Voice endpoint detection method and system

The invention provides a voice endpoint detection method and system, and the method comprises the steps: constructing an endpoint detection model based on a neural network, and the endpoint detection model comprises a preprocessor, a time domain encoder, a frequency domain encoder and a decoder; preprocessing the audio signal by using a preprocessor to obtain an audio time domain vector and an audio frequency domain vector; encoding the audio time domain vector by using a time domain encoder, and extracting a time domain feature vector; encoding the audio frequency domain vector by using a frequency domain encoder, and extracting a frequency domain feature vector; and decoding the time domain feature vector and the frequency domain feature vector by using a decoder to obtain a voice endpoint. The voice endpoint is recognized by using the time domain feature and the frequency domain feature of the audio signal, the robustness is high, the human voice and the environmental noise can be accurately distinguished