3 ms·
Synthetic Speech Generated from Brain Recordings
- est31 7y agoPrior discussion: https://news.ycombinator.com/item?id=19740923 https://news.ycombinator.com/item?id=19740923
- tomcam 7y agoRequires an implant, but cleverly they are using implants already required to help suppress epileptic seizures. Accurate to about 47% of a 50 word vocabulary, and only that high because they’re decoding signals thrown off by subvocalization.
- ralusek 7y agoThere's going to be something like Occlumency from Harry Potter. To avoid people reading your thoughts, you will need to practice thinking nonverbally. I realize that I already think pretty nonverbally, I don't really have an inner monologue so much as an inner visualization. Come at me NSA, you got nothing.
- andai 7y agohttps://youtu.be/jsp1KaM-avU https://youtu.be/jsp1KaM-avU
- pontifier 7y agoI've wondered whether meaning from heard words is partially decoded through a similar process. When actively listening to speech sounds, I sort of feel what it would feel like to mimic the words I hear, and this enhances my understanding of the speech. I also find that I often subvocalize the text I read. It's like the meaning of the words are somewhat connected with the pattern required to voice it.
- sitkack 7y agoIs not speech and language partially the process of replaying thought processes so that the receiver can experience the same path and progression. We don't communicate facts so much as paths through a winding trail of thought. Maybe if our neural networks had a more rigid pattern, the same activation fronts could be copied from person to person, creating cross person feelings and experiences.