SynthV Wiki:Sandbox/English phonetics

This article is still a work in progress

This article is slowly being improved for the benefit of all visitors. Please bear with us while improvements are being made. We apologise for the inconvenience this may cause in the meantime.
(What's Being worked on: Spelling/usage/modifications of ARPABET, Arpasing recording scripts, etc.)

(Inspired by VW page of the same name. Studio has no phonetic guide in it, and the older Editor is hosted in Animen's CDN but not accessible, so I thought it's better to have this page.)

About

(We can talk about the language, the history, and the sheer diversity of the language, though it's not the case (?) on SynthV because it seems it only aims on one certain accent)

Synthesizer V and the English language

(We can talk about the spelling, usage, and modifications of ARPABET –the basis for Arpasing–, and other technical stuffs)

English recording scripts

(Not fully unveiled but on AHS Stream #169, they showed a glimpse of it. Very likely using the same principle with Kanru Hua's experimental Arpasing recording scripts made before he went fully focused on SynthV (but not the stable Arpasing versions): actual words to ease the recording effort while also to cover as much phoneme transitions as possible <-- it's for "standard" voice databases, and likely the standard recording scripts. AI one, the collection of songs and other stuffs is yet to known.)

Notes on accents

Despite the general belief that singers completely lose their accents when they sing, this is not the case in every instance, and an accent is possible to be heard even in singing vocals.

However, the reason many are led to believe this is that there are several methods of training singers to disguise or otherwise hide their natural accents—they may even adopt an accent that isn't their own for singing. Samples include genres such as western or country, black music such as jazz or soul. Singing also uses different muscles to speech, resulting in difference of air pressure and way the throat moves. Genres such as opera are most likely to make a accent appear almost entirely absent thanks to the impact of the opera vibrato.

Synthesizer V will capture any form of accent quite easily at times. It depends on the recording method used by the voice provider, type of sound being recorded per sample (accent impact varies per sample and language) and –in the case of Standard voice databases– overall number of samples that make up the voice database (the more samples, the more chance of it slipping in). (Types of voice database produced: "standard" concatenative by "chanting" the scripts contains almost all possible phoneme transitions in particular pitches and tempos; or AI-based by actually sing, apparently also impact the accent.)

(Also the fact that Synthesizer V is using ARPABET for English, albeit modified, is notable for forcing the American accent to the voice provider).

Native accented

American-English Accented

• Eleanor Forte
• SOLARIA
• Kevin
• Natalie
• ASTERIAN
• SAROS
• NYL
• Felicia: Her unrevealed voice provider is described by Dreamtonics as an American person based in Japan.

Australian-English Accented

• Ninezero

Non-native accented

Synthesizer V Studio version 1.5.0 introduced the Cross-Lingual Singing Synthesis feature, which allows all AI voice databases to sing in all languages supported by the software, regardless of the language(s) in which they're recorded. All voice databases recorded in languages other than English are considered to be non-native accented, due to a deliberate decision to leave subtle accents in place.

• Tsurumaki Maki: Received a dedicated separate English voice database and is thus considered a bilingual vocalist in Japanese and English. Her voice provider, Manami Tanaka, is a Japanese voice actress who was known to have studied in the United Kingdom.

• Xingchen: Confirmed to be recorded with English training material to enhance Cross-lingual Singing Synthesis capabilities, though Xingchen Infinity was not considered to be an English voice database by Beijing Photek S&T Development Co., Ltd., labelling and categorizing her as a Mandarin Chinese voice library only. Her voice provider, Chalili, is a Chinese singer who was known to have lived abroad in the United States.

• Weina: Confirmed to be recorded with English training material to enhance Cross-lingual Singing Synthesis capabilities and is considered to be a bilingual vocalist in Mandarin Chinese and English according to Dreamtonics Co., Ltd.'s product description and categories. Weina Hu, Weina's voice provider, is a Chinese actress and singer. Her proficiency in English is currently unlisted.

• D-Lin: Confirmed to be recorded with English training material to enhance Cross-lingual Singing Synthesis capabilities, however, D-Lin is not considered to be an English voice database by Dreamtonics Co., Ltd., labelling and categorizing him as a Mandarin Chinese voice library only. Information about D-Lin's voice provider is unknown, including his proficiency in English.

• Sheena: Confirmed to be recorded with English training material to enhance Cross-lingual Singing Synthesis capabilities and is considered to be a bilingual vocalist in Japanese and English according to Dreamtonics Co., Ltd.'s product description and categories. Information about Sheena's voice provider is unknown, including her proficiency in English.

• Topaz: Her voice provider Elley, of Korean origin, is able to sing well in three languages, including English. Topaz was advertised by GEMVOX as a bilingual Japanese and English voice database, implying she was recorded with English training material to enhance Cross-lingual Singing Synthesis capabilities.

Accent unknown or undetermined

• ANRI
• JUN
• Ritchy: While Ritchy's voice provider was confirmed to be MBanja Ritchy, details about his accent in English is undetermined. MBanja Ritchy is a Cameroonian and French rapper/beat maker who lived abroad in Japan and is trilingual in English, French, and Japanese.
• Hayden
• NOA

Phonetic list

Symbol	Classification	IPA symbol/name	Example	Notes	Related phonemes
aa	vowel	ɑ open back unrounded vowel	palm		ao
ae	vowel	æ near-open front unrounded vowel	bat	In some dialects, it may be diphthongized into /eə/ or similar due to Æ-tensing.
ah	vowel	ʌ open-mid back unrounded vowel	but	In some dialects, fronted to [ɜ], or fronted and lowered to [ɐ].	ax (relaxed)
ao	vowel	ɔː open-mid back rounded vowel	bought	This vowel has a lot of variations depending on the dialect. In US dialects it varies between /ɑ/ for the cot–caught mergers and /ɒ~ɔ/ for the rest.	aa
aw	diphthong	aʊ̯	about
ax	vowel	ə schwa	rental		ah (stressed) er (r-colored)
ay	diphthong	aɪ̯	bite
b	consonant	b voiced bilabial plosive	buy		p (voiceless)
ch	consonant	ʧ voiceless postalveolar affricate	china		jh (voiced) sh (spirantizated) t (deaffricated)
cl	consonant	ʔ glottal stop	uh-oh	May introduce vocal fry to some voice databases. Can also be used as an allophone to k, t, and p in some extent.
d	consonant	d voiced alveolar plosive	die		t (voiceless) dh (lenited, lowered)
dx	consonant	ɾ alveolar flap	butter	Unstressed allophone of /t/ or /d/ phonemes (alveolar flapping).	t, d (allophone)
dr	consonant	d̠ɹ	drive	Initially introduced to differentiate between onset and interword consonant-cluster. Realizations may vary depends on voice databases, from making it postalveolar to completely dropping the "r".	tr (unvoiced)
dh	consonant	ð voiced dental fricative	the	Realizations may vary depends on voice databases. See Pronunciation of English ⟨th⟩ for more info.	th (unvoiced)
eh	vowel	ɛ open-mid front unrounded vowel	bet		ey (diphthongized)
er	vowel	əɹ, ɚ or ɝ (US)	bird	R-colored schwa.	ax (non-rhotic)
ey	diphthong	eɪ̯	hey		eh (monophthong)
f	consonant	f voiceless labiodental fricative	fight		v (voiced)
g	consonant	g voiced velar plosive	guy		k (voiceless) ng (nasalized)
hh	consonant	h voiceless glottal fricative	high
ih	vowel	ɪ near-close near-front unrounded vowel	bit		iy (tense)
iy	vowel	iː close front unrounded vowel	beat		ih (lax)
jh	consonant	ʤ voiced postalveolar affricate	just		ch (voiceless) zh (spirantizated) d (deaffricated)
k	consonant	k voiceless velar plosive	kite	Realizations (aspirated, unaspirated, unreleased) are depends on phonological contexts.	g (voiced)
l	consonant	l alveolar lateral approximant	lie	Realizations (whether it's "dark" or "clear") are depends on phonological contexts.
m	consonant	m bilabial nasal	my		n (alveolarized)
n	consonant	n alveolar nasal	nose		ng (velarized) m (labialized)
ng	consonant	ŋ velar nasal	sing		n (develarized)
ow	diphthong	oʊ̯~o (US)	boat
oy	diphthong	ɔɪ̯	boy
p	consonant	p voiceless bilabial plosive	pie	Realizations (aspirated, unaspirated, unreleased) are depends on phonological contexts.	b (voiced)
r	consonant	ɹ alveolar approximant	risk	Realizations may vary depends on voice databases. See Pronunciation of English /r/ for more info.
s	consonant	s voiceless alveolar fricative	say		z (voiced) sh (palatalized)
sh	consonant	ʃ voiceless postalveolar fricative	shy		zh (voiced) ch (affricated) s (depalatalized)
t	consonant	t voiceless alveolar plosive	tie	Realizations (aspirated, unaspirated, unreleased) are depends on phonological contexts.	d (voiced)
tr	consonant	t̠ɹ	tree	Initially introduced to differentiate between onset and interword consonant-cluster. Realizations may vary depends on voice databases, from making it postalveolar to completely dropping the "r".	dr (voiced)
th	consonant	θ voiceless dental fricative	thing	Realizations may vary depends on voice databases. See Pronunciation of English ⟨th⟩ for more info.	dh (voiced)
uh	vowel	ʊ near-close near-back rounded vowel	book		uw (tense)
uw	vowel	uː close back rounded vowel	boot		w (semivowel) uh (lax)
v	consonant	v voiced labiodental fricative	vote		f (voiceless)
w	consonant	w labio-velar approximant	wise		uw (syllabant) uh
y	consonant	j palatal approximant	yes		iy (syllabant) ih
z	consonant	z voiced alveolar fricative	zoo		s (voiceless) zh (palatalized)
zh	consonant	ʒ voiced postalveolar fricative	measure		sh (voiceless) z (depalatalized) jh (affricated)