/tech/ - Technology★

we are so close to the singularity that I can't tell if this belongs in here or /x/.

Is no one big brained enuff?

>>1017747 (OP)

Looks like a shitty version of Loglan/Lojban

>(((USCIIIIII)))

>not posting the originals

https://archive.md/YJs2r

https://archive.fo/It5Jk

>>1017817

Thanks the gods for your big mind brain, this smol brained nibba is forever in your debt

>>1017823

I also should add that what I describe here (or what he is claiming to have made, if I'm right) is entirely possible, but I'll believe it when I see it. Not aware of any attempts to do something like this outside of vocal synthesis tech, but this is more of a linguistics endeavor, and I'm not intimately familiar with the field.

>>1017825

Anything is possible if you set vague enough expectations! Believe in yourself!

>>1017825

I think we'd have better luck if we devised a way to capture the behavior of human vocal chords (something akin to motion capture but for the parts of the body responsible for speech and vocalization), trained a deep belief network on a representative sample of people speaking normally, but using the same script of phrases in multiple languages as well as reading a script of our best guesses at the full set of vocalizations that do not correspond to any specific language. Correlating binary representations of the possible positions of the human vocal chords to the binary representations of the auditory phenomenon they create would be quite the undertaking, but once accomplished could be fine tuned by taking multiple instances of this original generalized model and training it to reproduce it's best guess as to what the human vocal chord position and motion would be for each snippet. Maybe if we had access to PRISM's records of voice conversations. We could then further develop the model by placing sensors on the heads of test subjects and correlating their brain activity with the auditory data and vocal positioning data. This could lead to the ability to predict which parts of the brain are activating when someone is speaking, and perhaps even the ability to make inferences about what they are thinking depending on their tone and choice of words. This would not be dissimilar to how micro-expressions can reveal people's emotional state despite their best attempts to hide it.

>>1017830

What do you mean by having better luck, and what is exactly the end goal of the project? It sounds like your intentions are quite different from what interested me in this thread, but it's still interesting nonetheless. Are you the owner of the website, or a third party?

You aren't going to capture the majority of human speech by only monitoring the vocal chords. You will need to be able to capture the resonance of the mouth as well. Many of the vocalizations that we produce will look mostly the same with regard to the vocal chords, which are basically responsible for the tone and pitch of speech. The resonant properties of the mouth are what produces vowels and the other subtler variations of vocalization that are necessary for language. (This is why I brought up formants in my previous post)