OpenAI has developed a very powerful artificial intelligence model called Voice Engine for voice cloning. developed. On the other hand, this model is performing so well that the company fears a huge potential for abuse of the model. This made the company hesitant to make the model public, and instead the company only offered a preview showing the capabilities of a model called Voice Engine.
OpenAI Scares with Voice Engine!
The basics of artificial intelligence-based voice cloning technology are very simple. The model only needs a voice sample of the two original voices and a text for the synthetic voice to read. When the tool is fed with a sufficient number of samples, the result is truly astounding. Unlike other models currently available to the public, the Voice Engine only requires 15 seconds of audio from the original speaker. Despite very limited input, the resulting vocal expressions are incredibly realistic. The powerful tool can become a powerful weapon in the hands of malicious actors, especially as part of disinformation campaigns.
This is a cloned voice. OpenAI’s #VoiceEngine can create these convincingly realistic speech expressions using as little as 15 seconds of audio sample: pic.twitter.com/9LIGhBA3sL
— Martin Hodás (@Hody_MH11) March 31, 2024
Voice Engine was first developed in late 2022. It has since been used to support ChatGPT Voice and Read Aloud, in addition to the preset voices available in the text-to-speech API. Late last year, OpenAI began privately testing its voice cloning capabilities with a small group of trusted partners. The company says it is impressed by the applications this group has developed.
One reason for these tests is to find out how people and various industries can benefit from it. The other reason is to identify the potential for abuse and decide what steps to take. OpenAI thinks the wider rollout of this technology should go along with policies and countermeasures to prevent its misuse.
It is worth noting that the OpenAI model will not be the only publicly available voice cloning tool. The most popular one right now is ElevenLabs, but even with enough voice samples, the results are not always convincing. Voice Engine will be a big step forward, both in terms of ease of use and the resulting quality of the cloned voice.
