Hi everyone!
I’m excited to share FORMANT-1, a lightweight speech synthesis
engine built for the Bytebeat environment. This isn’t just a simple
wave generator — it’s a mathematical model that simulates the
biological mechanics of the human vocal tract.
In this demo, the engine says "HELLO".
Key Features:
LF-Source: Uses an asymmetric glottal impulse model for a natural
"chesty" male timbre.
Nasal Cavity: Features an independent resonance channel (~280Hz)
to add warmth and realism.
Dynamic Coarticulation: Real-time smoothing of formants for fluid
speech transitions.
🚀 How to run it:
Open any online Bytebeat player (e.g., Dollchan Bytebeat or
GreasyFork Bytebeat).
Set the Mode to JavaScript (or float).
Set the Sample Rate to 44100 (this is crucial for the filters to
sound correct).
Paste the code below into the editor and hit Play.
💻 The Code:
t === 0 ? (window.m = {h:new
Float32Array(16).fill(0),f1:500,f2:1500,p:100,a:0,n:0.1}) : 0,
sc =
[[500,1500,0,0,0.8,0],[500,1800,100,1,0.1,0.1],[400,1000,102,0.8,0,0.3],[450,850,95,1,0,0.1],[0,0,0,0,0,0],[0,0,0,0,0,0]][Math.floor((t/44100*1.4)%6)]||[0,0,0,0,0,0],