The rest of this stack, pointed at images, audio, text or a robot. The same shape repeats in each: running a model is the covered part, and the classical, non-learned half is where shelves thin out.

Speech has the most complete single answer, sherpa-onnxsherpa-onnx · Speech & audioversion1.13.8downloads347k / 90dstars14.8klicenseApache-2.0last commit2026-09-14published2026-09-11dependents15 owners Open sherpa-onnx in the catalogLast scraped on Sep 14, 2026, which ships official Rust examples for recognition and for several text-to-speech models1. Robotics is where Rust reached into a stack it did not start in: ROS 2 lists a Zenoh middleware among its own implementations2, and Zenoh is Rust.

  1. k2-fsa/sherpa-onnx — the toolkit's own Rust examples
  2. ROS 2 documentation — the ROS 2 project's own list