The simplest approach to use PaddleSpeech models.
paddlespeech help
paddlespeech asr --input ./test_audio.wav