Web and Media Search Lab
Web and Media Search Lab
The goal of this component is to perform the synchronization of the all audio files in an existing dataset.
Option 1:
Option 2:
On top of the txt files listed bellow, both options cut the audio files in the dataset according to the segments and save them in “sounds_cut/”.
Example: “sounds_cut/1/10.0” has all cut audio files that exist in cluster 1 starting on 10 seconds until the time stamp where the next segments starts.
Option 1:
Option 2:
Input:
Output:
For testing and validation purposes, a dataset with recordings crawled from YouTube was used. Each song had one professional recording (BBC) and a set of user-recordings.
Number of songs: 10 Number of samples: 90
