use case
verified July 2026
One museum, forty exhibits, twelve languages.
An audio guide is a localisation job wearing a small badge. A single venue is forty or fifty stops, each stop is a paragraph, and each paragraph has to exist in every language a visitor might carry. Multiply the stops by the languages and the word count is a book, priced by a meter that charges for exactly the thing the tour is made of.
01
The arithmetic of a rotating tour
A tour is a stop and a stop is a request, so a whole venue is a loop over requests and twelve languages is that loop run twelve times. On an unmetered seat the loop costs the same whether the museum has ten stops or fifty, and whether it ships in two languages or all twenty-three.
The saving lands hardest on rotation. Exhibits change, wall text gets rewritten, a new wing opens, and each of those edits re-renders one stop across every language at no line-item cost, so keeping the tour current is a normal task rather than a budget conversation.
02
What a multi-language tour needs
- 23 languages from one seat, so a twelve-language guide is one script rendered twelve times rather than twelve separate vendors and invoices.
- Instant voice cloning holds a single guide voice across every stop and every language, so the venue sounds like one narrator rather than a patchwork.
- A pronunciation dictionary fixes an artist's name or a place name once per request, before a visitor ever hears it read wrong.
- Break tags place the pause where a visitor walks to the next exhibit, so the pacing follows the room instead of the sentence.
03
Notes — an engineer's checklist
01An exhibit changed. Do I re-record the whole tour?
No. Each stop is its own request, so a rewritten label re-renders that one stop across every language, and the rest of the tour is untouched.
02How do the twelve languages stay in the same voice?
The voice is a request parameter conditioned on one reference, so every language and every stop is rendered from the same cloned guide and there is no drift between them.
03Does adding languages change the price?
No. The seat is a flat 150 dollars a month and is not metered per character or per minute, so a tour in twenty-three languages costs the same as a tour in one.
See also — related sheets
use case
1
Ship the voice in every market, not the top three
Localised product audio on a flat seat: one cloned identity across 23 languages, with the character count no longer deciding which markets get voice.
use case
23
The same voice, in a language it never recorded
One ten-second reference carries one identity across 23 languages, so a dubbed track keeps the original speaker instead of replacing them with a stranger.
capability
0
Voice cloning from ten seconds, in the request
Zero-shot voice cloning: a ten-second reference rides inside each request, no training job, and the identity holds across 23 languages.
capability
3
A text-to-speech API built for live audio
One REST, SSE and WebSocket endpoint for every text-to-speech workload: narration, dubbing, audiobooks, e-learning, localisation. Flat per-seat, usage unmetered.
A key and one seat to build this on — the same production API this page measures.
Request access for this use case