Project spoken words as readable captions at talks and panels.

Public Events and Spoken captions succeed or fail on decisions made before the doors open. Choose the screen the room will read from, the person who will produce the text, and the window that will hold it. Walk the venue and look at the screen from the back row, the side seats, and the accessible seating area. Check the font size, the contrast, and the angle from each position. Confirm power, cabling, and a backup device. Write the plan down and share it with the speaker and the venue staff. A caption setup that works only from the front row is not ready for a public event.
Pick a screen that can be seen without effort. A projector or a large display above the stage usually works better than a laptop turned toward the room, because the text stays in one place while the speaker moves. Use a plain, high-contrast font and a line length that does not stretch across the entire width. Check the room lighting: a dark hall needs a brighter display, while a bright hall may wash out a dim one. Avoid placing the screen where stage lights reflect into it. If the venue has multiple screens, use the same layout on each so the text is consistent from every seat.
Name one producer and one backup before the event. The producer listens to the speaker and speaks summary sentences into the caption window; the backup watches for gaps and takes over if the producer loses their voice or connection. Put both people where they can hear the speaker clearly and see the screen. Agree on a cue for slow down and a cue for repeat. Give the producer a printed run of show with segment times and names. A producer who has to guess what comes next will fall behind, and a backup who has never rehearsed will need those first minutes to learn the setup.
Build a run of show with times, names, and segment headings. List the welcome, each talk, the panel, the questions, and the close, with a target time for each. Add the names of every speaker and panelist, the terms likely to appear, and the transitions between segments. Share the run of show with the producer, the host, and the venue staff. At the event, the producer can add a heading at each transition, which gives the reader a landmark and shows where the text has moved. A simple, shared order of work is the backbone of a readable event. Rehearse the handoff at each heading so the producer and host move together.
Ask speakers for their slides, notes, and key terms in advance. You do not need a full script, but a list of names, product terms, acronyms, and unusual places helps the producer anticipate what will be said. Build a glossary with the expanded form and the short form of each acronym, and note any pronunciation that is easy to get wrong. Add the glossary to the caption window as a reference section, or keep it on a second screen the producer can see. A ten-minute conversation about vocabulary prevents most of the misreads that would otherwise appear in front of the whole room.
Brief speakers on the habits that keep captions readable. Ask them to speak at a steady pace, finish their sentences, and pause briefly between ideas. Ask them to say their name before a question and to avoid stacking several points into one sentence. If two speakers share the stage, ask them to take turns rather than finishing each other's sentences. Explain that the text is live and that a short pause helps everyone, not only the deaf attendees. Most speakers follow these requests when they understand the reason, and a written version on the green room table helps them remember.
Prepare the panel before the audience arrives. Confirm the seating order, the speaking order, and who will moderate. Ask the moderator to name each panelist before they speak and to restate a question before it is answered. Agree on a hand signal for wrap up, and decide who will manage time if a panelist runs long. If a panelist uses slides, confirm that the caption screen does not cover them. Seat the producer where the panel's voices are balanced, because a producer who can hear only one side of the stage will produce a one-sided text. A short walk-through of the first two minutes is usually enough.
Run a ten-minute rehearsal on the stage. Ask one speaker to read a short passage that includes a name, a number, and an acronym, then watch the text appear on the main screen. Check the size from the back row and the contrast against the stage lighting. Test the producer's shortcut and the backup device. Practice one handoff between the host and a panelist, and one question read aloud. Note any setting that needs to change and fix it before the doors open. A rehearsal that catches one readability problem is worth more than a polished plan that was never tested.
Start the captions before the first sentence. The producer adds the event title, the date, and the first speaker's name as a heading, so the screen is already readable when people finish settling in. Keep the heading short and use the same format for each segment. If the event opens with music or a video, leave the heading in place rather than filling the screen with guesses. When the host begins, the producer follows the opening remarks and adds a marker when the first speaker takes over. A clear start tells the room that the text is live and that it is worth watching.
Pace the captions in short chunks during the talk. The producer follows the speaker's main points and writes summary sentences rather than attempting every word. Lead with the headline, add one supporting detail, and mark the transition to the next idea. When a speaker gives a number, repeat it in the text and keep it visible for a moment. When a speaker uses an acronym, expand it the first time from the glossary. If the speaker speeds up, the producer should shorten rather than drop the point. Sparse, accurate text is easier to follow than a dense wall of half-captured phrases.
Carry the same discipline into the panel. The moderator names each panelist before they speak, and the producer adds the name at the start of the line. Ask panelists to take turns rather than speaking over each other; when they do overlap, the moderator restates the point and asks one person to repeat it. The producer should not try to capture a collision, because a guessed line is worse than a short pause. Mark questions with a Q marker and answers with the panelist's name. This simple structure lets a reader follow a fast discussion without losing track of who holds the floor. The producer can also add a one-line summary when the panel changes topic or a new question begins.
Plan for the sounds that are not speech. Applause, laughter, a dropped microphone, and an off-stage comment all arrive as gaps or noise. The producer should mark a pause rather than inventing text, and resume with the next clear sentence. When the host summarizes an audience reaction, the producer can capture the summary as the host's line. If a video plays without captions, say so in the window and note the time, so readers know the gap is expected. Honest gaps are easier to understand than a smooth text that quietly skips a segment of the event. That way, nobody assumes the event stopped.
Collect questions through one channel. Use a microphone runner, a card, or a queue at the side of the room, and have the host repeat each question before answering. Repetition is not redundant; it gives every attendee the same starting point and gives the producer a clear sentence to capture. Ask the questioner to say their name and keep the question to one or two sentences. If the question is long, the host summarizes it and says so. A single, well-run queue prevents the scattered shouting that makes both the room and the text hard to follow. Write the repeated question at the top of the answer so the pair stays together.
Bring the microphone to the questioner rather than asking them to shout. A handheld microphone improves what the room hears and what the producer can follow. Ask the questioner to speak toward the microphone and to pause before the question. If the venue has no runner, have the host move to the questioner. Repeat the question even when it was audible, because the back of the room may have missed it and the caption window needs a clean line. The extra ten seconds is the difference between a question everyone follows and a question only the front row understands. A wired microphone is more reliable than a handheld that loses charge mid-question.
Manage time and volume. Set a limit for each question, such as one minute, and ask the host to enforce it kindly. Group similar questions and answer them together rather than repeating the same exchange. Keep a visible list of questions that were not answered, and tell the room where the follow-up will appear. If a question is off topic, the host can note it for a later channel instead of taking the whole room down a side path. These choices protect the schedule and keep the captions focused on the points most people came to hear. Announce the limit before the first question so nobody is surprised.
Make the question segment repeatable. Put the producer's question workflow on the run of show: collect, repeat, name, answer, mark. When the same team runs another event, they can reuse the workflow without reinventing it. Voiskey runs at the system level, so the producer presses a shortcut, speaks, and text appears in the caption window, with the same habit across venues and apps. A regular producer can download Voiskey on a backup device and keep the workflow unchanged. Repetition turns a stressful live segment into a routine the team can trust. Keep the marker short, such as Q and A, so it does not compete with the text.
Debrief with the deaf attendees before they leave, or send a short follow-up the next day. Ask which parts of the event were easiest to follow and which segments failed. Ask whether the screen was visible from their seat, whether the text was large enough, and whether the pace matched the speaker. Write the answers down while the event is still concrete. Do not rely on a general impression from the organizing team, who watched from the front and may have seen a different screen. The people who read the captions are the only reliable source on whether the captions worked.
Record the terms that were misread and add them to the glossary. Names, acronyms, product terms, and place names are the usual problems, and each event adds a few. Note the segment where each problem appeared, so a future producer can anticipate it. If a correction pattern worked well, write it into the run of show. If a speaker ignored the pacing brief, note that too, and plan a different briefing method next time. A glossary that grows with each event is one of the most useful assets a caption team can keep. If the same term fails at two events in a row, ask the speaker to introduce it with a gloss.
Update the runbook and the kit list. Include the screen layout, the font size, the backup device, the cable types, the producer cues, and the question workflow. Add the venue contact and the person who controls the house lights. Check the kit after every event and replace anything that failed. Store the runbook with the glossary so the next producer finds both in one place. A runbook written from memory after a successful event tends to omit the small fixes that made it successful, so update it while the details are still fresh. Keep a printed copy in the kit and a shared copy the team can open from a phone.
Train the next producer before the next event. Have them read the runbook, watch one rehearsal, and run one segment with a backup beside them. Ask them to download Voiskey on the device they will use and practice the shortcut until it is automatic. Pair them with the regular producer for a full event before they take the lead. Review the setup with the venue and the deaf attendees after each event, and change one thing at a time. A caption routine improves through small, tested changes, not through a complete redesign after every show. Set the next event date before the rehearsal ends, so the practice has a deadline.
Voiskey is an AI voice typing app for computers. A producer presses a shortcut and speaks, and the words appear as text at the insertion point in a window that can be shown on the event screen. It removes filler words and false starts, so the projected line stays easier for attendees to read.
Voiskey removes filler words, stutters, false starts, self-corrections, and grammar slips from spoken input. It also adapts punctuation and formatting to the task. The producer should still summarize the main point rather than attempting every phrase, especially when a speaker moves quickly or two panelists overlap.
Yes. Voiskey works at the system level, so you press a shortcut, speak, and text appears at the insertion point in any app or text field. The event team can keep the notes window it already projects, with the same heading and question workflow described in this guide.
Voiskey learns names, jargon, acronyms, and spelling preferences. A presenter name that is corrected once is more likely to arrive as intended later. Add every presenter and panelist to the event glossary, and keep the three-step repair habit for unusual names: say the correct term alone, use it in a sentence, then continue.

Download Voiskey for Mac and Windows to project clear spoken captions at your next event.
Use it across macOS, Windows, iOS, and Android