Synthesize speech, singing and beatboxing via a vocal tract model in the browser
Run the free CHIMERA physical voice synthesizer - no installation, right in your browser - and get talking, singing and beatboxing digital vocals in minutes
- 1.Open CHIMERA at chimera-vocal-synth.vercel.app in Chrome or Edge. Wait for the WebAssembly engine to load - a head cross-section with tongue, lips, jaw and air pressure indicators will appear (pitch ≈ 110 Hz, breath pressure ≈ 8.0 cmH2O).
A secure context is required - open via https:// or on localhost. If the service is unavailable in your region, try alternative access via a mirror or local launch from the GitHub repository.
CHIMERA ↗ - 2.Switch to Speak mode (button on the top panel). Enter a phrase in English or IPA into the text field - the engine converts text into articulatory gestures. A gestural score and consonant table will appear on screen.
English text uses the CMUdict dictionary; unknown words are processed by a fallback «letter-to-sound» algorithm. For IPA, use IPA symbols directly.
CHIMERA ↗PromptHello world, this is my synthetic voice speaking through a simulated vocal tract.
What this prompt doesEnglish phrase for Speak mode. Replace the text with your own - CHIMERA will voice it through a physical model of the air tract. You can also insert an IPA transcription, e.g. [həˈloʊ wɜːld]. - 3.To sing, switch to Sing mode - a piano roll will open. Click the grid to add notes, and type syllables under each note. Consonants are automatically shifted before the beat so the vocals land rhythmically.
Pitch can be changed with the pitch slider; body type (male / female / child) is toggled with the corresponding button and blended continuously.
CHIMERA ↗ - 4.For beatboxing, open Beatbox mode - pads and a step sequencer will appear. Each hit is an articulatory gesture (glottal burst, fricative, nasal, etc.). Place steps in the sequencer and press Play.
You can connect a MIDI keyboard: the MIDI toggle is on the control panel - turn it on if you have an external controller.
CHIMERA ↗ - 5.Experiment with the tract shape: drag the tongue, lips and jaw directly on the head cross-section. Watch the spectrum and formants update in real time. Try Bestiary mode to hear vocal tract models of a bird, seal, alligator and elephant.
If the sound «breaks» at extreme settings - this is normal: the physical model can enter unstable regimes with unrealistic tract configurations.
CHIMERA ↗
- freeCHIMERA (web app)Completely free and open source under the MIT license - runs in the browser with no registration and no payment.
Why today
AI services change fast - interfaces and free limits may differ from what's described.