Helping actor Val Kilmer reclaim his voice

Request access

Try our hyper-realistic voices

John Flynn

CTO & Co-founder

August 9, 2021

Answering the call for a high-profile project

Last year, we were contacted with an exciting question: Could Sonantic build a custom AI voice model for acclaimed actor Val Kilmer? 

Kilmer, whose Hollywood career has spanned nearly four decades, has starred in scores of films, including blockbusters such as Top Gun, Willow, The Doors, Tombstone, Batman Forever, and Heat. But after undergoing a tracheotomy in 2014 as part of his treatment for throat cancer, Kilmer no longer has a voice that would be easily recognisable to fans. 

The actor has been reflecting on his career recently for the production of his autobiographical documentary, Val. But despite this recent focus on the past, Kilmer never stops moving forward. He and his team knew that building a custom voice model would help him explore new ways to communicate, connect, and create in the future.

From the beginning, our aim was to make a voice model that Val would be proud of. We were eager to give him his voice back, providing a new tool for whatever creative projects are ahead.

Collecting and processing recordings of Val’s voice

When we create voice models with actors who are able to perform their lines, we provide set scripts. The actors record performances of those scripts, then we feed the recorded audio into our proprietary Voice Engine, which trains the AI model. The Voice Engine provides a variety of manual and automated capabilities for checking the quality of the model. 

When an actor is unable to perform lines from our scripts—as in the case with Val—we need to do a bit more manual work. The first step is to gather audio recordings of the actor. For the best results, we need clean audio, with little background noise, in the highest-quality format. 

Our first hurdle with creating the voice model for Val was that his team provided a relatively small amount of recorded material, which we knew would make it more difficult to produce the model. We first cleaned the audio, carefully removing background noise without destroying the speech content. Next, we generated transcripts from the audio and paired the audio and text together in short chunks. The Voice Engine then began training the model with the resulting data—about 10 times less data than we might have used in a typical project.

With that limited amount of data, the algorithms we had been employing in our Voice Engine did not deliver the results we needed. We decided to research and implement new algorithms that could produce higher-quality options from the data we had available. In the end, we generated more than 40 different voice models and selected the best, highest-quality, most expressive one. Those new algorithms are now embedded into our Voice Engine, so future clients can automatically take advantage of them as well.

Fine-tuning performances

Once a custom voice model is produced, our desktop application enables creative teams to input text and then fine-tune the performance, adjusting a range of key parameters. In Director mode, for example, creative teams can modify the pitch and pacing of phrases to dial in the right delivery. 

As Val’s team experiments with the model, Director mode helps to produce authentic, highly realistic results. No matter how Val uses the voice model, he can now communicate using his own voice. 

Opening new creative possibilities

Producing a custom voice for Val Kilmer was more than a technical accomplishment—it resonated on a deep level with the Sonantic team. Sonantic Co-founder and CEO Zeena Qureshi explains her personal connection with this type of work: “I spent nine years helping children with autism learn how to use their voice as a better instrument for communication. The project with Val demonstrated again how empowering it can be when people overcome challenges with speaking.”

At the same time, this project aligned strongly with Sonantic’s passion for fostering creativity. “Val is an inspirational person with a deep soul. And to this day, he’s brimming with creative talent, in spite of everything he’s faced,” says Zeena. “We collaborated on this voice model because we believe it will help him connect with people and enable him to continue to tell stories that move us. I’m honoured to be part of this project with Val, and I know the team shares this feeling.”

Val himself has been extremely appreciative of the work. “I’m grateful to the entire team at Sonantic who masterfully restored my voice in a way I’ve never imagined possible. As human beings, the ability to communicate is the core of our existence and the effects from throat cancer have made it difficult for others to understand me. The chance to tell my story, in a voice that feels authentic and familiar, is an incredibly special gift.”

We’re looking forward to hearing how Val will use this voice model to explore new creative possibilities that entertain and inspire us in the future. 

→ Return to blog

Heading

Lorem ipsum dolor sit amet, consectetur adipiscing elit. Suspendisse varius enim in eros elementum tristique. Duis cursus, mi quis viverra ornare, eros dolor interdum nulla, ut commodo diam libero vitae erat. Aenean faucibus nibh et justo cursus id rutrum lorem imperdiet. Nunc ut sem vitae risus tristique posuere.

  • It prevents you to directly style elements of a RT when this RT is bound to the CMS. I hope this limitation will be removed soon, but in the meantime the workaround is pretty easy to handle. Proceed as follow:
  • CMS. I hope this limitation will be removed soon, but in the meantime the workaroun
  • Vents you to directly style elements of a RT when this RT is bound to the CMS. I hope this limitation will be removed soon, but in the meantime the wor