@rilshok if you know of a way to fix these SileroVAD tests so that they don't randomly fail in CI on trying to download the model, please make another PR with relevant changes and feel free to remove the xfail marker (xfail means the CI will pass even if they fail, which could also let regressions slip through, but unfortunately I don't have the bandwidth right now to find a proper solution and this part of the codebase is not changed very often anyway).
@rilshok if you know of a way to fix these SileroVAD tests so that they don't randomly fail in CI on trying to download the model, please make another PR with relevant changes and feel free to remove the xfail marker (xfail means the CI will pass even if they fail, which could also let regressions slip through, but unfortunately I don't have the bandwidth right now to find a proper solution and this part of the codebase is not changed very often anyway).