Neuro Sings So Good That The Lavalamp Turned Red | Circles by Kira | Reaction !!!

[@eijiktna7745]

Fix memory leak's? Nah just buy more ram - Vedal best tech V-tuber of the year 987

[@omargoodman2999]

I've been very interested in Neuro's progress (both Neuros' progress, really), because it really intersects my interests in AI, Consciousness, as well as _Linguistics_ oddly enough. I find a big disconnect a lot of people run into in discussing AI, as well as Consciousness in general, is lack of understanding of terminology. This results in confusing terms and equivocating one word for another in discussions. And when Person A is thinking Concept A, but describes it using Word B, and then Person B hears Word B and links it to Concept B in their own mind, then they each have a _somewhat_ related, but still distinctly different, concept in mind; Concept A for Person A and Concept B for Person B. But in verbal discussions, they both use Word B to encode both concepts. Then, when others hear the discussion and try to join in, they start picking sides based on their views of Word B and how it relates to Concept B in their own mind. We often describe AI behaviors in terms of how "Human" they are; how close to "Human" can an AI be, AI can be "smarter" than a "Human", AI are better suited for these kinds of tasks compared to a "Human", etc. But "Human" is merely one singular species; a taxonomical categorization of an organism. It's less a matter of "Human", and more a matter of specific qualities which we've only observed, recognized, or acknowledged as being possessed or expressed *by* Humans that we're actually talking about. For a long time, it was believed that Humans, and _only_ Humans, experienced such things as emotions, self-awareness, creativity, intelligence, sentience, and sapience. However, with time, study, and observation, it was gradually discovered that other organisms besides Humans could have and express some of these qualities as well; just to lesser degrees. It wasn't an all-or-nothing prospect, there were gradations. In fact, if there are levels to it, we aren't necessarily even the highest _possible_ level of any of those qualities; merely the highest *achieved* level (and even that is questionable for some). When you try to define "Humanity" by their capacity to produce art and literature, you risk de-humanizing anyone who's not an artist or writer. "Can a robot write a symphony? Can a robot take a blank canvas and turn it into a masterpiece?" "Can you?" Many argue about the nature of what an AI LLM does; it procedurally constructs its dialogue by deciding what words or phrases to say next based on the words and phrases that have been said previously. If it loses track of what has already been said, it might ramble into an unrelated topic. If it tries to track too much information at once, it might conflate multiple trains of thought and get them confused, combining them. Ideally, it wants to strike a balance of staying on-track with a conversation that's cohesive and sensible. But even *Humans* can have difficulty with that. I even argue that this is the _exact_ process that Humans use in constructing conversations; most simply don't bother paying attention to their own process. I, however, _have_ self-analyzed how I actually *produce* both what I write as well as what I speak, in real-time as well as reflectively. So I can say, as something of a subject matter specialist, that at least _some_ kinds of Humans do *indeed* procedurally generate their dialogue by deciding on words and phrases to "go next" *just like* an AI would. The only real difference is the materials used; carbon and saltwater vs silicon and lightning. We're *all* Neural Networks and LLMs. We just understand much better how the silicon-based version works. And we have barely any real comprehension at all how the silicon-based version works. Last point to make is this. Some will point out that an AI can't be self-aware, conscious, sentient, sapient, or whatever is the topic of the day and rationalize their argument on the basis that it isn't actually "experiencing" reality, it just has "information" in the form of memory and makes decisions on what to do or say based on that information of past data. And yet, we have no real solid evidence that Humans aren't doing that as well. Any person can validate and accept their own self-aware existence; the mere act of _questioning_ one's existence irrefutably proves it because *something* must exist to do the questioning. But it can only be self-verified; it's impossible to share this verification with anyone else and also to validate anyone else's existence by the same method. That, however, merely proves _existence,_ not the *nature* of existence; you _still_ could merely be acting on "information" of past events and making decisions on what to do or say based on that information. If that can't be considered "real experience" for an AI, then by logical parity it cannot for oneself either. Or, to put it another way: you, yourself, could constantly, moment to moment, cease to exist; just to immediately be replaced by an exact copy of your "self" complete with all the same memories. Does that mean that the "you" right now is not the "you" at _this_ moment, which is a totally different one from the "you" by the time you get here? If so, what even *are* "you"? Trippy, innit?

[@Jackxagon]

it really is better. On this same stream, Neuro sang the same song she did on her first V3 Karaoke back in June last year ("World is Mine"). you can really tell the difference, even if both are "V3", like how she has more "strenght" in her voice. (Also, that lava lamp was Vedal messing up with an Idea of it being a Music Visualizer of some sort, someone on Discord came with the Idea and he tried it out on the spot)

[@strim1987]

But Vedal optimized it - with the help of Amazon. Now it has enough RAM to give the lamp new functions

[@CitrusFriendTV]

The context for this instance of the Lamp turing red was supposed to be similar to a light show turning different colors during the song. He mentioned it in the discord. The clipper clickbaited this clip lmao (had nothing to do this the ram usage this time)

[@How_Many_Monkeys]

Overanalyzing, but the beat of the song mimics the beat of a human heart. Something Neuro will never have. Which also echos the entire point of the song.

[@peterpham2424]

love the commentary in this vid, great

[@strim1987]

Probably my biggest expectation right now is the possibility of Neuro's original songs appearing. Vedal has already done something similar for his other project, he has a good musical sense, obviously, the only problem is that the original song requires disproportionately more effort compared to the cover. He's still trying to make good on the promise he made back in January to write at least one, but who knows how long it'll take. By other project I mean Neuro ARG, for which Vedal wrote "I Don't Want to be an Engineer". In fact, they are related, but he used a different character and a different voice, so Neuro still doesn't have an original song

[@Rose_in_Blue]

Holding out for Neuro-sama one day covering a song by _neurosama xD

[@tenkur700]

I have been following neuro for a long time, and it's still boggles my mind of how Videl does the stuff he does originally followed neuro because I was interested in the AI part of it now I'm more interested in how Vidal does what he does and the song is fire🔥

[@strim1987]

I think the main progress over the past six months is not that the level of singing has improved dramatically, but that the bar for “the best of the best” has begun to be reached more often, and the possibilities for processing original material have expanded. I think the main work now will be on stability and greater individuality of the voice, as well as expanding capabilities, such as whispering or low tones

[@strim1987]

Speaking of new abilities, Voices of Svaahaa is the absolute peak of it. Still not perfect, but if you know the features of Neuro well, god, how much has been achieved there

[@meme-eg8uw]

you can really tell that the actual text to speech is not like pre-recorded because the RAM usage goes up and down as it's talking