The embodiment of the invention provides a voice recognizing method. The voice recognizing method comprises the following steps of splitting to-be-recognized voice into a plurality of voice paragraphs; respectively carrying out voice recognizing on the plurality of voice paragraphs by adopting the parallel manner; and integrating the voice recognizing results for the plurality of voice paragraphsto form a voice recognizing result corresponding to the to-be-recognized voice. According to the technical scheme, the long voice data are split to form the plurality of voice paragraphs, voice recognizing is respectively carried out on the plurality of voice paragraphs formed through splitting by adopting the parallel manner, and the obtained plurality of voice recognizing results are integratedto form the voice recognizing result corresponding to the long voice data. By spitting the long voice data into the plurality of voice paragraphs, a plurality of computing resources of a server can besufficiently utilized for carrying out voice recognizing simultaneously, and thus the voice recognizing efficiency is greatly improved.