Embodiments of the present disclosure disclose a speech interaction method and apparatus. The method includes: acquiring videos shot by a camera device in a target space and at least one channel of audio acquired by at least one audio acquisition device; determining to-be-recognized audio that respectively corresponds to each of the sound areas in the target space based on the at least one channel of audio; determining a target sound area from the target space based on the video and at least one channel of to-be-recognized audio; performing speech recognition on the at least one channel of to-be-recognized audio to obtain a recognition result; and controlling a speech interaction-targeting device in the target sound area for speech interaction in a preset mode according to the recognition result. The speech interaction method and apparatus in the embodiments of the present disclosure may detect a target object by the method of integration of an image and speech, a speech control mode corresponding to the target object is automatically entered according to a detection result, making objects on the speech recognition and corresponding speech control for various types more targeted, and helping to avoid misoperation caused by recognized sound of the target object during speech control.