The present invention discloses a subtitle generation method for Chinese online audio and video, including the following steps: S1, audio data extraction step, the server receives audio and video files, extracts audio data and converts it into a standard format; S2, noise reduction step, audio data Carry out noise reduction processing, obtain audio file; S3, data segmentation step, audio file is carried out endpoint segmentation, obtains audio sample; S4, segment recognition step, is further segmented to obtained audio sample, obtains voice segment, then Recognize the speech segment, sort out and obtain the recognition results of all audio data; S5, the subtitle generation step, integrate and analyze the text and the corresponding time axis, obtain the subtitle file, and match the subtitle with the audio data according to the generated subtitle file. The method of the invention can automatically complete the voice recognition of audio and video information and subtitle generation, effectively making up for the shortcomings of traditional manual shorthand in conversion efficiency in subtitle generation.