Voice Over

Preview image of “Voice Over”
  • Speakers’ voices are automated by AI
  • Photos: Shutterstock
Technology means that digital voice assistants sound more human

Pushed by a wave of startups, robot actors are using Artificial Intelligence to take over voiceover work.

Acting has always been a tough industry, but securing gigs as a voice actor could get harder, as AI services get better at replicating human speech. Electronic voices in the past always tended to sound rather alien, with strange emphasis, pronunciation, and rhythm. But some companies, such as WellSaid Labs in Seattle, are tackling this by modelling voices based on real human actors. They take the real voice, with the actor’s consent, and create individual AI actors “for hire” with various personality types. Deep learning techniques are breeding voices that sound more and more human, including aspects like pauses, breathing, and better intonation.

Video game developers, corporations, and filmmakers can save serious money with computerised voices. Instead of hiring actors and studios, they can simply enter text into a field on-screen and adjust whenever
script changes or different emphasis are required. Brands under pressure to produce voice content, for example for digital assistants or tutorial videos, can save hugely by producing their own AI voice actor, who
speaks in any language and delivers any content instantly. And it makes their brand communication more consistent.

Sonantic is one provider of AI voice services, whose founders Zeena Qureshi and John Flynn have backgrounds ranging from speech and language therapy to Hollywood sound production. They claim to have built a “Photoshop for voice”, and cite more than 1000 companies on their waiting list. They believe “using technology to augment actors’ voices will be the new normal within five years time.” And they emphasise a winwin for both studios and actors, with the latter standing to gain “passive income…”

Some providers are exploring profit-sharing models to pay actors every time a client licenses their specific synthetic voice – meaning a potential new way for voice actors to earn. And this is something many of these
companies are stressing – fairness. Some have approached unions to work out how to justly buy rights and compensate human voice actors for their work. Perhaps they are worried about future lawsuits, such as the 2018 suit from Galit Gura-Eini, the voice of Hebrew Siri, who sued Apple because people were making Siri say bad words.