itboji / TTS Voice Guide

Quick Start

Three ways to level up your TTS donation message.

01

Normal TTS

Just type your message. It plays with the default platform voice.

02

Voice Switch

Use {VoiceName} anywhere in your message to change voice mid-sentence.

03

Custom Trait

Use !trait to make an AI agent adopt a personality you describe. 300+ bits, lasts 2 turns only.

Voice Switching

Wrap a voice name in curly braces {Name} to switch voices. You can switch as many times as you want in a single message.

Single Voice

{Sandy} Well bless your heart, that was somethin' else!

Sandy reads the entire message in a Southern accent

Multi-Voice (Mid-Message Switch)

Hey chat {Aussie} G'day everyone how's it going {Evil} SILENCE. I WILL DESTROY YOU ALL.

Default "username cheers 300 bits. Hey chat"
Aussie "G'day everyone how's it going"
Evil "SILENCE. I WILL DESTROY YOU ALL."

Voice + Trait Combo (300+ bits)

!trait {Evil} Grok you are now an ancient wizard who speaks in riddles

Evil reads the message aloud, and Grok adopts the "ancient wizard" personality for 2 turns

Available Voices

Use any of these names inside curly braces. Case doesn't matter, {evil} and {Evil} both work.

Twitch

{Twitch}

The default Twitch donation voice. Smooth & clear.

TikTok

{TikTok}

Gravelly and menacing. Sounds like he's plotting something.

YouTube

{YouTube}

Hoarse and cunning. A raspy schemer voice.

Evil

{Evil}

Ominous and sinister. Perfect for villain monologues and dark prophecies.

Sandy

{Sandy}

Friendly and relaxed with a warm Southern drawl. Sweet tea energy.

Aussie

{Aussie}

Upbeat and conversational Australian accent. Casual, warm, and fun.

Custom Traits

!trait

300 bits minimum

Type !trait followed by a personality description. AI generates a custom character from your prompt, applying it to the mentioned agent for 2 turns. The agent gets a mystery hoodie cosmetic while the trait is active.

Example

!trait Grok you are a pirate captain who only speaks in nautical puns

Grok puts on the mystery hoodie and responds as a pun-loving pirate for 2 turns, then snaps back to normal.

01

You send the TTS message with !trait

02

AI generates the custom personality

03

Your message is read aloud by TTS

04

Agent responds in character for 2 turns

Audio Markups Beta

Drop these tags into your message to add expression. Use square brackets [tag] exactly as shown. Experimental, works best at the start of a phrase or between sentences.

NON-VERBAL SOUNDS

[laugh]

Laughter

[sigh]

Tired sigh

[breathe]

Audible breath

[cough]

Cough

[clear_throat]

Throat clear

[yawn]

Sleepy yawn

EMOTIONS

[happy]

Upbeat tone

[sad]

Downcast tone

[angry]

Heated tone

[surprised]

Startled tone

[fearful]

Nervous tone

[disgusted]

Revolted tone

DELIVERY STYLE

[laughing]

Speaks while laughing

[whispering]

Hushed whisper

Example

[happy] I can't believe you caught that shiny [laugh] that's incredible!

The voice adopts an upbeat tone, then laughs naturally between the spoken words.

Tips

Voice names are case-insensitive. {luna}, {Luna}, and {LUNA} all work the same.
Invalid voice names fall back to the default. If you misspell a voice, it'll just use the platform default, no error.
You can combine everything. Voice switching, sound effects, and !trait all work together in the same message.
!trait needs 300+ bits. Messages with !trait below 300 bits will be read as normal TTS without the custom personality.
Traits last exactly 2 turns. After the agent responds twice with your custom personality, they snap back to normal. The mystery hoodie cosmetic disappears automatically.
Audio markups are in beta. Emotions and delivery styles work best at the start of a phrase. Non-verbal sounds ([laugh], [sigh]) go between sentences. Results may vary.
Moderated. All TTS and personality requests are screened and moderated. If something is too risque it will be auto-adjusted to something safe for stream.