Load your script
Import a WebVTT file prepared by a human captioner. Speaker labels like RACHEL: or 1ST WITCH: are supported. That file is the quality bar for every show.
Watch us make magic
Your script, in time.
Make theatre events accessible to deaf and hard of hearing audiences. Display the right caption line at the right time automatically.
ASR transcript
Waiting for speech…
Decision
Current caption
Waiting…
Live theatre captioning is hard.
Quality is human. Scale is the bottleneck. Every performance should be captioned.
-
01
A human captioner owns the words
Someone skilled must prepare the caption file. They are the arbiter of quality for the audience: clarity, speaker labels, and the text that make the performance accessible. The output is only as good as that input. Watch us make magic does not replace that expertise; it depends on it.
-
02
Today, every performance needs an operator
Current practice usually needs a caption outputter in the room for each show, advancing lines by hand. That is a skilled role, and costly. As a result, the number of accessible performances stays low. Most nights of a run never get captions at all.
-
03
Invest once in the file. Caption every night.
Watch us make magic asks for a little more care up front: a human captioner prepares the caption file, then you can tune the matching settings from a rehearsal recording. After that, every single performance can be captioned: the same approved text, followed live, without a dedicated outputter for each performance.
How it works
Four steps from script to every screen in the building.
Connect your audio
Stage mic, mixing desk, or induction loop audio feeds into the app. Any audio input device works - the better the audio, the better the captions.
Start listening
On-device speech recognition finds where the performance is in the script, then advances, holds, catches up, or marks off-script - all without the operator needing to do anything.
Captions everywhere
Accurate and timely captions are provided to the audience via a projector, their own phone, or venue-provided display devices. Captions can also be broadcast on the venue network, to OSC show control, or via webhooks.
Built for the operator, not against them
Automation that waits for evidence, with a human always one keypress away.
Script following with Laya
A dedicated decision model follows the performance line by line.
It identifies patterns, spots ad-libs and pauses when needed. When lines are skipped, it automatically catches-up and shows any missed text at a readable pace.
Automated cues like [APPLAUSE] and *BANG*can be timed to display after the line they follow.
- Carefully progressing through the script, never jumping ahead of the performers
- Recovers when the company returns to the script
- Shows your approved spelling and names, not ASR improvisation
Operator control
Live status shows ON SCRIPT, UNCERTAIN, or OFF SCRIPT with confidence. The decision panel lists competing lines and matched words. Overrides with ← / → or Prev / Next. Never lose script position when you type a manual announcement.
Audio for real rooms
Induction loops and mic'd performers. Speech gate adapts to room noise; gain boosts quiet voices up to 20 dB. Choose a speech model for speed or accuracy, plus custom vocabulary for character names and odd spellings.
Tune from a rehearsal recording
Tuning does not write the caption file. You bring an approved WebVTT and a rehearsal recording; Watch us make magic transcribes the audio and searches for matcher settings that fit that show. It is designed to penalise early jumps, so captions stay with the performers rather than racing ahead. After a real performance, session logs let you replay every recognition pass, decision and manual override against new settings.
- Uses your existing caption file plus a rehearsal take
- Optimises settings; favours accurate timing over early advances
- Replay a logged show to refine before the next house
Where it fits
Anywhere a prepared script meets a live room.
- Captioned theatreStage screens, seat-back devices, or patrons’ phones.
- Talks and panelsCultural institutions with prepared transcripts.
- Touring showsTune once, run every night with less operator strain.
- Opera, musicals, verseExact text matters more than a rough transcript.
- Ceremonies and servicesAwards, graduations, AGMs, set readings.
- EducationLectures and talks with scripts or slide notes.
- Multi-screen venuesStage, lobby, stream, and phones from one operator.
Nothing leaves the building.
Language models are downloaded and run locally, Watch us make magic runs fully offline. No cloud, no per-minute fees, no performance audio on someone else's servers. Decisions land in about 50–120 ms, as fast as speech is recognised.
- Platform macOS desktop, Apple Silicon
- Script WebVTT with optional speakers
- Models Whisper + Laya, local after download
- Network Only needed for first model fetch
Outputs that meet the rest of the show
Same caption line, four routes.
Full-screen window
Native caption display for the operator or house screen.
Local network page
Any phone, tablet, or screen on the venue LAN. No app install.
OSC
UDP out to QLab, lighting desks, and show control.
Webhooks
JSON or plain text with custom headers for streams and custom accessibility platforms.
Why not the usual tools?
| Compared with | Watch us make magic |
|---|---|
| Live speech-to-text | Shows your script text: correct names and spelling |
| Manual caption operator | Automates line-following; the operator supervises |
| Cloud captioning | Fully offline, private, no recurring per-minute cost |
| Simple timecode playback | Follows the performers’ pace, including skips and pauses |
FAQ and specs
What platforms does it run on?
macOS desktop app for Apple Silicon. Audio from any input device: USB interface, mixer, or loop receiver.
What script format do I need?
WebVTT (.vtt), with optional speaker labels. Load once, reuse for every performance of the show.
Which languages are supported?
Speech recognition is production-ready for English (UK/US). Additional languages including French, German, Spanish, Italian, Portuguese, Dutch, Japanese, Korean and Chinese are available in the speech models; English remains the best-tested path today.
Can I tune it from a rehearsal recording?
Yes. Tuning uses your existing caption file and a rehearsal recording. It does not create the captions; it searches for matcher settings that favour accurate timing over early jumps. Session logs let you replay a real show against new settings.
How is it priced?
Subscription per application for commercial use. Registered charities and Arts Council England or Wales funded venues (or equivalent cultural bodies elsewhere) can apply for a free license. We ask those free-license partners to share testing feedback from real performances.
Who qualifies for a free license?
Registered charities, and cultural venues supported by Arts Council England or Arts Council of Wales, or by an equivalent public cultural funding body in another country. If you are unsure whether you qualify, get in touch and we will say plainly.
Can you help set this up at our venue?
Yes. We can help get Watch us make magic ready for your performance: connecting the audio feed, routing captions to your screens or LAN displays, and tuning against your caption file. Say if you want on-site or remote support when you get in touch.
Free for charities and cultural venues.
Registered charities and venues funded by Arts Council England or Wales (or an equivalent cultural body outside England and Wales) can use Watch us make magic under a free license. We can also help set things up in your venue for a performance: audio feed, display routes, and a tune against your caption file. In return, we ask for honest input from rehearsals and shows so the tool gets sharper for D/deaf and hard of hearing audiences everywhere.
Apply for a free license- Who Charities and Arts Council (or equivalent) cultural venues
- What Free application license for captioned performances
- Help On-site or remote setup for your venue and show
- Ask Feedback from real testing, not a marketing quote
- Also Commercial venues: subscription per application
Put the right line on screen, every time.
Book a demo, apply for a free license, ask for venue setup help, or join as a testing partner.