Bringing real-time captions to 7 million local government constituents

The company

Tightrope Media Systems builds accessibility technology for local government: the city and county communications teams, clerks, accessibility coordinators and IT leads who run public meetings, alongside the PEG (public, education and government) stations that record many of those same meetings for broadcast.

Local government meetings are among the hardest rooms to caption well. Officials, staff and residents speak in turn and over each other, accents and first languages vary, and the person running the system is often a clerk already juggling several other things during the meeting. Whatever appears on screen is what a resident gets, the first time.

Its MediaScribe platform runs across 22 government and institutional customers in 13 states, reaching communities of more than 7 million residents.

The challenge

Two different needs land on the same small team. Captioning for deaf and hard-of-hearing residents is an ADA-driven obligation: under ADA Title II, public entities face enforceable deadlines for accessible meeting video, with WCAG 2.1 Level AA as the technical standard.

Language access for residents who don't speak the language a meeting is conducted in is a different matter — most cities pursue it as equitable communication rather than compliance. Both have to happen in the room, while the meeting is underway.

Live human captioning covers the obligation, but it's costly and hard to schedule across every meeting, in every jurisdiction, every week. Cheaper automated alternatives tend to fail on exactly the audio these meetings produce: regional accents, cross-talk during public comment, and a dense stream of local names and agenda items a general-purpose model has never encountered.

Unlike a recorded transcript that can be cleaned up later, in-room captioning has almost no margin for error. If the caption is wrong, the person reading it gets the wrong information in real time. A system that only performs for some accents leaves the jurisdiction exposed under the same law it was meant to satisfy.

The solution

MediaScribe Live was built for that room, not for broadcast audio in general. Its primary job is in-room accommodation: dedicated caption displays, live captions pushed to attendees' own phones, and real-time translation for residents who don't speak English. Broadcast captioning — the version PEG stations air or stream afterward — runs off the same system as a secondary output.

Under the hood, an on-premise hardware gateway ingests live meeting audio and feeds a cloud-based captioning engine built on Speechmatics' speech recognition, so stations turn on captions, translation or transcription without adding a new vendor relationship or separate hardware.

Accuracy on that difficult audio starts with the Speechmatics engine itself, tuned further by two features built into MediaScribe: a custom vocabulary list that stations populate with local names, agencies and recurring agenda terms, and a caption editor that lets an operator catch and fix errors.

Heavy accents, cross-talk and unfamiliar local names are exactly the conditions Speechmatics' speech recognition is built to handle — which is what MediaScribe draws on rather than a general-purpose transcription model.

One government video team described speakers with heavy accents where MediaScribe stayed consistent and ran faster than most of the live captioning they'd used before.

The results

That accuracy is now measurable at scale.

MediaScribe's 22 government and institutional customers represent communities totaling more than 7 million residents across 13 states. Applying U.S. Census ACS demographic rates to those communities, an estimated 1 in 6 residents — more than 1.2 million people — has a hearing-related or language-access need that live captioning directly serves.

Tightrope knows that this figure understates the real need. Fewer than 20 percent of those with hearing loss seek accommodations for their condition.

The benefit isn't confined to city halls, either. At a Minnesota Association of Community Telecommunications Administrators (MACTA) conference, Eric Strouse, station manager at CTN in Coon Rapids and MACTA's president, pointed out that captions help everyone in a noisy room, not only attendees with hearing loss: glancing at accurate captions kept audiences engaged when the room itself made listening hard.

That combination of accuracy and reach has drawn recognition beyond Tightrope's own customers. In April 2026, NAB Show named MediaScribe a Product of the Year winner. By Tightrope's own review of NAB's award history, no other recent honoree has been built specifically for local government.

For smaller stations, the difference shows up in what's now possible at all:

What's next

Tightrope is extending the same speech recognition foundation into new formats. MediaScribe Live's spoken translation feature pairs that engine with neural text-to-speech, so attendees can hear a meeting in their own language rather than only read it, available today in 19 of the platform's 70+ supported languages.

Some residents can't use captions at all, and giving them a spoken option asks more of the speech-to-text underneath, because a listener can't go back and reread a garbled name the way a caption reader can.

For a resident trying to follow a zoning decision or a school board vote, that principle comes down to something simple: a caption or voice translation, accurate enough to trust, is the difference between being in the room and being able to take part in what happens there.