https://x.com/VladTheInflator/status/2114453016984330274they sound just like real humans. he should have named it bixnood
>>544955812sorry wrong link this is the correct one:https://x.com/Timcast/status/2114355996453646819
>>544955812Sources tell me it is stable diffusion. Please God
>>544955874kek
>>544955874heh
>>544955874fucking kekigger.but in all seriousness imagine how fucking funny this shit would be if it answered Jeet scammers and transcribed the interaction in real time.
>>544955874Kekeke
>>544955812>bixnood
>>544955812>>544955874little timmy TT theme song: https://youtu.be/P2iB7uUzB6c
>>544955874Heh
>>544955812It would be great to finally have decent voice synthesis.But aside from that, I don't see what is particularly surprising about it.
>>544959204>I don't see what is particularly surprising about itIt is made without llm tokens meaning the entire model was trained on voice conversations. No textual input whatsoever. It's an llm but without text. Not really that impressive now that the method of making large language models is known.
>>544955812Indian scamming is going to get even worse
>>544955812So it's an advanced voice synthesizer. Which we have had for decades now. How does this differ from UTAU or Vocaloid?
>>544961054>How does this differ from UTAU or Vocaloid?It takes a lot more GPU power to do.
>>544955874zoz
>>544961131Oh so it is still AI garbage.
>>544960182Anon, it's just a waveform LLM.
>>544955874Clever joke. Have a (you)
>>544961195That's what I said.
>>544955874Kek
>>544959204*decent local voice synthesis>>544960182I seeI suppose I've only heard of multimodal models and not pure non-text models. (Though I guess sora also falls in that camp)
>>544955874
>>544961023Anyone who isn’t a retard can still tell the difference (boomers are retarded)
>>544955812Why do right wingers want an imaginary friend so badly?
