The Voice Clones of Hollywood

James Earl Jones voiced Darth Vader for 45 years. Now, he has given the reigns over to an AI to continue his vocal legacy.

Jones, who is 91, had mentioned he was looking into winding down this particular character. The answer was Respeecher, a voice cloning product that uses deep learning to model and replicate human voices in a way that is nearly indistinguishable from the real thing. Working from archival recordings of Jones, Respeecher created a voice model that could be “performed” vocally by another actor using the company’s speech-to-speech technology.

Previously, LucasFilm had used Respeecher to clone Mark Hamill’s voice for The Mandalorian. – Ars Technica

The whole concept of living forever through AI-assisted legacies is actually becoming a reality. It makes me wonder what Walt Disney would think about all these "memorable" technologies, considering he was (supposedly) cryogenically frozen in order to be revived in the future.

In a similar instance this year, Val Kilmer had his voice cloned for a scene in Top Gun: Maverick. Kilmer lost his voice due to cancer in his throat. However, a company called Sonatic was able to build an AI model of his voice through his previous recordings/movie scenes. Thus, giving him the ability to speak again in this movie.

Ads, Animations, and Assistants

Hollywood is clearly gravitating toward the power of AI-generated voices. It makes sense since celebrity drives viewership and there’s a lifespan (and career span) to every celebrity. But through voice clones, a celebrity could voice characters forever and at a rate more than humanly possible.

Voiceover roles are likely to be the first area to feel the impact of voice clones. That goes for advertisements and animated films/shows. This technology will speed up production while also giving more creators access to top-tier talent.

In the next decade, I think we’ll see an entire animated movie that is voiced entirely by AI.

The value of having a distinct voice will become increasingly lucrative, as your sound could be licensed out endlessly through this technology.

Naturally, there will be a demand to clone the iconic voices of our time, like Morgan Freeman. I remember Mark Zuckerberg showing the world in 2016 that his AI assistant, Jarvis, was voiced by Freeman. But I think there’s a future where even your uncle with the funny voice might one day have the chance to clone his voice and license it to a voice service to use in ads, animations, assistants, etc.

Overcoming the Stigma

There’s still an aversion to the idea of cloning a person’s voice. Replicating anything that makes us distinctly unique individuals goes against the reality of life and death. I remember the public outrage when Prince was turned into a hologram at Super Bowl 52, against his own wishes.

Not long ago, it was widely reported that Bruce Willis sold his likeness to an AI deepfake company. Within a day, these claims were refuted by his rep. Whether Willis’s team was reacting to the negative press and pulling out of the licensing deal or never did it to begin with, this shows that there are a lot of negative connotations to overcome before this is something people are proud to do.

The companies working in this field will need to warm the public up to this idea for some time.

Regardless, the applications of AI voice generation are vast. I cloned my voice so that my production team could edit my podcasts and generate synthetic videos of me without needing me to do another recording. In that same Everydays note I mention how voice clones will immortalize loved ones.

Because a lot of this technology will be open-sourced, AI-generated voices will bring about a lot of societal problems. There have already been numerous online robberies that used voice clones. Inevitably, voice clones will make it simpler to spread misinformation. But that's the story of any technology. It can be used for good just as much as for bad.