v21’s Twitter Archive—№ 54,106

hot take: one of the biggest problems with speech synthesis is that the output stresses the wrong words - either you need to do elaborate and delicate annotation or machines need to have something like a sense of the meaning of what they read