NeuralEngine

Transforming Natural Language into real-time software actions through local ONNX-driven intent classification.

Capture

Whisper AI Transcribing

Interpret

ONNX Intent Analysis

Discover

Property Mapping

Execute

Zero-Latency Dispatch

Supported Intents

Examples of natural language commands mapped via ONNX

Transport

"Start recording"
"Set BPM to 124"
"Quantize track"

Mixing

"Mute master"
"Add Reverb to track 4"
"Low-pass at 500Hz"

Theory

"Create Cmaj7 chord"
"Transpose +1 octave"
"Generate bassline"

Modular Intent Injection

NeuralEngine utilizes the IIntentHandler interface, allowing for seamless injection into the system core.

This architecture decouples the AI interpretation from the target host, making it possible to control Ableton, Logic, or any external software with the same neural logic.

Real-time Discovery

Our DiscoveryService dynamically maps intent to software properties.

By processing ONNX models locally, we match transcribed voice commands to live-parameters in real-time, eliminating the need for static MIDI mappings or cloud processing.

Musical Intent to Execution

The NeuralEngine isn't just a bridge; it's a context-aware controller. By combining OpenAI Whisper for transcription and custom ONNX models for classification, we enable a fluid workflow where your voice directly manipulates the DSP environment. No cloud, no latency—just intelligent, local execution.

.NET 10Whisper.NETONNX RuntimeAbleton LiveOSC

Neural Harmonic Engine

Test Neural Engine directly. Click the microphone button and see the result. For now its only mockup data, but the call to the service is real. When the service runs locally this will also be a Live service.

.NET 10Ableton LiveMidi generation