Top Gun is back and for some reason very successful. Artificial Intelligence plays a part in that: without AI audio, Val Kilmer probably wouldn’t have returned.
With “Top Gun: Maverick”, the sequel to the 80s cult film “Top Gun – They Fear Neither Death Nor Devil” recently opened in theaters. Director Joseph Kosinski delivered: On the review site Rotten Tomatoes, “Maverick” currently receives a tremendous 97 percent from 309 critics with an audience approval rating of 99 percent.
Returning from the original cast, alongside Tom Cruise, is Val Kilmer. As “Iceman” Tom Kasanzky, Kilmer plays the (former) rival to Tom Cruise’s “Maverick.”
Deepfake for good: AI brings back Val Kilmer’s voice
The fact that Kilmer was able to take on the role at all is thanks to medical and technological advances made in recent years. Kilmer suffered from throat cancer around 2014 and lost his voice after an operation. Since then, Kilmer can only speak with an auxiliary device and a heavily distorted electronic voice.
In 2018, however, one of Kilmer’s assistants had an idea: He heard about advances in synthesized media and specifically AI for audio. He turned to the British company Sonantic, which trained a neural network using Kilmer’s voice from archival recordings. You can listen to the result in the video below.
Because of the comparatively small amount of data, Sonantic says the project was one of the most complex of the past year. Among other things, the recordings had to be cleared of background noise. The company created a total of 40 different voice models, from which it then worked with Kilmer’s team to select the best.
In Top Gun: Maverick, AI speaks for Kilmer
When the collaboration with Kilmer was announced, it was already said that the AI voice would be used in the new Top Gun movie. During production, those responsible can additionally modify the synthetic voice with a software tool, for example to adapt emphasis and height or depth to the respective scene.
“I’m grateful to the entire team at Sonantic who masterfully restored my voice in a way I’ve never imagined possible,” Kilmer said at the unveiling of his AI voice last fall. He can license the voice to studios and also use it in everyday life.
Sonantics chief technology officer John Flynn describes the collaboration in detail on the company’s blog. Flynn sees other uses for AI voices, such as earning money with virtual work: Actors could go on a shoot, for example, while their AI voice dubs a commercial. Bruce Willis did something similar last year, but he deepfaked his face for a Russian commercial.
Among other things, modders use AI-generated voices from voice actors to give their game modifications an authentic soundtrack. Sonantic also offers synthetic voices for games.