3 things that do not multiply a ten-language dub

The leftover /cost/dubbing-a-video-into-10-languages meter is job × languages. Which language you picked, how many speakers are in the source, and the lipsync per-second rate card do not multiply. A resubmission is a new job. Japanese and Spanish are the same charge.

That cost page is the live formula. This post is the leftover: three variables people still multiply anyway, and the one event that does mint a new charge. Dub a video is the editor-shaped door. Dub a video into another language is the named agent job. The dubbing tool is the same pipeline. Confirm the per-job figure in-app before you fan out. Then multiply only by how many targets you actually submit.

The leftover meter is job × languages

The formula on the cost page is credits = dub_job_price × languages. dub_job_price is one flat charge per submitted dub. languages is one job per target. That second term is the only multiplier. Ten markets is ten identical charges. Five markets is half of ten, exactly. There is no batch discount and no volume break.

The per-job figure is quoted at submission rather than published as a rate. Check the estimate on one dub, then multiply by your market count. That is the whole model. Do not reach for a per-second card to "sanity check" it. Do not assume Japanese is dearer. Do not assume a panel discussion is dearer because more people talk.

A resubmission is not a discount on a language you already ran. Re-running a target you already dubbed is a new job at the same price. Approve the master before you fan it out. A typo caught after ten dubs costs ten dubs. Translating on-screen text is a separate captioning job per language. That is another meter, not a reason to multiply the dub line.

Which language you picked

Nothing in the pricing reads the target language. There is no rare-language surcharge, no multiplier for a smaller market, no discount for a common one. Japanese and Spanish charge the same. A language your audience is small in costs exactly what your biggest market costs.

Difficulty is not a pricing dimension. The target is a label on the job. It is not a rate.

Capability can still rule a target out. The default ElevenLabs engine is the more permissive on input and on the language list. HeyGen supports a subset of languages, is video-only, caps at eight minutes, and does not trim. That is an engine constraint, not a price multiplier. Pick the engine for what it does to the video, then confirm the charge. Two dubbing engines and their hard limits is the capability map. It does not change languages as the only term that multiplies.

The dub target list is narrower than the voice list. Voice coverage is not a dub. A missing target is a zero, not a surcharge. Rank markets by expected return, not by language difficulty.

How many speakers are in the source

num_speakers is a hint on the job so a panel or interview keeps voices distinct. It is not a multiplier. One host and a four-person panel are the same dub job on the language axis. The cost table on the live page is explicit: number of speakers in the video does not multiply the charge.

This meter is not a studio. It is one submitted dub per target language. The engine does not bill a second stem because a second person talked. Heavy noise is a quality problem, not a rate. Fix the master before you fan out. Trimming on the default engine can happen in the same submission. HeyGen does not trim. Neither engine turns speaker count into a second languages term.

Ten languages is ten jobs whether the source is a monologue or a panel. The source has to be media already hosted on Versely. External links, YouTube included, are not accepted. That constraint does not change the meter. It only decides whether the job can start.

The lipsync per-second rate card

This is the expensive mistake the cost page exists to kill. Lipsync bills per second of finished clip. Across the lipsync catalogue that rate spans a documented spread — lipsync credits ignore the script, the voice and the language walks the rate × seconds shape, including a 1-to-30 credits-per-second range on that card at the time it was written. A three-minute video across ten languages, under that arithmetic, looks catastrophic. Dubbing does not work that way.

A dub is its own job type with its own flat charge. It is not derived from the lipsync rate card. They share neither the meter nor the rate. HeyGen's capability is lip-synced translation. That is still a dub job, billed as a dub job, not a per-second lipsync render. Choosing HeyGen because the mouth has to match does not import the talking-head rate card.

If you actually need a lipsync render — a face already on tape, new audio, per-second billing — that is AI lipsync, and it is a different cost page. Two meters. Confirm one language in-app. Multiply by the number of targets you will ship. Ignore the language name, the speaker count, and the lipsync card. Resubmit only when the master is actually wrong.

FAQ

Do rare languages cost more than common ones?

No. Language is not a pricing dimension. Japanese and Spanish are the same charge. The only term that multiplies is how many target languages you submit.

Can I estimate dubbing cost from the lipsync per-second rates?

No, and doing so overstates it badly. Lipsync is a per-second render of a face against an audio track. Dubbing is a flat per-job charge. They share neither the meter nor the rate card.

Does the number of speakers in the source change the bill?

No. num_speakers is a hint so voices stay distinct. It does not multiply the dub charge. A panel is still one job per target language.

Is a resubmission cheaper if I already dubbed that language?

No. A resubmission is a new job at the same price. There is no batch discount and no loyalty rate for a language you already ran. Approve the master first.