Subtitles translate the dialogue into another language as text. Captions render the same-language audio as text, including sound cues, for viewers who can't hear it. Dubbing replaces the spoken audio with a recorded performance in the new language, and voice-over lays one narrating voice over the original. The distinction matters in practice, so naming which deliverables a release needs is the first thing we settle on any project.

For most US television programming, and for that programming when it moves online. Federal Communications Commission (FCC) rules require captions on new English- and Spanish-language TV programming and set quality standards for accuracy, timing, completeness, and placement, and the same captions must follow the programming to internet delivery at the same quality. In National Association of the Deaf v. Netflix (2012), a court applied the Americans with Disabilities Act (ADA) to a streaming service, and the consent decree required captioning. If your content never airs on US television, the rules may not reach it, but the audience expectation they set is now the baseline.

The method sets the price. Subtitling is quoted from the runtime and the languages; voice-over adds talent and recording time; full dubbing adds casting and direction for every speaking part, so it costs the most. Repeated wording costs less on ongoing work, because translation memory, the record of everything already translated for you, means a recurring title or a re-used segment is not billed as new work. Send the cut and the release date, and the quote comes back with a recommendation per piece.

For an internal review cut, yes, and we'll say so. For anything your audience sees, the record argues against it: auto-generated captions still misread names, accents, and technical terms, and a machine-translated subtitle publishes those mistakes to every viewer at once. There's also an accountability gap: machine output is anonymous. Every Argo Translation delivery arrives with a signed record naming the credentials behind the work, which is something you can check on your first project.

From a working roster covering 80+ languages, cast against the content rather than assigned by language alone. A safety course, a character performance, and a promo each call for a different read, and the same project can need several voices, languages, and personality types at once. Recording runs remote or in studio, and your team hears the voices in context before the version ships.

No. An AI voice can be the right call for internal training, drafts, and high-volume content where the budget decides what gets translated at all, and we'll tell you when it is. The line we hold is audience-facing work: when the voice is the performance, a professional records it. If you're weighing the two, send one piece and we'll quote it both ways.