Listen to Google DeepMind generating human speech, composing music, plotting SkyNet takeover - Polidicks

[QUOTE=TurtleeyFP;51032145] [vid]https://storage.googleapis.com/deepmind-media/pixie/knowing-what-to-say/first-list/speaker-1.wav[/vid] [vid]https://storage.googleapis.com/deepmind-media/pixie/knowing-what-to-say/first-list/speaker-5.wav[/vid] [/QUOTE] Huh... English really does sound like a mix of Germanic and Latin languages.

We could probably immortalize voice actors with this. Imagine years after the original voice actor is dead you can still have him speak lines for free. you can bring Mel Blanc back from the dead with this.

I want to use it for Text to Speech. I really suck at reading stuff, and natural Text-to-Speech software are usually wayy overpriced. This sounds great.

We're rapidly approaching the point where machines will be indistinguishable from humans. What comes after will be [I]very[/I] interesting.

[QUOTE=ZombieDawgs;51041105]We're rapidly approaching the point where machines will be indistinguishable from humans. What comes after will be [I]very[/I] interesting.[/QUOTE] I wouldn't say rapidly, due to the fact that robots are still many years off from being able to talk with humans as if they themselves are humans. We're approaching, but I wouldn't say super rapidly.

[QUOTE=ZombieDawgs;51041105]We're rapidly approaching the point where machines will be indistinguishable from humans. What comes after will be [I]very[/I] interesting.[/QUOTE] progressive-democratic party platform 2064: android rights libertraian party platform 2064: android segregation

[QUOTE=Daniel Smith;51036871]Huh... English really does sound like a mix of Germanic and Latin languages.[/QUOTE] My first thought was a new way for the Sims to generate their gibberish language.

[QUOTE=phygon;51041214]I wouldn't say rapidly, due to the fact that robots are still many years off from being able to talk with humans as if they themselves are humans. We're approaching, but I wouldn't say super rapidly.[/QUOTE] i give it ten or twenty years, which is fucking fast as hell considering what we had 100 years ago

[QUOTE=Gmod4ever;51041254]My first thought was a new way for the Sims to generate their gibberish language.[/QUOTE] simese

[QUOTE=ZombieDawgs;51041105]We're rapidly approaching the point where machines will be indistinguishable from humans. What comes after will be [I]very[/I] interesting.[/QUOTE] I don't think we have robots that are as complicated as a living mouse yet or a similar simple minded creature for that matter. I think it's a long way off still. I think we might have the tech to replicate voices perfectly soon, though and I wonder what this will say about the future of the voice acting industry. Maybe the voice acting of the future will just be you "leasing" out your voice, like they would take it and you would just get money while they use programs to manipulate your voice to read out the script naturally. I wonder if anyone will ever have a voice and voice something a cartoon that gets popular without them thinking it would be which would end up with them not being able to voice in anything after that since they sold the "right" to use their voice to some company :v:.

[QUOTE=phygon;51041214]I wouldn't say rapidly, due to the fact that robots are still many years off from being able to talk with humans as if they themselves are humans. We're approaching, but I wouldn't say super rapidly.[/QUOTE] These developments have come around in the last decade, give us another 10-20 years and we'll either be there or well on our way; that's pretty fast.

[QUOTE=phygon;51041214]I wouldn't say rapidly, due to the fact that robots are still many years off from being able to talk with humans as if they themselves are humans. We're approaching, but I wouldn't say super rapidly.[/QUOTE] There already exists algorithms that can learn itself basic algorithms from input data. This algorithm is called neural network based Turing machine. (and it was invented around 2014..)

[QUOTE=Fourier;51043112]There already exists algorithms that can learn itself basic algorithms from input data. This algorithm is called neural network based Turing machine. (and it was invented around 2014..)[/QUOTE] Well yes but convincing humans at a machine is a real human in realtime is more difficult than regression like that (If I'm using the right term). It's similar, but vastly more complex. [QUOTE=ZombieDawgs;51043013]These developments have come around in the last decade, give us another 10-20 years and we'll either be there or well on our way; that's pretty fast.[/QUOTE] When I hear people talking about "rapidly advancing tech" I usually figure that they don't know what they're talking about and that they're speaking in the manner of like 5 years ~15 years seems reasonable, even if only within the context of supercomputers

[QUOTE=phygon;51043318]Well yes but convincing humans at a machine is a real human in realtime is more difficult than regression like that (If I'm using the right term). It's similar, but vastly more complex. [/QUOTE] True, non-idiot human could easily test the AI subject if it is human.

very interesting, i have an assignment where i have to program a basic neural network at the moment.

Neural networks are great at predicting outcomes but are terrible at creating. Ergo, a chatbot is a terrible use for a NN, while a TTS is a great use for one (because you have existing input)

[QUOTE=Black;51042432]I don't think we have robots that are as complicated as a living mouse yet or a similar simple minded creature for that matter. I think it's a long way off still. I think we might have the tech to replicate voices perfectly soon, though and I wonder what this will say about the future of the voice acting industry. Maybe the voice acting of the future will just be you "leasing" out your voice, like they would take it and you would just get money while they use programs to manipulate your voice to read out the script naturally. I wonder if anyone will ever have a voice and voice something a cartoon that gets popular without them thinking it would be which would end up with them not being able to voice in anything after that since they sold the "right" to use their voice to some company :v:.[/QUOTE] I doubt human actors will be replaced in any sense because we still haven't breached the uncanny valley in any real way. Further, if you look back even a year ago, you can see the wrinkles and issues with top of the line graphics trying to mimic real life. Its the same kind of argument over CGI and Practical. Plus we already use this kind of tech to simulate large gatherings of people to avoid having to actually pay actors.

I don't really like talking about my mental health in public but the no guidance speech sounds a lot like what I hear when hearing voices. I have it with a more Norwegian accent though.

That music is pretty intimidating tbh.

The name of the project, deepmind, sounds pretty intimidating as well

[QUOTE=Zombie_2371;51049285]The name of the project, deepmind, sounds pretty intimidating as well[/QUOTE] Maybe it's based off Deep Thought, from Hitchhiker's Guide to the Galaxy