OpenAI Has Introduced An AI Tool That Can Mimic the Human Voice OpenAI has released a new tool that can mimic the human voice. Writing a blog post, OpenAI has released a tool called ‘Voice Engine’ for voice clones.
Currently, OpenAI is privately testing this tool with a few groups of small businesses. It is assumed that the company has done this because it can imitate the voice of a real person based on some samples, which raises ethical questions. But this tool is just an extension of the ‘text to speech system’ already made by OpenAI.
Looking at some of the samples in OpenAI’s blog post, it can be heard that this tool can reproduce almost the same sound as zinc. The cloned voice can be heard speaking on different topics and even using the same voice in different languages.
In the sample audios provided now, you can hear the tool creating the same sound for the same amount of time-based on 15 seconds of real sound.
OpenAI also stated that the company has considered this matter and is working on a system of watermarking to prevent such possible abuse. The company believes that it will solve ethical problems.
However, there are areas where this type of tool can be useful for many. Now the streaming platform Spotify has used this kind of tool to make the audio available in different languages. But OpenAI has responded that it has not yet thought about making this tool available for public use.
The company had previously released Sora, a tool for creating realistic videos of up to one minute based on prompts. The tool has not been made publicly available yet.