Gibber Voice Generator — Free Browser Tool for Game Character Voices (Celeste / AC Style)
Gibber Voice Generator is a free browser tool that generates character dialogue sounds for games, animations, and interactive projects — without recording a single word.
It doesn't do text-to-speech. Instead, it reads the rhythm, punctuation, and structure of your script and generates a "performance" — something that sounds like a character talking, but carries no real language. The same approach used in Celeste, Animal Crossing, Undertale, and Don't Starve.
——————————————————————
TWO MODES
🔵 Synth Mode
Generates voice entirely from Web Audio synthesis — no audio files needed.
Control pitch, hoarseness, speaking speed, syllable length, amplitude envelope, filter envelope, and a 16-step pitch trend grid.
🟤 Sample Mode
Upload any audio (WAV / MP3 / M4A), or use a built-in demo sound.
The app slices it into grains and triggers them according to your script's rhythm.
Supports transient / even slicing and random / coherent grain flow.
——————————————————————
HOW IT WORKS
Every line you type is parsed into a performance plan:
— syllable count → controls note count
— punctuation → ? rises, ! hits harder, … trails off
— Chinese tones → optional Tone Match mode maps tonal contour to pitch
That plan drives both the audio engine and the on-screen mouth animation simultaneously.
——————————————————————
FEATURES
✔ 8 built-in character presets (name, colour, pixel avatar, saved lines)
✔ Bilingual UI — English and Chinese
✔ Export WAV — use directly in your game, DAW, or animation timeline
✔ Runs entirely in the browser — no account, no install
✔ Free beta — Pay What You Want on itch.io
——————————————————————
LINKS
▶ Try it free on itch.io → https://fredyfeng.itch.io/gibber
——————————————————————
Built with React, TypeScript, and Web Audio API.
This started as a Max/MSP experiment for a game jam. So I rebuilt it as a proper web app.
Currently in beta. If you find bugs or have feature requests, drop them in the comments.