harmoniq

Live demo

A quantum synth that sonifies quantum states. Build quantum circuits and hear them play.

Harmoniq maps the probability distribution of an n-qubit quantum state to sound. A playhead scrubs through a circuit step by step, and at each step the state is evaluated in both the Z-basis and X-basis. Each basis state is assigned a note from a configurable scale, and its probability controls the volume of that note's oscillator.

How it works

• Drag-and-drop circuit editor -- place gates (H, X, Z, T, control, M) on qubit wires; drop a control (●) in the same step as another gate to make it controlled (●+X = CNOT, ●+●+X = Toffoli, etc.)
• Two-basis sonification -- Z-basis drives sine/square/saw/triangle oscillators; X-basis drives a separate layer one octave up by default
• Measurement gate -- collapses a qubit mid-circuit
• Live visualization -- probability bar chart, real-time waveform oscilloscope, state vector display, entanglement and correlation meters

Quantum simulation

All quantum simulation runs through Quantum Forge via the quantum-forge npm package. There is no custom quantum math in this project -- harmoniq is a pure consumer of the public quantum forge API.

Setup

• ensureLoaded() loads the quantum forge WASM binary
• useQuantumForgeBuild("qubit") selects the qubit-optimized variant (d2n20)
• The vite plugin serves WASM files during dev and copies them to dist/ on build

Per-step evaluation

A single QuantumPropertyManager lives for the lifetime of the engine. As the playhead advances, the engine applies new gates incrementally; on loop, qubit reset, or circuit edit, properties are released back to the pool (which resets them to |0⟩) and re-acquired.

1. manager.acquireProperty() allocates n qubit properties, each starting in |0⟩
2. For each circuit step, gates are batched and applied via the WASM module (manager.getModule()). Controls in the same step become predicates for every non-control gate at that step:
   • m.hadamard(prop, 1, predicates) -- H (controlled if predicates present)
   • m.cycle(prop, 1, predicates) -- X
   • m.clock(prop, 1, predicates) -- Z
   • m.clock(prop, 0.25, predicates) -- T (π/8 phase)
3. m.probabilities(qubits) reads the Z-basis probability distribution without collapsing the state
4. Hadamard is applied to all qubits, probabilities are read again (X-basis), then undone with m.inverse_hadamard()
5. m.reduced_density_matrix(qubits) extracts the full density matrix, from which harmoniq reconstructs the state vector
6. m.forced_measure_properties([prop], [value]) handles measurement gates by forcing specific outcomes to simulate branching into a mixed state

Development

npm install
npm run dev

Contributing

Contributions are welcome via pull request. Please open an issue first for anything substantial so we can discuss the approach.