dawn zealot
One of the reasons Neuro and evil stick to thier role even when the human part of the stream seems to be down,angry,sad or wants to speak about a serious stuff.
Is, that neuro and evil cant detect emotions in the voice of a human they take everything like they are reading it from a book.
Here are some projects that maybe can help with that https://medium.com/@marko.briesemann/emotion-recognition-with-ai-c7f831332ed3
calm mauve
I think this might be solved with intelligence upgrades and time
quasi maple
I think this might be solved with intelligence upgrades and time
There’s only so much you can get from text. Even humans can’t fully understand emotions through it.
calm mauve
Oh right she doesn't just get the audio she has to convert it to text I'm stupid
pliant osprey
She actually can recognize some sounds so... maybe we're on the right track? But maybe is just another AI saying: "Someone clapped"
So... she actually needs to... "listen" like us 
More work for vedal...
rich fractal
She’ll probably be completely multimodal kinda like chatgpt 4o eventually when its cheaper and more efficient to do so. Rn tho, something like that probably adds latency and would somehow be unethical or something
calm mauve
Am i missing something or isn't she already multimodel
dim estuary
imagine the l*tency