Mundo Claro
municipal captioning
Interactive product demo

One broadcast, every viewer.

Mundo Claro adds live captions in two languages and a real-time dubbed voice to a municipal broadcast — on your hardware, while the meeting is still happening. Below are three real programs it processed. Press play, then switch the layers on and off.

▶ Startup Portugal hearing (PT → EN) ▶ The Frame, Framingham (EN → PT) ▶ Matosinhos assembly session (PT → EN)
What the buttons add
Layer 1

Captions — original language

Streaming speech recognition on the program feed, delivered as broadcast captions in the language being spoken.

Layer 2

Captions — translated

The same speech, machine-translated live. Both caption tracks stay in lockstep because they come from the same recognized segment.

Layer 3

Dubbed audio

A neural voice speaks the translation as a second audio program, a few seconds behind the podium. Synthesized audio carries an inaudible provenance watermark.

Sample 1 · Portuguese → English

Assembleia da República — Startup Portugal on the sovereign fund

Source: ARTV — parliamentary committee hearing of the president of Startup Portugal on the Banco de Fomento fund of funds · Clip: 83 s · Pipeline output captured live — an MP yields the floor, and the president's reply airs in English seconds behind the Portuguese.
EN DUB
Layers
Playing the original broadcast — no accessibility layers.
Sample 2 · English → Portuguese (Brazil)

The Frame — Framingham News in Focus, Access Framingham TV

Source: weekly news program — the council debates staffing for the new Mary Dennison Park · Clip: 66 s · Same pipeline, reversed — for Framingham's large Portuguese-speaking community.
PT-BR DUB
Layers
Playing the original broadcast — no accessibility layers.
Sample 3 · Portuguese → English

Assembleia Municipal de Matosinhos — the Ribeira da Reguinha debate

Source: live assembly session, 20 July 2026 — a deputy takes the floor on the stream polluting Matosinhos Beach · Clip: 78 s · Pipeline output captured live — captions and the English voice below aired seconds after the words were spoken.
EN DUB
Layers
Playing the original broadcast — no accessibility layers.
On air

What each layer looks like to a viewer

Frames from the samples above, exactly as the caption layers render them on the broadcast.

Committee member listening in the parliamentary hearing room, no captions
Original broadcast. The parliamentary committee room as ARTV airs it today — reaching only viewers who follow spoken Portuguese.
President of Startup Portugal speaking, with English translated caption on screen
I will start by saying that Startup Portugal is aware of the fund of funds mechanism that
English captions on. The live translation as it aired, seconds behind the speaker.
Same hearing with Portuguese caption above the English caption
Deputado as várias questões que fez.Member of Parliament, the various questions you asked.
Both languages at once. Same-language captions for deaf and hard-of-hearing viewers, the translation for everyone else — one recognized segment feeds both.
Aerial view of Mary Dennison Park with Brazilian Portuguese caption on screen
Isso é ridículo, é tudo o que posso dizer, não posso acreditar que isso seja apresentado para dedicar 10 pessoas a um parque, isso é
Reversed direction. The Frame in Brazilian Portuguese — captions and a dubbed voice for Framingham's Portuguese-speaking community.
How it works

Sovereign by design

The whole pipeline runs on hardware you control — nothing leaves the building. Median end-to-end delay from podium to translated voice is about ten seconds.

Program feedSRT / SDI ingest
→
Streaming ASRspeech → text, live
→
Machine translationopen EU model, on-prem
→
Captions + neural voiceCEA-608 / DVB · dub track
→
Back to airsame stream, new audiences

After the live session, the same run packages video-on-demand deliverables automatically: SRT caption files for YouTube, Scenarist SCC masters for broadcast, and embedded closed-caption cuts.