Abstract: For most military speech datasets, only “weak” textual annotations are available meaning that the content of the speech is known but the exact on- and offset of each word is not. This is ...
Some results have been hidden because they may be inaccessible to you
Show inaccessible results