In the ever-evolving world of artificial intelligence, OpenAI has been quietly working on a groundbreaking tool called Voice Engine. This AI-powered technology has the remarkable ability to clone a person’s voice after listening to just a 15-second audio sample. While this capability holds immense potential, it also raises serious concerns about privacy, consent, and potential misuse.
OpenAI recently conducted a small-scale test of Voice Engine, and the results were nothing short of astonishing. The model can take text input and a brief audio clip, and then generate speech that sounds strikingly similar to the original speaker. Even with such a short sample, Voice Engine can create voices that sound emotionally nuanced and remarkably realistic.
Developed in late 2022, Voice Engine is already being utilized to power preset voices in text-to-speech applications like ChatGPT Voice and Read Aloud. However, OpenAI recognizes the risks associated with the widespread release of this technology and is proceeding with caution. They aim to foster a dialogue about the responsible use of synthetic voices and how society can navigate these new capabilities. The feedback gathered from these conversations and the results of their tests will inform OpenAI’s decision on if and how they should make Voice Engine more widely available.
“We recognise that generating speech that resembles people’s voices has serious risks, which are especially top of mind in an election year,” OpenAI wrote in a recent blog post, acknowledging the possibility of abuse. To make sure we are taking their input into account as we develop, we are interacting with domestic and foreign partners in the fields of government, media, entertainment, education, civil society, and other areas.”
Despite the concerns, Voice Engine also holds promise for numerous beneficial applications. It can assist individuals with reading difficulties, including children, by providing natural-sounding voices to read text aloud. Companies like Age of Learning are already leveraging this technology to create educational content. Additionally, Voice Engine can translate videos and podcasts into different languages while preserving the original speaker’s voice, making it easier to reach global audiences. Visual storytelling platforms like HeyGen are already utilizing this feature.
Voice Engine is being piloted in the clinical setting to assist patients in regaining their speech. According to a case study on OpenAI’s blog, medical professionals utilised the technology to give a patient who had lost their voice due to a brain tumour back.
While acknowledging the potential benefits, OpenAI is keenly aware of the risks of misuse, especially during election seasons. They are collaborating closely with various partners and implementing safety measures to prevent misuse. Additionally, they are suggesting precautions such as phasing out voice-based authentication for sensitive information and educating the public about the capabilities and limitations of AI.
In order to address the opportunities and difficulties presented by synthetic speech technology, OpenAI is willing to have discussions with academics, developers, and policymakers. OpenAI feels that it is imperative that people grasp the potential of this technology.
