This application discloses an audio signal processing method, a model training method, and related apparatus. The audio signal processing method includes inputting a first audio input signal into a machine learning model to obtain a first howling point, and obtaining a first gain value according to the first howling point, the first howling point corresponding to an effective audio signal in the first audio input signal, and the first gain value indicating a suppression parameter of the first howling point; processing the first audio input signal according to the first gain value to obtain a second audio input signal; detecting a second howling point, and obtaining a second gain value according to the second howling point, the second howling point corresponding to a frequency band of an ineffective audio signal in the second audio input signal; and processing the second audio input signal to obtain an audio output signal.