DEFINITION

ONNX

Open Neural Network Exchange

Definition

An open-source format for representing machine learning models that enables deployment across different platforms and runtimes. ONNX models are lightweight, portable, and can run on CPU with minimal latency, making them ideal for edge and real-time inference scenarios.

Examples in the Wild

  • Example 1:1MB voice gender classifier running in 4ms on CPU
  • Example 2:Silero VAD (Voice Activity Detection) models in ONNX format
  • Example 3:Cross-platform deployment of ML models in production voice AI systems