the content of the audio at once for recognition. It supports multiple clients sending at the same time. https://k2-fsa.github.io/sherpa/onnx/pretrained_models ...