Sherpa-ONNX prebuilt libs
This is not a model checkpoint — it’s a distribution repo for prebuilt sherpa-onnx native libraries across versions and platforms (the file tree is organized by sherpa-onnx version and target). If you’ve ever tried to embed speech features into an app, you’ve probably hit the “native build tax”: pulling in ONNX Runtime, building for Android/iOS/Linux, and keeping those artifacts consistent across CI.
What makes this interesting is the pragmatic packaging: you can grab a tarball for a specific sherpa-onnx version and platform (e.g., Android builds that statically link ONNX Runtime, or Linux aarch64 bundles) and plug it into your application build. That’s especially handy for mobile apps or edge deployments where you want offline ASR/TTS pipelines without fighting toolchains.
If you want to try it, start by picking the exact version you’re targeting (e.g. 1.12.x) and the artifact that matches your environment (CPU vs GPU, architecture, and the ONNX Runtime variant). Then integrate the extracted shared/static libs into your build system (NDK/CMake/etc.) and confirm you’re pairing them with compatible model files from the broader Sherpa-ONNX ecosystem.
Quick stats from the listing feed: 6 likes.
Source listing: https://huggingface.co/models?sort=modified