ChatGPT makers OpenAI seem to be on a roll as the company just unveiled its own voice-cloning initiative.
While the project is said to be both useful and intimidating at the same time, it’s only up for grabs to a limited number of the company’s partners, as of right now.
The latest endeavor is dubbed Voice Engine and it’s designed to clone a user’s voice by simply hearing a voice note comprising just 15 seconds. Additionally, the voice that’s replicated is designed to convey all sorts of emotions and intuitive human speech along the way as well.
This makes the whole AI-themed dialogue very realistic in nature as well. Moreover, the firm adds how it was initially curated in the latter part of 2022 to better support the AI tool, ChatGPT, and its inert ability to convert text to speech. But instead of rolling out Voice Engine to the masses, it has opted to take a halt in that regard as perhaps the world is not quite ready, just yet.
The company rolled out a statement on the matter recently and spoke about how they hope to take on a more cautious and careful approach that’s designed to better cater to respective target audiences.
This carries the potential for voice misuse as mentioned in the blog post that revealed several types of Voice Engine in effect.
Speech generated through AI means is quite different from that seen in referenced audio. And that is the scary part because it does come with its fair share of threats online.
We’re talking about examples where it could be placed in the wrong hands and therefore sure technology might be utilized to remove deepfakes and hence can misinform the audience. However, keeping the risks of misuse on the side, the company feels Voice Engine might be a really useful and successful endeavor for society today.
Whatever the case may be, it’s now getting rolled out in users’ direction from all over the globe including a small figure of trusted individuals. Meanwhile, the company is currently boasting about providing reading assistance to a group of school kids. This would serve as translators too as one person’s voice is made use of to communicate in various kinds of languages.
Meanwhile, another case entails giving rise to voice cloning tech for those individuals who lose their voice, which happens to be a concept that’s quite similar to tech giant and iPhone maker Apple.
Furthermore, the company says it is rolling out Voice Engine through preview modes so that partners can benefit, even if they promise never to make use of this technology for reasons dubbed unauthorized like when impersonating the other.
Partners are said to disclose the respective audience and the voices that are being heard by the software as ‘AI-produced’ to ensure complete transparency when using the technology. In the same manner, OpenAI has added watermarking systems to better detect AI-based audio through Voice Engine.
Still, the company is not providing any kind of guarantee about this being rolled out as a bigger release for all. The voice cloning program is a trial and if things go as planned, we shall be seeing it released across the board. But until then, it all depends on how the whole society generates a response to the growing popularity of AI.
Did we mention how technology is already making it hard to differentiate fact from fiction?
Let’s wait and watch for this experiment to roll out and then see what sort of response comes from the company’s partners.
Image: DIW-Aigen
Read next: Google Has Launched a New AI System in DeepMind Called SAFE that is Claimed to be Better than Humans in Fact-Checking
While the project is said to be both useful and intimidating at the same time, it’s only up for grabs to a limited number of the company’s partners, as of right now.
The latest endeavor is dubbed Voice Engine and it’s designed to clone a user’s voice by simply hearing a voice note comprising just 15 seconds. Additionally, the voice that’s replicated is designed to convey all sorts of emotions and intuitive human speech along the way as well.
This makes the whole AI-themed dialogue very realistic in nature as well. Moreover, the firm adds how it was initially curated in the latter part of 2022 to better support the AI tool, ChatGPT, and its inert ability to convert text to speech. But instead of rolling out Voice Engine to the masses, it has opted to take a halt in that regard as perhaps the world is not quite ready, just yet.
The company rolled out a statement on the matter recently and spoke about how they hope to take on a more cautious and careful approach that’s designed to better cater to respective target audiences.
This carries the potential for voice misuse as mentioned in the blog post that revealed several types of Voice Engine in effect.
Translating content : You can use this for translation in tk multiple language. You can clearly see the voice is close to refrence audio in all languages.
— AshutoshShrivastava (@ai_for_success) March 29, 2024
HeyGen is one the ealry adopter of this. pic.twitter.com/86V5E5tpQ7
Speech generated through AI means is quite different from that seen in referenced audio. And that is the scary part because it does come with its fair share of threats online.
We’re talking about examples where it could be placed in the wrong hands and therefore sure technology might be utilized to remove deepfakes and hence can misinform the audience. However, keeping the risks of misuse on the side, the company feels Voice Engine might be a really useful and successful endeavor for society today.
Whatever the case may be, it’s now getting rolled out in users’ direction from all over the globe including a small figure of trusted individuals. Meanwhile, the company is currently boasting about providing reading assistance to a group of school kids. This would serve as translators too as one person’s voice is made use of to communicate in various kinds of languages.
Meanwhile, another case entails giving rise to voice cloning tech for those individuals who lose their voice, which happens to be a concept that’s quite similar to tech giant and iPhone maker Apple.
Furthermore, the company says it is rolling out Voice Engine through preview modes so that partners can benefit, even if they promise never to make use of this technology for reasons dubbed unauthorized like when impersonating the other.
Partners are said to disclose the respective audience and the voices that are being heard by the software as ‘AI-produced’ to ensure complete transparency when using the technology. In the same manner, OpenAI has added watermarking systems to better detect AI-based audio through Voice Engine.
Still, the company is not providing any kind of guarantee about this being rolled out as a bigger release for all. The voice cloning program is a trial and if things go as planned, we shall be seeing it released across the board. But until then, it all depends on how the whole society generates a response to the growing popularity of AI.
Did we mention how technology is already making it hard to differentiate fact from fiction?
Let’s wait and watch for this experiment to roll out and then see what sort of response comes from the company’s partners.
Read next: Google Has Launched a New AI System in DeepMind Called SAFE that is Claimed to be Better than Humans in Fact-Checking
