Communication methods

Last updated: October 17, 2026

A voice protocol is how a creator communicates. It might be their natural voice, typed TTS, speech-to-text-to-speech, voice conversion, on-screen text, chat text, soundboards, gestures, sign language or a mix of several.

Voice Protocol focuses on creators who communicate beyond their natural voice. This page explains each method in plain terms: where the message starts, what happens to it, and what reaches the audience.

A method describes how a message travels. It doesn't tell you why someone uses it, and it isn't a label for the person. The same setup can come from privacy, character design, comfort or simple preference, among other reasons.

Natural voice

Natural voice means speaking on stream with your own voice. It's the most common voice protocol, and many creators mix it with the methods below. Voice Protocol focuses on those other methods.

See creators who use this →

Typed TTS

Typed text-to-speech, or typed TTS, turns written messages into a spoken voice. The creator writes each line, and a speech engine reads it aloud. The words belong to the creator. The software supplies the sound.

Input: A typed message. Some creators send saved phrases or use another text-entry device instead of a keyboard.
Processing: A speech engine converts the text into audio.
Delivery: A synthetic voice in the stream, often with the same text shown on screen.
What viewers experience: A consistent voice that can become part of how they recognize the creator. Replies arrive after a short pause while each line is written and sent.
Common misconception: "TTS means the creator can't speak." TTS tells you how the voice is made. It doesn't tell you whether the creator can speak or why they chose it.

Read more: What is a TTS VTuber?

See creators who use this →

Speech-to-text-to-speech

Speech-to-text-to-speech, often shortened to STT→TTS or STTTS, starts with the creator speaking. Software transcribes the words, and a synthetic voice reads the transcript. Viewers hear the new voice instead of the microphone, as long as the audio is routed that way.

Input: Spoken words into a private microphone.
Processing: Speech recognition turns the speech into text, then a speech engine voices that text.
Delivery: Synthetic speech, sometimes with captions.
What viewers experience: Spoken replies without keyboard composition, plus the occasional misheard word that the voice reads out anyway. Some creators turn those slips into running jokes.
Common misconception: "Every TTS creator types." This method begins with speech, so a synthetic voice alone doesn't tell you how the words went in.

Read more: What is a TTS VTuber?

See creators who use this →

Voice conversion

Voice conversion changes the sound of live speech. The creator talks, and software transforms the audio into a different voice in real time. There is no text stage, so it is a separate method from TTS, even though both change what viewers hear. Products sold as voice changers may use voice conversion or simpler effects such as pitch shifting.

Input: Live speech into a microphone.
Processing: The audio itself is transformed. TTS starts from text, while voice conversion starts from speech (UnifySpeech research paper).
Delivery: The creator's own timing and phrasing in a different voice.
What viewers experience: Live conversation with the rhythm of ordinary talk.
Common misconception: "A voice changer is a kind of TTS." They can sound alike, but they work from different inputs.

Read more: What is a TTS VTuber?

See creators who use this →

On-screen text

On-screen text puts the creator's words on screen for viewers to read. It might appear in a chat bubble beside the avatar, in a text panel or in stream chat. No voice is needed at all.

Input: Typed messages.
Processing: Display software shows the text. A basic version can be as simple as an OBS text source that reads from a file.
Delivery: Written words on screen.
What viewers experience: A stream they read as well as watch. Wording, punctuation and timing all become part of the creator's style. The text needs to be large enough, and on screen long enough, to read on a phone.
Common misconception: "Creators who don't speak all use TTS." Plenty of streams run on text alone, and a text-led stream never has to become a TTS stream.

See creators who use this →

Chat text

Chat text means the creator replies by typing directly into the stream's chat or a Discord channel, with no synthetic voice and no on-screen overlay. Viewers read the creator's messages alongside their own.

Input: Typing into the platform's chat (Twitch, YouTube, Discord).
Processing: None beyond the platform. Messages may be highlighted or badged as the broadcaster's.
Delivery: Text in the chat window, which viewers watching full screen, on a TV or in a VOD or clip may not see unless the chat is shown on stream.
What viewers experience: A conversation that happens in the same space they type in.
Common misconception: "Chat text and on-screen text are the same thing." On-screen text is part of the broadcast and shows up in VODs and clips. Chat text lives in the platform's chat unless the creator captures it.

See creators who use this →

Soundboards and prepared phrases

A soundboard plays prepared audio at the press of a button. A clip might be recorded words, a signature sound, a saved TTS phrase or a sound effect. Soundboards are built for speed and for the things a creator expects to say again and again.

Input: A button or hotkey.
Processing: Playback of a clip that already exists.
Delivery: Recorded speech or sound.
What viewers experience: Greetings, reactions and catchphrases that land on time, often filling the gap while a longer message is typed.
Common misconception: "A soundboard is the same as live TTS." A clip is prepared in advance. Typed TTS voices a new message written in the moment.

See creators who use this →

Gestures and expressions

Gestures and expressions let the avatar carry meaning through its face and body. A creator can put expressions and short animations on hotkeys, or use tracking to bring their own movements onto the model.

Input: Hotkeys, face tracking or hand tracking.
Processing: Avatar software maps the input to expressions, animations or finger movement. VTube Studio, for example, offers experimental webcam hand tracking that can move a model's fingers or trigger hotkeys from gestures.
Delivery: Visible reactions and movement on the avatar.
What viewers experience: Instant emotion. A wave, a laugh or a shocked face reads in a second, even while a longer reply is on its way.
Common misconception: "Gestures are just decoration." For some creators the avatar's body is the main channel, with text or sound in support. Tracking can drop out during fast hand movements, so it is worth testing with the gestures you actually use.

See creators who use this →

Sign language

A sign language is a complete language, not a set of gestures. American Sign Language, for example, has its own grammar, uses both the hands and the face, and is a different language from British Sign Language (NIDCD). There is no universal sign language.

Input: Signing, captured by a camera or a hand tracker.
Processing: The creator's real hands appear on a hand camera, or tracking software maps the signing onto the avatar.
Delivery: Signed language, often alongside captions or on-screen text for viewers who don't sign.
What viewers experience: For viewers who sign, direct communication in their own language. For everyone else, captions or text carry the message.
Common misconception: "Avatar tracking captures signing perfectly." Fingerspelling depends on clear handshapes, and the face carries grammar too, such as raised eyebrows marking a question. Tracking may miss that detail, so a hand camera or captions can help.

See creators who use this →

Mixed setups

A mixed setup, sometimes called a hybrid, combines several methods and switches between them by moment or context. A typed line explains a plan, a soundboard clip lands the joke, and an expression carries the reaction. Some creators run their speech-to-text-to-speech output through voice conversion as well, reshaping the synthetic voice into something more distinctive.

Natural voice can be part of a mix too. Some creators speak in one setting and use text or TTS in another, and a setup can change over time.

Input: Whatever each moment needs, from typing to hotkeys to speech.
Processing: Each part of the mix follows its own path, described in the sections above.
Delivery: Several channels at once, such as voice, text and avatar.
What viewers experience: A stream with more than one way in. Viewers can read, listen and watch the avatar react.
Common misconception: "A creator has to pick one method." Methods aren't a ladder or a ranking, and moving between them doesn't change who the creator is.

See creators who use this →

A contrast: AI VTubers

Some VTubers speak through TTS but don't write their own words. With an AI VTuber, software generates the dialogue as well as the voice. In every method above, a person supplies the words. With an AI VTuber, a system does.

Input: A generative system produces the line.
Processing: TTS voices it, and the avatar performs it.
Delivery: Synthetic speech written by software.
Common misconception: "A synthetic voice means it's AI." The sound alone can't tell you who wrote the words, and an AI-themed character design isn't proof either.

AI VTubers appear here only to show the difference.

Read more: What is a TTS VTuber?

A look-alike: viewer chat TTS and TTS sidekicks

Many streaming tools come with text-to-speech that reads viewers' chat messages or donation alerts aloud, so "TTS on stream" often means the audience's words, not the creator's. In those streams the host usually talks normally, and the synthetic voice belongs to the viewers.

Some creators also give a pet, mascot or sidekick a TTS voice to banter with. That character is a co-star, not how the creator communicates.

A TTS VTuber is different because the synthetic voice carries the creator's own words.

Frequently asked questions

What is a voice protocol?

A voice protocol is how a creator communicates. It might be their natural voice, typed TTS, speech-to-text-to-speech, voice conversion, on-screen text, chat text, soundboards, gestures, sign language or a mix of several.

What is the Voice Protocol?

Voice Protocol is an independent index and journal of creators who communicate beyond their natural voice, whether by necessity or by choice. It profiles creators by how they communicate and publishes stories and guides about those methods.

Does a creator's method say why they use it?

No. The same setup can come from privacy, character design, comfort or preference, among other reasons. Voice Protocol describes the method, not the reason.

Are TTS VTubers AI?

Not unless software writes the words. In a human TTS setup, the creator supplies every line and the software only produces the sound.

Are gestures and sign language the same thing?

No. Gestures and expressions add meaning to what the avatar does. A sign language is a complete language with its own grammar.

Can a creator use more than one method?

Yes. Setups often combine methods, and a creator's mix can change over time.

Explore creators by method in the Voice Protocol directory.