When I'm working with noisy audio, like something from a sporting event with crowd noise behind it, Descript often can't detect the speaker's voice and assumes there is no speaker for long stretches, or sometimes, the entire audio file. That's understandable, but it's really frustrating that Descript does not allow me to manually add a transcript to that file via "correct text" mode. It makes this audio kind of impossible to work with, especially when I'm trying to export my script to share with my editor (the script ends up just being blank there).