Abstract: Using larger scale datasets in the training stage of speaker verification model usually leads to better performance. However, when the speaker number of the training dataset becomes extreme ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results