■ SAFE RISK ■ Sports & Entertainment
No. Synthesis makes fake voices trivial, which is exactly why watching a person produce one with their lips shut, live, in the same room, still sells tickets.
“AI can synthesize any voice. But throwing your voice with a wooden dummy? Peak human absurdity.”
Our AI replacement risk score — how we score jobs
The craft is technical before it is funny. A ventriloquist trains substitution sounds for the labials — B, F, M, P, V — that the lips would otherwise betray, holds a stable voice placement while breathing normally, and coordinates puppet manipulation so the figure's mouth, head tilt and eye focus land on the beat. Then there is the comedy: writing two-hander material, building a distinct character with a consistent worldview, and running crowd work where the puppet insults an audience member and the ventriloquist reacts as though genuinely surprised. Professionally the job also means touring logistics, corporate and cruise-ship bookings, figure maintenance, and a social media presence that increasingly drives bookings more than clubs do.
Automation is visible in adjacent territory. Voice cloning and animatronics can produce a talking character without a performer, animated and AI-voiced puppet content competes for attention on short-form video, and virtual characters host streams. Writing tools can generate joke drafts. Corporate entertainment budgets can now buy a synthetic mascot instead of a live act.
The reason the score stays low is that the act's premise is human limitation. The audience is watching someone do something difficult with their own body, in real time, with a piece of carved wood — and reacting to this specific room. Remove the person and there is no trick, just a cartoon. The strongest performers already lean into this by improvising with the crowd, the one thing a pre-rendered character cannot do. The economics reflect that split. Recorded character work and corporate voiceover money has thinned considerably, while theatre and cruise contracts for performers who can hold a room for ninety minutes have held up, because those buyers are purchasing a live event rather than a piece of content.
Automatability: our editorial assessment of current and near-term AI capability
Resilient for the foreseeable future. The pressure that exists is on recorded and corporate work, where synthetic characters and animation are already cheaper, and that squeeze tightens through the 2030s. Live performance is the opposite case: the value proposition is watching a person pull it off in front of you, which grows more novel as synthetic media becomes ordinary. Expect fewer voice-for-hire gigs, steady stage demand.
It threatens recorded character voice work, not the stage act. Synthesis makes producing any voice trivial, which devalues the novelty of a strange voice and simultaneously increases the value of doing it live with visible effort. Audiences pay to see a difficult physical skill performed in front of them and to be part of the improvised bits. A cloned voice offers neither.
It is a niche one that supports a small number of people well and many people partially. Income tends to come from corporate events, cruise ships, family and school shows, theatre touring, and increasingly from video platform reach. The performers doing best treat social video as the top of the funnel. The economics are difficult in the way all live comedy is — not because of technology.
The business scaffolding and the content pipeline: booking admin, contracts, invoicing, video editing, thumbnail and promo generation, subtitle and clip production. Writing assistants can produce serviceable joke drafts to punch up. All of that saves hours. The performance itself — lip control, timing, manipulation and reading a room — is untouched.
As a production department. Use it to cut promo videos, generate captions and clips, draft premises you then rewrite in the puppet's voice, and handle scheduling and outreach. Some performers experiment with synthetic voices as a bit within the act, which lands well precisely because the audience knows the difference. The trap is outsourcing character voice, which is your signature.