Far-field pickup speech recognition method and system based on self-organizing microphone array

The invention relates to a far-field pickup speech recognition method based on a self-organizing microphone array, and belongs to the technical field of speech recognition. Comprising the following steps: S1, establishing a point sound source and a self-organizing microphone array comprising M microphones, and collecting voice information through the self-organizing microphone array; s2, preprocessing the voice signals with different intensities acquired by the microphones of different channels by using a filter with a uniform threshold value; s3, filtering noise reduction and voice enhancement are carried out by adopting an improved filtering mode combining adaptive filtering and Wiener filtering; and S4, voice recognition is carried out through the trained WeNet network framework. According to the scheme, the recognition accuracy of a voice recognition algorithm under long-distance and complex noise backgrounds can be improved. 本发明涉及基于自组织麦克风阵列的远场拾音语音识别方法，属于语音识别技术领域。包括以下步骤：S1、建立一个点声源和一个包含M个麦克风的自组织麦克风阵列，通过自组织