AI Startup Aims to Give Voice Actors a Digital Twin

Morpheme strives for a talent friendly approach to applying AI to films and video games

(Bloomberg) — Over the years, Cissy Jones, a voice actor with many credits, including video games Starfield and Baldur’s Gate III, has performed countless fictional mishaps. Death by gunshot, she said, is easy. Whereas death by fire can be long and tedious. One time, she played an astronaut in space who woke up breathing tensely. Ten minutes into the scene, Jones says she came close to passing out.

“Imagine your paycheck being dependent upon being able to sustain that,” Jones said

If Jones and her business partners succeed, sometime soon she may no longer have to. Such work, she said, will simply be outsourced to her “digital twin,” a vocal doppelganger powered by artificial intelligence.  

Jones is the co-founder and vice president of strategic partnerships at Morpheme, a startup aiming to harness artificial intelligence to reshape how vocal performances are used in everything from animated series to video games. It’s the kind of venture that lately has tech investors excited — and all manner of creative professionals, from screenwriters to actors, on edge. Within the video-game industry, workers are concerned that AI could eliminate entire job categories, such as quality-assurance testers. 

Morpheme’s AI software records audio from actors and then creates a model of their voice that can be used to alter, expand and enliven future productions. Recently, Morpheme has been demonstrating the technology to entertainment businesses, including several top gaming companies that it declined to identify, citing nondisclosure agreements. 

Among other features, Morpheme is developing a library of exertion sounds, such as heavy breathing or shrieking, that clients will be able to reuse throughout the development process. That way, actors like Jones won’t have to strain themselves as much, and producers can get all the bloody screams they need. The software will also be able to render a performer’s voice into additional languages, making it cheaper and faster for companies to translate products across cultures. 

Morpheme is entering an increasingly crowded market. In 2022, investors poured $378.6 million into voice-related AI startups across 47 deals, according to data from PitchBook. Some are using AI to clone voices, alter voices and generate audio from text.

Recently, labor unions have raised urgent concerns about how such AI-technology will be incorporated by entertainment giants. It remains a point of contention in the ongoing strike by Hollywood actors and is flaring up on another front, as well. 

In September, video-game voice actors with the SAG-AFTRA union authorized a strike amid ongoing negotiations with top gaming companies including Electronic Arts Inc., Take-Two Interactive Software Inc. and Activision Blizzard Inc. About 2,600 performers work under the Interactive Media Agreement. AI has been a sticking point.  Actors fear entertainment companies will use AI to reproduce their voices without permission or payment, pushing down the value of their work. 

“The unregulated use of artificial intelligence poses an existential threat to anyone who makes their living using their voice, image or performance,” union leader Ray Rodriguez said.

Jones, for one, believes that she and her colleagues may be better off embracing the technology, albeit under the right conditions. “It’s not going backwards,” she said. “It’s going to get better, cheaper, faster.” 

Morpheme, which is privately funded and seeking additional investors, is banking on the idea that amid the mounting anxiety and potential work disruption, entertainment companies will embrace some rules of engagement. Currently, Morpheme’s 20-plus employees are in regular conversation with voice actors, Hollywood agents, entertainment companies and SAG-AFTRA, hoping to stake out some middle ground between the competing interests.

A spokesperson for SAG-AFTRA said in a statement that the union has had “ongoing conversations,” with Morpheme, noting that “any use of digital replicas” must include a range of provisions, including “safe storage” and “appropriate compensation.” A spokesperson for the gaming companies negotiating the interactive media agreement did not provide comment.

A key part of Morpheme’s sales pitch is that it has created a pay model that is fair to actors.  Under a broad and flexible contract, Morpheme will let performers approve of how their AI-engineered voice will be deployed and will compensate them for its continued use through a licensing fee. After a year or so, they will have the option to renew or renegotiate. To keep score, Morpheme is also developing technology that will monitor and track how the companies are using the AI-generated or modified audio. 

Recently, Morpheme signed on to a new campaign from the National Association of Voice Actors (NAVA) that has  a mantra of “consent, compensation and control” around the use of AI.

“If we as humans mess this up, we’re gonna destroy ourselves,” said Morpheme Co-founder and Chief Executive Officer Brett Shapiro, who has worked in tech and entertainment for decades, including a stint at the gaming company Cosmic Forces. 

Unlike their kin in film or television, voice actors for video games don’t receive residual payments after their recording sessions. Now, some gaming actors are looking at the emerging AI technology as an opportunity to potentially collect extra payments down the road on top of a base minimum. Under Morpheme’s contract, actors who are unavailable or unable to work on a new project can put their “digital twin” to work and, in exchange, receive additional money.  

“There’s a licensing opportunity for me as a voice actor to get paid for the generation of that audio,” said Tim Friedlander, co-founder and president of NAVA. “The more my character becomes popular and people interact with it, the more I get compensated.” 

Meanwhile, unauthorized uses of AI technology are already proliferating. 

“We’re seeing people rip off publicly available content to create whatever they can in whatever fashion they want,” Jones said. 

Jones has noticed fans on TikTok remixing her past scenes with words she “never said before, very clearly robotic.” Some creators have honored her takedown requests, and others haven’t. 

“There’s no consideration for the human behind the original art,” she said. 

More stories like this are available on

©2023 Bloomberg L.P.