless plain become voiced (chain shift, i guess). but this way works, too.

* loss of final *i

loss of *š universally

that's weird! you'd expect some reflex of it, not a complete disappearance (since it's a noisy sound). maybe you get š -> h -> 0?

* loss of *h universally? debuccalization of *f > h

shift of /n/ to /m/ when initial or preceding nearby /n/+a very strange change to be sure, but the distribution bears it out

that makes sense, at least the "preceding /n/" part, that's just dissimilation. the initial part is weirder, but i suppose it could be bullied into happening by the dissimilation change.

loss of gemination

*kʲ > [ts] allophonically before front vowels /e i/; gʲ > /dz/ universally

Later final /e/, /a/ were lost (compare literary fíre with modern fír "tree"), and more importantly the entire length contrast.

looking forward to more!