Skip to main content
All CollectionsCreating Courses
Audio: Text-to-Speech Pronunciation tips
Audio: Text-to-Speech Pronunciation tips

Correct and customize pronunciation of your script

Updated over a year ago

Here are some best practices for adjusting pronunciation to get the AI to say a specific word how you need it to sound 🤗

Pronunciation

Acronyms as Words

For an acronym as a word, try spelling the word the way it sounds. For example, FEMA sounds better when you enter it as feema.

Shorter Words

For a shorter word, try spelling the word the way it sounds. For example, edamame sounds better when you enter edamawmay.

Longer Words

For a longer word, try breaking it up by syllables and adjusting the syllable that doesn't sound right.

For example, triiodothyronine sounds better when you enter it as "try idoe thighro neen" (including quotation marks).

Emphasize a Word or Phrase

Add quotation marks ("") around a word or phrase to let the AI know to pay particular attention.

Upward inflection in questions

  • Try recording multiple takes. You’ll notice subtle differences in voice avatar performances, even if you don’t make any changes.

  • Add a comma before the final word or leave off punctuation at the end of the sentence to help guide the inflection. For example: How are you doing, today? or How are you doing today

  • Use a combination of punctuation marks like commas, ellipses, and quotation marks to emphasize certain words or phrases. For example: How are you doing, "today"? or How are "you" doing, "today"?

  • Place quotes around words in the middle of the sentence to emphasize them. For example: How are "you" doing today?


    ​

Emphasis of a word or a phrase

You can try placing the syllable, word or phrase you’d like to have emphasized in quotation marks. If you want the AI to say “PROduce,” as in vegetables, but it’s saying “proDUCE” instead, try entering “pro”duce.

Longer pauses

  • Commas ​add pauses in places where a voice actor would make a small, subtle pause.

  • Periods ​create a pause as well as a downward inflection. They are best used to break a long sentence up into two pieces.

  • Using an ellipse (...) or a combination of punctuation marks (".,.,.,.") can create "breathing room" and add a slight pause between sentences. Add the ellipses or a combination of punctuation at the end of a sentence, like this: Remember...It's essential to create an environment that encourages growth and success.
    ​

Numbers

Just like real voice actors, AI voices need cues to understand if a number is a reference number, a value, an address, a dollar amount, a year, a phone number, etc.

Reference numbers or pages:

You may need to use spaces between numbers to help the AI make accurate predictions. If the reference number 1246 should be read as “one two four six,” enter it as "1 2 4 6"

Dollar amounts:

With dollar amounts: If $1,200,0000 should be read as “one point two million dollars” enter it as "$1.2 million" If it should be read as "one point two million" enter "1.2 million"

Years:

If you want 2022 to be read as "two thousand and twenty-two," you would enter it as "2,022 "

Phone numbers:

A simple way to get the AI to read a number as a phone number ending with double digits is to add spaces between the numbers. For (206) 555-3131 you would enter it as "2 0 6 5 5 5 31 31"

Acronyms & Initialisms

If you want an acronym pronounced like a word, spell it as a word or use Respellings to specify each syllable, for example: ASAP →​ e​igh-sap


If you want an initial pronounced as individual letters, add spaces or dashes between them or use a Respelling: MOM →​ ​M O M or M-O-M.
​

Respellings

Respellings give you a way to format a word that tells the AI exactly what sound each syllable should make and which syllables should be emphasized.

Every respelling will be contained within the two colons with all lowercase letters except for the emphasized syllable or syllables. For example, ::STOO-dee-oh:: for Studio.

Reference Charts

Use the column headers like a fill-in-the-blank sentence:
​“To hear “a” as in all, type AW. For example, ::AWL::”

VOWELS

To hear

as in

type

Example

To hear

as in

type

Example

a

ant

A

::ANT::

i

in

IH

::IHN::

a

spa

AH

::SPAH::

i

isle

Y

::YL::

a

all

AW

::AWL::

o

oat

OH

::OHT::

a

eight

AY

::AYT::

o

ooh

OO

::OO::

e

egg

EH

::EHG::

o

foot

UU

::FUUT::

e

ease

EE

::EEZ::

u

up

UH

::UHP::

CONSONANTS

To hear

as in

type

Example

To hear

as in

type

Example

b

bunk

B

::BUHNK::

qu

quote

KW

::KWOHT::

ch

chart

CH

::CHAHRT::

r

rain

R

::RAYN::

d

dust

D

::DUHST::

s

slice

S

::SLYS::

f

first

F

::FURST::

sh

shy

SH

::SHY::

g

glow

G

::GLOH::

t

tarte

T

::TART::

h

horse

H

::HORS::

th

though

DH

::DHOH::

j

jell

J

::JEHL::

th

think

TH

::THIHNK::

k

kite

K

::KYT::

v

van

V

::VAN::

l

laugh

L

::LAF::

w

win

W

::WIHN::

m

mask

M

::MASK::

x

axe

KS

::AKS::

n

nest

N

::NEHST::

y

yes

Y

::YEHS::

ng

ring

NG

::RIHNG::

z

zen

Z

::ZEHN::

nk

rink

NK

::RIHNK::

zh

measure

ZH

::MEH-zhur::

p

pop

P

::PAHP::

VOWELS COMBOS

To hear

as in

type

Example

To hear

as in

type

Example

ar

car

AR

::KAR::

ow

cow

OW

::KOW::

er

error

ERR

::ERR-ur::

oy

oy

OY

::OY::

or

more

OR

::MOR::

ur

urn

UR

::URN::


Respellings for Acronyms

When you need to respell an acronym or initialism, where each letter is spoken as itself (ASPCA, LCD, BBC, FBI), use the following guideline:


ABCDEFGHIJKLMNOPQRSTUVWXYZ → ::ay-bee-see-dee-ee-ehf-jee-aych-y-jay-kay-ehl-ehm-ehn-oh-pee-kyoo-ar-ehs-tee-yoo-vee-DUH-buhl-yoo-ehks-wy-ZEE::

Did this answer your question?