Voice generation AI ``Voicebox'' developed by Meta can ``read sentences with other people's voices without permission'', Meta avoids public disclosure because it is too dangerous



Meta, which operates Facebook and Instagram and is also focusing on AI research, announced a speech generation AI `` Voicebox '' on June 16, 2023. If you use Voicebox, you can not only read out the sentences you entered, but also perform operations that are likely to be abused, such as ``editing part of the audio'' and ``reading sentences with other people's voices without permission''. Meta acknowledges the potential for Voicebox to be abused and refrains from publicly disclosing Voicebox model data and code.

Introducing Voicebox: The Most Versatile AI for Speech Generation | Meta
https://about.fb.com/news/2023/06/introducing-voicebox-ai-for-speech-generation/

Introducing Voicebox: The first generative AI model for speech to generalize across tasks with state-of-the-art performance
https://ai.facebook.com/blog/voicebox-generative-ai-model-speech/

Voicebox ``Reads input sentences in a natural voice'' ``Records other people's voices and reads the input sentences'' ``Records other people's voices and reads the input sentences with a specified intonation'' It is a voice generation AI that allows operations such as 'recording other people's voices and editing parts'. You can understand the high performance of Voicebox in one shot by playing the movie included in the tweet below.




Many demos of the function 'Record other people's voices and edit some' are published on the following pages. When you play each demo, you can see that high-precision editing is possible so that you can not tell where the edited part is.

Editing
https://voicebox.metademolab.com/edit.html



In addition, it is also possible to read out long sentences just by recording someone else's voice for about 3 seconds. On the demo page below, you can confirm that you can read sentences freely just by recording a short voice like a tweet.

Zero-Shot TTS
https://voicebox.metademolab.com/zs_tts.html



If you check various Voicebox demos, you can see that the risk of being abused is high because the operation of ``reading sentences with other people's voices'' can be performed with extremely high accuracy. Meta also acknowledges the risk of Voicebox being abused and refrains from publicly disclosing Voicebox's learning model and code. However, Meta states that it has also developed an `` effective system for distinguishing between genuine speech and voice generated by Voicebox '', and in the future, a mechanism to distinguish between genuine speech and AI-generated speech will be built. There is also a possibility that

Related Posts:

in Software,   Video, Posted by log1o_hf