How to read a script on camera
Put the next line near the lens. Give yourself enough time to say it.
Free toolTeleprompterRead your script at your own pace. Rehearse a paragraph, mirror the reader and keep your place as you resize.Open the teleprompterSet the screen before you set the speed
Place the reading point close to your camera lens, then test it from your actual recording position. Enlarging the text helps only if you can still follow a line without obvious sideways glances.
- Set your camera at the height you intend to use.
- Put the reader immediately below or beside the lens. Start with a narrow text column.
- Read one paragraph aloud while recording a test.
- Watch the test for downward glances and rushed phrases. Adjust your setup before recording the full script.
A phone on a stand below a separate camera can be a simple reading screen. Reading from that phone and recording in another app on it is a different setup. A web page cannot overlay itself on the camera app.
Choose notes, a scrolling reader or prompter glass
Use the least structured aid that keeps the facts right. A full script suits precise wording; notes leave more room to speak freely.
| Method | When it helps | What to check |
|---|---|---|
| Short notes beside the lens | You know the subject and need reminders. | A glance can be obvious. Keep notes short enough to recognise quickly. |
| Browser teleprompter | You need specific wording or want to rehearse a timed script. | The screen position still controls your eyeline. A timed reader cannot hear hesitation. |
| Beam-splitter glass | You want to read a reflection directly in front of the lens. | Mirror the screen and check alignment. Reflections and rig position need a physical test. |
| Voice-following app | You need the text to wait for your delivery. | Confirm language support, correction controls and where recognition processes your audio. |
Autocue’s explanation of prompter glass describes the reflection in front of the lens. Mirror mode reverses text for that reflection. It does not make direct screen reading look more natural.
Find your pace with an actual paragraph
Time yourself reading a representative paragraph aloud. Divide the word count by the seconds, then multiply by 60 to find a starting WPM.
For example, 70 words read in 35 seconds gives 120 WPM. If you need extra pauses for emphasis, allow for them when planning the recording.
| A 120-word script at | Calculated reading time |
|---|---|
| 100 WPM | 1 minute 12 seconds |
| 120 WPM | 1 minute |
| 150 WPM | 48 seconds |
| 180 WPM | 40 seconds |
These are arithmetic examples, not measurements of a speaker. A countdown adds preparation time. Numbers, acronyms, pauses and pronunciation can change how long the spoken version takes.
Check what a WPM control actually promises
A constant pixel speed is not a constant speaking pace. Larger type creates more lines, and short lines hold fewer words.
floi advances through word positions. Each rendered line receives time in proportion to its words. It remeasures the text after a size or window change and keeps the same position in the script.
- A short final line passes more quickly than a full line.
- Blank lines create a visual paragraph gap. They do not insert a timed pause.
- All displayed words count, including stage directions. Use Pause when you need time outside the script.
What our browser runs showed
On 30 October 2026, we read a synthetic 600-word script at 300 WPM in Chrome on an M1 Max MacBook Pro. We sampled the displayed word position twice during each run.
| Setup | Observed interval | Average pace |
|---|---|---|
| 42px type, 28-character column | 160.47 words over 32.174 seconds | 299.25 WPM |
| 88px type, 12-character column, both mirrors | 198.44 words over 39.631 seconds | 300.43 WPM |
The position display updates every 100 ms, and browser-control timing adds uncertainty. These checks support similar average pace across these two layouts. They do not measure a speaker or establish performance on every browser.
Make a paragraph easy to retake
Separate the script into small paragraphs that make sense on their own. When you stumble, stop and begin that paragraph again.
In floi, a blank line creates a paragraph shortcut. Choose it, look back at the camera and start. The countdown applies to that deliberate retake, while ordinary Resume continues immediately.
- Rewrite any sentence that is awkward to say aloud.
- Keep one idea in each paragraph.
- Leave a short quiet moment before speaking again. That gives you a clearer editing boundary.
- Review the recording before deleting earlier takes.
Keep your script out of the shared picture
Share a specific tab or window that contains the presentation. A whole-display share includes any visible prompter on that display.
floi’s desk can put the reader and recorder beside each other, but that arrangement does not hide either window from capture. A separate monitor or separate device can give you a clearer boundary.
The Screen Capture API documentation explains that the browser asks you to choose the captured surface. floi cannot silently choose a safer surface for you.
Compare the prompting options
Choose a tool around the part of recording that needs help. Several free readers already cover scrolling and mirror mode.
| Option | Useful capability | Boundary |
|---|---|---|
| floi | Paragraph retakes, WPM pace, local TXT/MD import and a separate reading view. | Timed scrolling only. No camera capture or voice following. |
| Tella’s free reader | A compact WPM setup and separate reading player. | Our synthetic script reached the player and its countdown. File import was not visible. |
| LilPrompter | Several scripts and two reading-setting profiles. | Its scroll speed uses a numeric scale. Our script opened in its separate app and played. |
| CuePrompter | Horizontal and vertical mirror, colours and text alignment. | Our script opened in a separate player with a promotional banner. We did not verify its recording output. |
These interface observations used an original 35-word script in Chrome on 30 October 2026. They do not establish recording quality, speech recognition accuracy or the privacy of each service.
Understand voice following before enabling it
Voice following requires a speech-recognition path. Browser recognition can send audio to a service, depending on implementation.
MDN documents a processLocally setting for on-device recognition, with limited browser availability. Its existence does not prove a particular tool uses it. floi’s teleprompter uses no speech recognition.
Keep an independent script copy
Save the final script as text before your session. Browser storage can be cleared, and a private browsing session may not retain it.
In floi, typed or pasted scripts save to your private account when signed in. Imported text and its edits stay on this device. Opening a TXT file never makes it an account draft.