What Dubbing Really Costs and Takes

Numbers in this article come from two places: our own pipeline, measured repeatedly on real clips (dates included, because this tech moves), and the publicly visible structure of the studio/freelancer market. Where we don't have a real number, we say so instead of inventing one.

Route 1: the studio

Professional dubbing is priced per finished minute, and "per finished minute" hides a lot: casting voice actors, a director, an engineer, a studio room, mix sessions, and usually a lip-sync pass for video. Rates vary wildly by language pair and country, but the shape is universal — the minimum engagement costs more than a year of any self-serve tool. Studios are also slow in calendar terms: days to weeks per deliverable, because humans have schedules.

When is it worth it? Hero content where voice performance is the product — flagship ads, theatrical shorts, anything broadcast — and when you need a named talent's voice. For a content library, course catalog or channel, it's hard to justify.

Route 2: the freelancer or vendor

The middle route: a translator plus a voice artist, or a small dubbing vendor, quoted per video or per minute. Cheaper than a studio, better than raw machine output, and the quality ceiling depends entirely on who you get. The hidden cost is iteration: every correction is another round, and rounds are billed. This route shines for a handful of videos a month; it collapses at twenty.

Route 3: self-serve AI

The economics flipped when per-line voice generation became good enough: the marginal cost of a dubbed minute fell to server time, and the marginal cost of a redone minute fell to zero. That second part matters more than the first. In our own pricing, Premium is $12.99 a month for videos up to 30 minutes with a 10-minute-speech daily allowance; there's a free tier for short clips. Per-video math stops being the point — you're buying the ability to redo anything, anytime.

The timings, measured

Here's what our own pipeline actually does, measured end to end on an eight-second, two-speaker clip (Russian to English, run October 2026, warm server):

Scaling follows speech volume, not runtime. A 40-minute lecture with continuous talking is roughly forty times the voice generation of our eight-second clip; a 40-minute webinar with long pauses is half that. The studio-time model ("minutes per finished minute") maps onto AI as "roughly linear in spoken seconds, with a fixed overhead per video."

The full cost picture, honestly

One more cost people forget: the cost of not dubbing. A video that only works in one language is a ceiling on its audience, and every month it stays untranslated is a month of that ceiling. When the marginal dubbed language costs a subscription and an afternoon, the expensive decision is increasingly the decision to skip it.

When DIY stops making sense

Honesty requires the other side of the ledger. Self-serve dubbing hands you the quality controls, which means you also hold the quality responsibility. The crossover points are predictable:

• Volume beyond your review capacity. If you ship more dubbed minutes per week than you're willing to listen to, either cut volume or hire the review out — at which point you're halfway to the vendor route anyway.
• Performance-critical voice acting. A script that lives or dies on comic timing wants a human performer. AI dubbing is good; it's not directed.
• Languages you can't quality-check at all. If nobody on your team or circle can spot-check a language, the native-review line in the budget stops being optional — for any route.

For everything else — the long tail of tutorials, talks, interviews, product videos — the arithmetic is lopsided in a way it wasn't two years ago. Try it on the free plan with your shortest video and check the clock yourself; the fastest way to budget this is to own one real measurement from your own content.