EVA to IPA

[ErinaBee]

EVA to IPA:



VOWELS:

a = /a/

e = /ɛ/ (or /e/ acceptable)

i = /i/

o = /o/ (or /ɔ/ acceptable)

u = /ɯᵝ/ (like u in Japanese)

y = /ə/

ee = /u/

ii = /y/

eee = /ø/ (/œ/ also acceptable)

an = /ɑ̃/

ain = /ɛ̃/

aiin = /ẽ/ (/ẽj̃/, /ej̃/ also acceptable)

aiiin = /ãj̃/ (/aj̃/ also acceptable)

am = /ɐ̃w̃/ (/ɐw̃/ also acceptable)

aim = /ɛ̃w̃/ (/ɛw̃/ also acceptable)

aiim = /ẽw̃/ (/ew̃/, /ĩw̃/, /iw̃/ also acceptable)

on = /ɔ̃/

oin = /ɔ̃j̃/ (/ɔj̃/ also acceptable)

oiin = /õj̃/ (/oj̃/, /ũj̃/, and /uj̃/ also acceptable)

oiiin = /ɥĩ/ (/wĩ/)

om = /õ/ (/ũ/ also acceptable)

oiiim = /ɔ̃w̃/

in = /ĩ/

im = /ĩw̃/ (/iw̃/ also acceptable)

iin = /jẽ/

CONSONANTS:

p = /p/

t = /t/

k = /k/

b = /b/

d = /d/

g = /g/

l = /l/

r = /r/ (any rhotic acceptable)

c = /ʦ/

ch = /ʧ/

j = /ʣ/ (/ʤ/ also acceptable)

s = /s/

Sh = /ʃ/

f = /f/

cTh = /n/

cKh = /m/

cPh = /h/

cFh = /ŋ/

q = /x/ (/χ/ also acceptable)

z = /z/

v = /v/

x = /ʝ/

Ih (capital I, not lowercase L) = /ç/

This is not a proposed solution, just an idea of how to make "spoken EVA" sound good

[ReneZ]

So.... could online AI-based tools be instructed to use this, so that we can actually hear what this sounds like? I am just thinking that this should be within their capability.

To avoid misunderstanding, for me this would be 'just for fun', but it does not seem to be a worse use of these tools than any of the recent 'AI-based' solutions.

[ErinaBee]

I'm actually making a demo, with Synthesizer V Studio 2

[dashstofsk]

make "spoken EVA" sound good

I doubt whether this would make a nice sound. Most syllables in most languages are CV syllables, a consonant sound followed by a vowel sound. This is not surprising since this the way that we as babies first experimented with speech, making the sounds that are natural due to the way that the mouth muscles have evolved, and hence the way that language sounds evolved. "Ba-ba, goo-goo, da-da". In order for your VMS speech to flow nicely it will have to follow some similar structure.

[Jorge_Stolfi]

Most syllables in most languages are CV syllables, a consonant sound followed by a vowel sound.

You are not allowed to view links. Register or Login to view.
Gvprtskvni - how is this even a word, Georgian!?

[ErinaBee]

make "spoken EVA" sound good

I doubt whether this would make a nice sound. Most syllables in most languages are CV syllables, a consonant sound followed by a vowel sound. This is not surprising since this the way that we as babies first experimented with speech, making the sounds that are natural due to the way that the mouth muscles have evolved, and hence the way that language sounds evolved. "Ba-ba, goo-goo, da-da". In order for your VMS speech to flow nicely it will have to follow some similar structure.

it does actually sound really good. Not harsh or strained at all. I just did the beginning quote from the manuscript and it works

[dashstofsk]

it does actually sound really good.

You will need to post some recordings. I would be interested to hear them.

[ErinaBee]

You are not allowed to view links. Register or Login to view. here it is  this is an interpretation of a few lines on the beginning of the manuscript, made using Synthesizer V Studio 2, particularly the voicebank Feng Yi 2. Synthesizer V Studio 2 is similar to Vocaloid, you basically insert the phonemes and it synthesizes the vocals.

[Mauro]

You are not allowed to view links. Register or Login to view. here it is  this is an interpretation of a few lines on the beginning of the manuscript, made using Synthesizer V Studio 2, particularly the voicebank Feng Yi 2. Synthesizer V Studio 2 is similar to Vocaloid, you basically insert the phonemes and it synthesizes the vocals.

It's cool 

[MarcoP]

You are not allowed to view links. Register or Login to view. here it is  this is an interpretation of a few lines on the beginning of the manuscript, made using Synthesizer V Studio 2, particularly the voicebank Feng Yi 2. Synthesizer V Studio 2 is similar to Vocaloid, you basically insert the phonemes and it synthesizes the vocals.

I think it would be even better without the music and as a video with EVA subtitles (or highlighting the words on an image of the manuscript text).