AI and machine learning technologies are leading the change in almost every area of life. From business and medicine to entertainment and education, AI disrupts how we use tech. One of the most notable examples of this shift in technology is the multimedia industry, where AI continues to deliver results that were simply unimaginable. One of the newest innovations is that of generating AI voices.

Three of the most impactful instances of technological innovation for multimedia in 2021 took place when AI helped clone the voices of cultural icons for social and entertainment initiatives.

AI voice cloning technology at a glance

In short, speech-to-speech voice conversion allows audio engineers to clone the voice of any person so that the cloned voice is indistinguishable from the original.

In practice, this occurs via the AI system, which analyzes two voices — the target and the source and then transforms the source’s voice into the target’s voice. For example, the voice of a random person can be transformed into the voice of the intended celebrity.

The resulting recording will contain all the emotional intonations the source person spoke with and will come out the other end as the target’s voice.

Giving Voice to Michael York in Healthcare Initiative Fighting Bone Marrow Cancer

Michael York, a famous Hollywood actor, was suffering from bone marrow cancer ten years ago. When a non-profit organization decided to create an educational program about the disease, Michael York kindly offered his unique voice for the narration.

The only problem was that the actor had utterly lost his voice when the project went into production. When changes needed to be made to the film that had already been completed, Michael York could not dub the sections the production team added to the original animation.

Thanks to AI voice cloning technology, Respeecher was able to produce a solution to the problem.

Based on earlier recordings of the actor’s voice, Respeecher generated a model of Michael’s voice and then synthesized the actor’s speech so precisely that the audience was none the wiser.

The film was a success at conferences, meetings, and clinics worldwide.

With tremendous gratitude to the voice cloning team, Michael York said: “Bravo to the entire team for their extraordinary achievement! The quality of the re-worked passages is indistinguishable from the original.”

To listen to snippets of Michael’s synthesized voice, read the original story in this case study.

Synthesizing Young Luke Skywalker’s Voice for The Mandalorian Series

The Mandalorian is a Star Wars television series created by Jon Favreau for the Disney+ streaming service.

The Mandalorian was the first film production to use real-time rendering for a realistic parallax environment. It is also one of the first TV shows to utilize Respeecher’s voice cloning technology in a Hollywood studio production.

The challenge was to make 68-year-old Mark Hamill sound like a 20-year old version of himself. The producers decided to use deepfake technologies for the visuals and voice to create the younger version of Luke Skywalker.

Jon Favreau’s team got together with Lola VFX for video production and Respeecher to produce the young Skywalker’s speech synthesis.

After analyzing the younger voice of Luke Skywalker, Respeecher recreated a replica that was indistinguishable from earlier authentic versions.

“Something people didn’t realize is that his voice isn’t real. His voice, the young Luke Skywalker’s voice, is completely synthesized using an application called Respeecher,” says Jon Favreau.

To learn more about the project, read Respeecher’s initial story.

Resurrecting a Legendary NFL Coach Vince Lombardi

The NFL is often the stage for some of the most memorable commercials before and during the season’s final game of the year, the Super Bowl. In 2021, the NFL brought back the American football legend Vincent Lombardi — the iconic coach and NFL executive spoke on the stadium’s screen minutes before the song “America the Beautiful.”

To create a genuine vocal model of Lombardi, Respeecher had to rely on the original audio material of the coach who had passed away in 1970. Needless to say, speech recordings made in the pre-digital era were not a reliable source of data for voice synthesizing.

Researcher did an extraordinary job processing and restoring Lombardi’s original speech recordings for this project. Where AI typically requires only an hour-long speech record, dozens of hours of recordings have been used to build Lombardi’s voice model.

You can watch the original Super Bowl commercial here and read more about the story on the Respeecher blog.

This article was initially published on News Nation USA.