per editor: Bi Luming
According to reference news on the 13th, citing the American Interesting Science website, which reported on July 10, scientists said that the new artificial intelligence (ai) speech generator developed by Microsoft "has reached a level comparable to that of humans" ...but it is too dangerous to be released publicly."
reported that vall-e 2 is a text-to-speech (tts) generator, can reproduce the speaker's voice using only a few seconds of audio. The effect is so realistic that it cannot be released to the public.
Microsoft researchers said in a paper published on the preprint repository on June 17 that vall-e 2 can "accurately and naturally generate speech that is exactly the same as the original speaker, and the speech it generates is comparable to that of humans. Comparable". In other words, this new AI speech generator is lifelike enough to fool people into thinking it’s a real person speaking—at least that’s what its creators think.
Image source: Visual China
Microsoft researchers wrote in the paper: "vall-e 2 represents the latest progress in neural encoding and decoding language models. It marks a milestone in zero-shot text-to-speech synthesis and reaches the same level for the first time. "Comparable to human speech."
The paper also says: "Furthermore, vall-e 2 is able to consistently synthesize high-quality speech, even for sentences that are traditionally challenging due to their complexity or repeated phrases."
The researchers used audio samples from the speech libraries librispeech and vctk to evaluate how similar Vall-E 2 was to recordings of the speaker. They also used ELLA-V, an evaluation framework for measuring the accuracy and quality of generated speech, to determine how effectively VALL-E 2 can handle more complex speech generation tasks.
researchers wrote: "Our experiments based on the librispeech and vctk datasets show that vall-e 2 surpasses previous zero-shot TTS systems in terms of speech robustness, naturalness, and speaker similarity. It is "The first system of its kind to achieve human-level performance on these benchmarks,"
Microsoft researchers wrote in a blog post: "vall-e 2 is purely a research project. At this time, we have not implemented vall." -e2 plans to incorporate it into products or make it available to the public, there could be potential risks in terms of the model being misused, such as tricking speech recognition systems or impersonating a specific speaker."
However, they did say that the AI voice technology is There may be practical applications in the future. The researchers said: "vall-e 2 can synthesize speech that maintains the identity of the speaker. It can be used for education and learning, entertainment, journalism, self-created content, accessibility features, interactive voice response systems, translation, chat robots, etc. field. "
According to public reports, on June 19, at the 2024 Cannes Lions International Festival of Creativity, Tesla CEO Elon Musk accepted the CEO of WPP, a world-renowned marketing services company. Interview with mark read. When talking about "whether artificial intelligence will really handle everything for us," Musk said: We cannot be blindly optimistic or overly pessimistic. I strongly agree with Jeff Hinton (the godfather of artificial intelligence). He believes there is a 10% to 20% chance that some worrisome scenario will arise. However, we should pay more attention to the 80% positive possibility.
Therefore, I think the most likely future is that we will enter an era of extreme material abundance, and goods and services will be available to everyone on the planet. This doesn’t just mean universal basic income, but high income for everyone. Work will become optional and people will have the freedom to choose whether to work or not.
Such a prospect, while alluring, could also trigger a crisis about the meaning of life. When artificial intelligence can do everything we can do, and even do it better, what is the motivation and meaning of what we do? This can lead to an existential confusion as to why we do anything at all?
So, I think we are heading into an era of unprecedented prosperity. At the same time, we are also in the most interesting moment in history.There’s a saying that goes, “May you live in interesting times.” It may seem like a curse, but I think we’re in such an era right now. I mean, even if faced with the worst possible outcome of artificial intelligence, which is the annihilation of the human race, I would choose to face it rather than run away from it. I think I might actually want to see this unfold.
Daily economic news comprehensive reference news, public information
Daily economic news
-
-
-
-
-
-
-
-
-
-