OpenAI Released A Voice Cloning Model That Needs Only A 15-Sec Clip

OpenAI has introduced Voice Engine, a revolutionary text-to-voice generation platform. This technology can create lifelike synthetic voices based on a mere 15-second audio clip of a person’s voice, offering a plethora of possibilities across various industries. Simply put, OpenAI will now reproduce a voice just from a 15 second clip. Think about how this would potentially change the voice over game especially in the film and general media industries. In a sample that OpenAI provided via a blog post, it gave 15 second clips in English and Spanish and then the regenerated AI versions. It was hard to tell really and that’s the issue with AI – deepfakes. The first one below is the reference, 

Voice Engine enables the creation of AI-generated voices that can seamlessly read out text prompts in multiple languages, mirroring the accent and intonation of the original speaker. OpenAI emphasized the positive impact of Voice Engine, highlighting its potential applications in education, healthcare, communication, and beyond.

The platform has already garnered attention from leading companies, including Age of Learning, HeyGen, Dimagi, Livox, and Lifespan, who are utilizing Voice Engine to enhance their offerings. Age of Learning, for instance, is leveraging the technology to generate pre-scripted voice-over content and deliver personalized responses to students, powered by the advanced capabilities of GPT-4.

Voice Engine represents the culmination of extensive research and development efforts by OpenAI, with the model trained on a diverse dataset comprising licensed and publicly available data. Jeff Harris from OpenAI’s product team underscored the significance of this innovation, noting its integration into ChatGPT’s Read Aloud feature and the text-to-speech API.

While AI text-to-audio generation continues to evolve rapidly, ethical considerations remain paramount. OpenAI has implemented stringent usage policies to ensure responsible deployment of Voice Engine. Partners are required to obtain explicit consent from original speakers, refrain from impersonation, and disclose the AI-generated nature of the voices. Additionally, watermarking and active monitoring mechanisms are employed to track the usage of audio clips.

As the technology landscape evolves, OpenAI advocates for proactive measures to mitigate potential risks associated with AI voice technologies. This includes phasing out voice-based authentication for sensitive transactions, implementing robust policies to safeguard individuals’ voices, raising awareness about AI deepfakes, and developing systems to track AI-generated content.

With Voice Engine poised to redefine the boundaries of synthetic voice generation, OpenAI continues to lead the charge in driving innovation while prioritizing ethical considerations and societal well-being.