A demonstration video featuring an AI-generated imitation of the renowned naturalist David Attenborough narrating a company’s video feed has surfaced, showcasing developer Charlie Holtz’s innovative use of technology. By combining GPT-4 Vision (referred to as GPT-4V) and ElevenLabs voice cloning technology, Holtz created an unauthorized version of Attenborough narrating his actions on camera. The video quickly gained traction, amassing over 21,000 likes within a short period.
In the demo, the fabricated Attenborough character describes Holtz as a remarkable specimen of “Homo sapiens,” noting his distinctive silver spectacles and tousled wavy hair, speculating about the blue cloth he is wearing as part of a display. The narration continues in the style of a BBC documentary, observing Holtz’s facial expressions and gestures with a tone reminiscent of Attenborough’s iconic voice.
The process involves a Python script named “narrator,” capturing images from Holtz’s webcam every five seconds and sending them to GPT-4V for text generation. The generated text is then adapted using ElevenLabs AI tone profiling, trained on Attenborough’s voice samples. Holtz has shared the code for this project on GitHub, highlighting the integration of costly API tokens from OpenAI and ElevenLabs.
This creative application of technology underscores the potential of combining different AI capabilities through API integration. While these technologies have been available independently, developers are now exploring novel ways to merge them, leading to unexpected and entertaining outcomes like Holtz’s AI-generated Attenborough narration.
The video showcases Holtz engaging in mundane activities, such as drinking from a bowl, accompanied by the faux Attenborough’s humorous and insightful commentary. Similarly, another example involving a cloned voice of Steve Jobs critiquing design styles in Figma demonstrates the versatility of these AI tools in mimicking iconic personalities.
However, the ethical implications of voice cloning technology, particularly in creating convincing deepfakes, raise concerns about privacy and intellectual property rights. While ElevenLabs prohibits the unauthorized replication of individuals’ voices, enforcing such restrictions poses challenges in practice.
The demo has elicited mixed reactions, with some viewers amused by the novelty of the AI narration, while others express discomfort over the unauthorized use of Attenborough’s voice. As discussions on the ethical boundaries of such technologies continue, it is essential to consider the implications of AI-generated content on privacy and intellectual property rights.
