According to one aspect of the invention, there is provided a method for providing a voice recognition trigger, comprising the steps of: estimating a first distance, which is a distance between a device and a user, on the basis of proximity information detected by the device, and estimating a second distance, which is a distance between the device and a location where a voice detected by the device is uttered, with reference to information on the voice detected by the device; and determining whether the voice detected by the device is an object of voice recognition, with reference to similarity between the first distance and the second distance.