Documentation

How it works, step by step.

How to use Freally Teleprompt, from the first launch to a read on the rig.

Getting started

Freally Teleprompt is a teleprompter for Windows, macOS and Linux, for presenters, streamers, podcasters, actors, singers and rappers. It opens as one window — your scripts, the editor, the preview and the projector controls — and needs no account.

The first time it opens, it shows its licence agreement. Scroll to the end, then choose I Agree, or Decline & Quit. The app cannot be used until you agree; your answer is remembered, and you are asked again only if the agreement changes.

A one-minute tour then shows the editor, the pace controls and the projector. Type or paste a script, set your pace and press play.

There are no downloads on this site, and the source code is private.

Write a script

Scripts are .ftscript files: plain text in one folder, readable in any text editor. Open Scripts on the toolbar to create, open, rename and delete them. A recent list takes you back to what you were working on, everything is saved as you type, and the app reopens the script you had open last.

Type in the editor or paste from anywhere. Under the read-time estimate, a live word and character count shows what you will actually perform — skipped labels and pause marks are not counted — and a warning marks any line long enough to be a whole paragraph.

Import a document

Press Import on the toolbar and choose a Word document (.docx), an RTF, a PDF, a plain text file or a Markdown file. You see the text it produced, and a report, before anything is saved.

Paragraphs are kept; bold, fonts and colours are flattened, because a prompter shows one typeface. Curly quotes, long dashes and ellipses become plain ones, and a file saved in an older encoding is read correctly, and the report says so.

Pictures, footnotes, comments, headers and footers, link addresses and embedded objects are counted and listed, never dropped quietly. A PDF has no such structure, so for a PDF the report says it could not tell.

Name the script — a name is suggested from the file — and press Import. It is added to your library and opened, and it never overwrites a script that already exists. Markdown headings arrive as # lines, so a Markdown script comes with its section markers.

A typeset dash becomes an ordinary hyphen, and two of them never turn into a pause, so an imported play does not arrive full of pauses nobody wrote. A text file you wrote with real pause marks keeps them.

Find and replace

Press Ctrl+F, or Find on the toolbar, to search the open script. The match count updates as you type; Enter goes to the next match and Shift+Enter to the previous one, wrapping round at both ends.

Match case and whole words only narrow the search, and what you type is searched for literally. Replace changes the current match and Replace all changes every one; either is undone with a single Ctrl+Z.

Section markers

Begin a line with # — # Verse 2, # Q&A, ## Bridge — and it becomes a section marker. Any line that is one of your skip words, such as [Chorus], is a marker too. There is nothing to set up.

Markers appear in a jump list beside the seek bar, which shows the section you are in, and are ruled onto the seek bar itself. Next and Previous walk them; Previous goes to the top of the section you are in first, like a previous-track button. Choosing one moves every screen, the projector included.

Pause chips

Type -- (a space, two dashes, a space) for a short pause, or --N for a hold of N seconds. Each becomes a chip in the editor: the cursor sits only before or after it, one Backspace removes it, and Shift with an arrow key selects it whole. Set the length of a plain -- under Settings → Reading.

Copy and paste carry chips intact, and pasted text has its spacing tidied so every pause lands as a clean chip. While the script plays, each pause lights up as the read reaches it and the scroll holds for its length — on the preview, the projector and the network mirror alike.

Pacing: characters per second or BPM

Set the speed in characters per second: a real reading rate that every screen follows, whatever its font size. Or switch to BPM, from 20 to 250, to read lyrics and verses in time with a track.

Both drive the same engine, and changing either while the script runs re-paces it at once.

Tempo, bars and the click

A tempo becomes a reading speed through one number: how many characters you read in a beat. Under Settings → Timing → Your own tempo, press Tap three times or more at the tempo you perform at, then choose Use this tempo; Back to the default undoes it. The BPM box then offers only the tempos the engine can reach.

With BPM on, or the click switched on, the bar and beat appear above the seek bar and bar lines are ruled onto it. Set the beats in a bar, from 2 to 12, under Settings → Timing.

Tick Play a click at the current tempo for a quiet tick on every beat, higher on the first beat of each bar; the start countdown becomes its count-in. The app makes the sound itself and times every tick against the audio clock, so it does not drift. It stays silent while read-aloud is speaking.

Rehearsal and the timing report

Tick Rehearse, and time my read under the transport, press play and read aloud as you would on the day. Pause, go back, change the speed: all of it is part of the measurement.

Untick it and the rehearsal report opens: one row per section, with the time it should have taken at your pace, the time it took and the difference. Sections are your paragraphs, or your section markers when the script is divided by them. If you read at a different pace, the report offers that speed, and one click applies it.

While you rehearse, a quiet warning appears once the read is three seconds or more behind or ahead of its plan. Rehearsal records timings and nothing else: it opens no microphone and keeps no audio, and it cannot run together with read-aloud.

Words you do not perform

Section labels — Chorus, Verse 2, Bridge — are landmarks you see coming but never say. List them under Settings → Timing → Words you do not perform, one per line. The list starts empty, because it changes how a script is timed.

A line that is nothing but a label is skipped whole: brackets, hashes, a colon and a trailing number are seen through, so Chorus, [Verse 1], ## Bridge and Chorus: all match. The same word inside a real line skips only itself. Matching ignores capitals and only matches whole words.

Skipped text stays on screen, dimmed, so the talent still sees it coming, and read-aloud never says it. Because a label takes no time, the scroll passes it at once — and your lyrics stay on the bar you wrote them for.

Playback and the transport

Play, pause, stop, rewind and fast-forward are on-screen buttons; hold rewind or fast-forward to keep going. Spin the mouse wheel to scrub, click a word to start from it, or hover over the seek bar to see the words at that point.

The seek bar shows the elapsed and total read time, pauses included. A reading guide keeps the current word on the reading line, and the last lit word never scrolls out of sight.

Shortcuts, pedals and hotkeys

Open Shortcuts on the toolbar. Press the binding you want to change, then press the key, remote or pedal you want; Escape leaves it as it was. Search by a command's name or by the key itself. If two commands share a key, both rows say so; clear one with ×, or choose Restore defaults.

Most foot pedals and presenter remotes send ordinary keys, usually F13 to F24, so they bind with no driver and no setup. Bindings follow the physical key, not the letter printed on it, so they survive a change of keyboard layout.

The Anywhere column asks your system to deliver a key even when another app is in front — a camera app, a DAW or OBS. Nothing is claimed until you set it, and a key another program already owns says so. Under Wayland on Linux, apps cannot claim keys system-wide, and the dialog tells you.

While you are writing, the space bar, the arrow keys and Home type as normal: a bare key runs its command only outside the editor.

Start countdown

Set Countdown before starting under Settings → Reading, and pressing play counts down that many seconds first, on the preview and the projector, so the talent has a moment before the first word moves. With the click on, the countdown is also the count-in.

The projector

Press Projector and choose the display the talent reads from: the script fills that screen edge to edge. Turn on the mirror for beam-splitter glass, where the reflection would otherwise read backwards. The projector has its own transport, seek bar and keys for the person reading.

The preview and the projector lay the script out at the same width and scale it to fit, so they break lines in the same places. Check that the mirrored text reads correctly through your own glass before a shoot.

The network mirror

To show the script on a phone, tablet or second laptop, open Settings → Network → Mirror to my network and tick Mirror the script to devices on my network. Tick Allow other devices, not just this computer as well — without it, the mirror answers only on this computer — and press Apply. A link and a QR code appear: scan the code, or open the link in a browser on the same network.

The mirror follows the same engine, so every device is on the same word as the operator. Each device sets its own text size with A− and A+ without changing anyone else's screen.

The page is served by the app itself, straight to your devices — never through this website, which a browser would not let reach a device on your network anyway. It is read-only: no browser can start, stop, seek or edit anything. Its link carries a key that changes every time the app starts. It is plain HTTP, so the key is not encrypted on the way: use it on a network you trust, not on public Wi-Fi.

If a device cannot connect: check both switches; make sure it is on the same network, not on mobile data or a guest network that keeps devices apart; allow the app through your firewall for private networks when your system asks; and if Settings says the mirror could not start, change the port and press Apply again.

Read aloud

Tick Read aloud with per-OS speech synthesis to hear the script in your system's own voices: Windows OneCore or SAPI, macOS AVSpeechSynthesis, or Speech Dispatcher or espeak-ng on Linux. No voice is bundled with the app.

The transport, the seek bar and a click on a word drive the speech, and the highlight follows it — word by word where the system reports word boundaries, at your pace elsewhere. Pauses become short silences, and -- is never read out as "dash".

Read-aloud is for the preview only: it never moves the projector. The checkbox is locked while a read is playing or paused, until you press Stop. Check that it speaks on the computer you will present from, especially on Linux.

Autocomplete

As you write, the rest of the word appears dimmed ahead of the cursor; finish a word and press space, and it can offer a whole common phrase. Tab accepts, Esc dismisses, and typing on ignores it.

Each of the 18 languages has its own word and phrase lists — 463,985 words and 344,891 phrases in all — inside the app, and only the language you write in is loaded. Under Settings → Editor, turn it on or off and choose its language, which can differ from the app's.

The lists are built from the Tatoeba sentence collection (CC BY 2.0 FR), credited in the NOTICE file that comes with the app. It is a dictionary lookup — instant, predictable and offline, with no model.

Dictation

Open Settings → Voice, tick Write my script by speaking and press Apply. A round record button appears above the script: press it and talk, and what you say is written where your cursor is. Press it again to stop; the microphone is open only in between.

Dictated text is ordinary script text, and Ctrl+Z takes back one thing you said, not the whole session. Expect to tidy punctuation and names: it is a quick way to a draft, not a finished script.

Dictation understands English only. It uses Vosk, an open-source speech recogniser, with a small English speech model — a trained neural network of about 40 MB — that ships inside the app. It runs on your computer, works with no connection, uploads and downloads nothing, and saves nothing but the text you see. It is off until you switch it on. If the script is mirrored to your network, dictated words reach those devices as they are written, like typing.

Appearance, languages and settings

Under Settings → Appearance, set the typeface, weight and text colour, the side margins, the line spacing and where the reading guide sits; the font size is under Reading. They apply to the preview, the projector and the network mirror together.

Settings work on a draft: nothing changes until you press Apply, and Cancel restores what you had. Choose one of 18 languages, or Same as my system. The theme is Dark, Light or Same as my system; the reading surface stays high-contrast whichever you choose.

With Minimize to the system tray on, under Settings → General, minimising puts the app in the tray instead of the taskbar. Show the tour again, also under General, replays the welcome tour.

The tour and accessibility

The first time the app opens, after the licence, a four-step tour shows the editor and its pause marks, the pace controls and the projector. Skip, Esc or a click outside it closes it for good; Settings → General → Show the tour again brings it back.

Every control shows a focus ring when you reach it from the keyboard. Tab stays inside an open dialog, Esc closes it, and focus returns to the button that opened it. The seek bar is a real slider: the arrow keys move one character, Page Up and Page Down a screenful, and Home and End go to either end.

If your system asks for reduced motion, animations and blur are dropped while the script itself still scrolls; with reduced transparency, dialogs dim the background instead of blurring it.

Problems and updates

When something goes wrong that the app can tell you about, its message has a Report button beside it. After a crash, the next launch shows a strip with Report and Dismiss.

Report opens the Freally Teleprompt report page in your browser, with the technical details already filled in: you can read them, but not change them. Add what you were doing and, if you like, an email address. Nothing is sent unless you press Send on that page, and nothing is ever sent in the background.

Once per launch, the app asks GitHub whether a newer version exists. If there is one, you see its version number and release notes and choose yes or no; an update is checked against a signing key inside the app before it installs. When you are offline or already up to date, it says nothing. Check for updates on the toolbar asks again at any time.

Privacy and the network

No account and no telemetry. Your scripts are .ftscript files on your computer, where you save them, and the app's author never receives them.

The only request the app makes to the internet is the update check, which sends no script, no settings and no identifier. A problem report leaves only from the report page in your browser, when you press Send. The projector and the network mirror stay on your own displays and your own network.

Machine learning is used in one place: dictation's speech model, which runs on your computer and is off until you switch it on. Autocomplete is a dictionary lookup, read-aloud uses your system's own voices, and nothing you type or say is sent to an AI service.