| 摘要 |
A system and method of identifying a sound signal, by receiving a sound signal; converting the sound signal into an audio signal; identifying a first keypoint in the audio signal; calculating first keypoint data from the audio signal, the first keypoint data being associated with the first keypoint; comparing the first keypoint data to at least one stored keypoint data, the at least one stored keypoint data each being associated with at least one audio recording; determining a first similarity value for each audio recording based on said comparison; assigning a likeness score to each audio recording whose first simlarity value is within a threshold range; identifying a second keypoint in the audio signal; calculating second keypoint data from the audio signal, the second keypoint data being associated with the second keypoint; comparing the second keypoint data to the at least one stored keypoint data; determining a second similarity value for each audio recording based on said comparison of the second keypoint data; updating each likeness score based on the second similarity value; identifying at least one of the audio recordings as a candidate audio recording, based on the updated likeness score; and providing the identity of the identified candidate audio recording as the identity of the received sound signal. |
