PPASR provides an end-to-end Chinese speech recognition framework supporting both streaming and non-streaming inference modes. It implements multiple neural network architectures including DeepSpeech2, Conformer, Squeezeformer, and Efficient Conformer with various decoding strategies such as CTC greedy search, prefix beam search, and attention rescoring. The project runs on PaddlePaddle deep learning framework and can be deployed on servers and edge devices.