Air-ground conversation alarm prediction method and system, and storage medium

The invention discloses an air-ground conversation alarm prediction method and system and a storage medium, and the method comprises the steps: obtaining a voice signal which is an audio data signal collected in real time; after converting the voice signal into a digital signal, carrying out denoising processing on the digital signal to obtain a sub-voice signal; processing the sub-voice signals by adopting a CNN-based twin neural network model to obtain a semantic text and a data type corresponding to the voice text; according to the data type of the semantic text, a standard voice signal corresponding to the semantic text is matched, the standard voice signal and the semantic text are compared and judged, and an air-ground notification alarm prompt is obtained; the beneficial effects of the invention are that the audio signal processing time can be reduced and the safety of air-ground flight can be ensured through a method of converting the audio data collected in real time and comparing the audio data with