Everything Talk2Camera does

Script library

Write, edit and search your scripts. Favourites, folders, word count and an estimated read time on every one.

Import a script

Bring in plain text, Markdown, Word (.docx) or PDF. PDF text is extracted on the device, and it says so plainly when it meets an encrypted file or a scan with no text layer.

Script templates

Eight starting points — product announcement, tutorial, weekly update, a two-presenter interview and more — each a real spoken scaffold with slots to fill, in your own language.

Cloud script sync

Your scripts follow you between devices, reconciled by one shared ruleset on both platforms. Part of a subscription.

The built-in demo script

A two-minute script written to be read aloud: it tells you not to read like an announcer, invites you to stop mid-sentence so you feel the prompter wait, and leaves you with a real take for the editor. Hide it if you like — it never deletes.

Voice follow

The script scrolls as you speak. Pause and it waits; skip a sentence and it catches up. Recognition runs on the device.

Manual and timed scrolling

Drive the script yourself, or set a pace and let it run. Progress bar with the time remaining.

Typography and accessibility

Text size, line spacing, side margins and alignment, with five faces including Lexend and OpenDyslexic. Right-to-left scripts lay out correctly without a setting.

Prompter colours

Text, background, and a separate colour for words you have already said — guarded so a colour picked for one background cannot vanish on the other.

Prompter size and cover mode

Set how much of the screen the script takes. At the top of the range it covers the camera entirely and you show through behind it — for people who would rather not watch themselves while reading.

Mirroring

Horizontal and vertical, for beam-splitter teleprompter rigs.

Rig Mode

Prompter only — full screen, no camera, nothing recorded. It never even asks for camera access.

Cue markers

A line like [SHOW THE PRODUCT] becomes a stage-direction chip in the prompter: shown, acted on, never spoken — it can’t reach voice follow or the captions.

Keyboard and controller remote

Scroll, pause, change speed and start recording from a Bluetooth keyboard or a game controller.

Dialogue scripts for two presenters

Tag lines with a name and a colon and the prompter colours each speaker, announces every hand-over, and keeps voice follow exact — the tags are never fed to the recognizer. The AI writer drafts in this format on a toggle.

Apple Watch remote

Scroll, pause, change speed and start recording from your wrist.

Keyboard and game-controller remotes work on Android; there is no Wear OS app yet.

A second device as your teleprompter

Put the script on a tablet or spare phone and leave the recording phone where the shot wants to be. Both screens stay in step, and if the link drops the script carries on scrolling rather than stopping.

Pairing by scanning a code

Point one device at the other’s screen. The code works once and stops working after three minutes, and it carries everything the two devices need to find each other.

Works with no Wi-Fi network

The two devices connect to each other directly, so a field with no router and no signal is fine. No account, and nothing about your script leaves the two devices.

Two Apple devices pair with no network at all. Pairing with an Android second screen, the Android device is the one that shows the code — scan it with the iPhone and you are connected.

Switch cameras mid-recording

Switch between front and rear cameras while recording. The file stays one continuous clip — no cut, no second file, nothing to stitch.

Switch styles

Instant cut, a 100 ms quick blend, or a 250 ms dissolve. The style shapes the reveal; the recording never stops either way.

Resolution and frame rate

720p, 1080p and 4K at 24, 30 or 60 fps. Only what the camera actually reports is offered — nothing greyed in that the device cannot do.

Manual camera controls

ISO and shutter, exposure bias, exposure lock, manual focus, tap to focus, white-balance lock and stabilisation. Every control is re-checked against the incoming camera on a switch, so a value cannot carry over out of range.

Microphone selection

Choose your input. Switching it re-routes live capture without interrupting a take.

Pause and resume

Pause mid-take and pick it up again. The paused time is excluded from the file rather than appearing as dead air.

Audio-only takes

Record just the sound, with the same prompter and the same pause behaviour.

Two-presenter recording

Plug in a dual wireless kit — a RØDE Wireless GO II or a DJI Mic in split mode, or any stereo interface — and each presenter lands on their own channel. Selecting the interface is the whole setup; the built-in mic stays mono on purpose.

Recordings library

Every take with its own poster frame, length, size and date. Open any of them straight into the editor.

Select and delete

Tick several takes and remove them together. The confirmation says plainly that exports are separate files and stay where you put them.

Storage accounting

What your recordings cost, counted separately from exports and other files, with a warning before the disk becomes the problem — including how much room is actually left.

A take remembers its script

The script is saved beside the recording, so captions and title suggestions still work when you open that take next week — not only right after shooting it.

Trim on a zoomable timeline

Pinch to zoom. At full magnification one pixel is milliseconds, so an in-point lands on the frame you meant.

Cut a selected stretch

Drag across the waveform and cut. The timeline closes up, captions retime, and the cut stays switchable off afterwards.

Remove pauses

Long silences are found and proposed as cuts you review one at a time, with undo. Nothing is removed without you agreeing to it.

Edit the video by editing the words

The take as tappable sentences: tap one, cut it, and that stretch of video leaves the edit. Struck-through lines come back with a tap, every cut is reviewable and undoable, and the recording itself is never touched.

Sentences come from the script the take was read from, timed by the prompter, so it works on takes recorded with the prompter.

Lower third

A name and handle card over the opening seconds — two to eight seconds on screen with a gentle fade, remembered across takes, burned in at export like captions. The recording stays untouched.

Music that ducks under your voice

Lay a music track under the edit and it dips while you speak and returns in the pauses — broadcast fader discipline, applied for you. Breath-length gaps stay ducked so it never pumps.

Captions

Burn captions into the video or export an .srt, with style controls and per-line timing.

Timed from where you actually reached in the script while voice follow was on; estimated otherwise. Every line stays adjustable in the editor.

Merge takes

Join several recordings with a hard cut, a cross dissolve, or a dip through black — a true black frame at the midpoint, the chapter break of talking-head grammar. Audio crossfades under the join.

Looks and colour

Five presets with a strength control, plus exposure, contrast, saturation, temperature and tint. Two slots to save your own graded look and recall it on any take.

Transparent background export

A real alpha channel, so another editor can drop you straight onto their footage — no keying, no green spill.

Not offered on Android. Use the broadcast green or blue background instead and key it in your editor — the result is the same.

Background blur and replacement

On-device person segmentation — no green screen. Blur the room or replace it with a colour or a photo, including broadcast green and blue for keying elsewhere. If the device cannot produce a clean matte the export finishes with the original background and tells you.

Per-presenter levels

A two-presenter take splits into one lane per person, each with its own level line. Pull one presenter down without touching the other — the balance is set in the edit, not lost at the receiver.

Studio sound

One tap reduces room noise and evens out speech levels at export — a high-pass, a gentle gate and a slow leveller, tuned so nothing ever pumps. The recording itself is never touched.

Word-by-word caption highlight

Burned-in captions can pop word by word as they’re said — the style short-form feeds expect — timed inside each line’s real window, with the same styles and presets.

Caption translation

Turn your caption track into a second .srt in another language — same timings, one line per line, never merged or split. Part of a subscription’s AI allowance.

Two-track audio with level lines

Your take’s sound and an attached track, each with a line you drag across its waveform — the way a mixer works, not a fader. The bars redraw at the level you set, so a bed pulled under speech looks quieter as well as sounding it.

Attach an audio track

Lay music, a voiceover or a cleaned-up re-record under the edit, positioned where you want it. Preview and export both honour it.

Thumbnail and title help

Candidate frames scored for sharpness, exposure and whether you are looking at the lens, plus title suggestions from your script.

Chapters from your script

Start a line with # in your script and it becomes a YouTube chapter, timestamped to the moment you actually said it. Cut footage and the timestamps move with the edit. Copy the finished list straight into your description.

Your logo on your exports

Paid plans replace the Talk2Camera mark with your own logo in the same slot. Free exports carry a small mark that walks between the corners every few seconds — it opens bottom-left, clear of every platform’s buttons, and the preview shows exactly the walk the export takes.

Auto-reframe

Export a portrait take to square or 16:9 with a crop that follows you — a slow glide, never a snap, framed with proper headroom. It holds when you leave frame, and captions and marks stay put.

Platform export presets

TikTok, Instagram, YouTube and LinkedIn, each with its own resolution cap, bitrate and duration guidance. The vertical feeds get a 9:16 frame; the others keep your shape.

No re-encoding when nothing needs it

A plain trim that already matches the preset copies its frames instead of compressing them again — faster, and no generation of quality lost for nothing.

Cancel an export

Stop a long export mid-run. When it finishes it tells you what actually happened — re-encoded or copied, at what size, and whether the frame was cropped.

Publish to YouTube

Upload and post without leaving the editor.

Share, or save to your photos

Send the export anywhere, save it to your photo library, or take the original unedited file somewhere else untouched.

Export subtitles as .srt

A separate subtitle file for platforms that want one, instead of burning the words into the picture.

Stream to any RTMP server

Paste the URL and key from your platform’s live dashboard — YouTube, Twitch, Facebook, Instagram or your own server. Reconnects automatically with backoff.

Record and stream at once

Start and stop the recording without touching the stream. Camera switching works mid-stream too.

One-tap platform connect

Sign in to YouTube, Twitch or Facebook and the app creates the broadcast for you. Part of Studio; pasting a key from your own dashboard stays free on every plan.

Coming soon as a Studio feature. Streaming with your own stream key from your platform’s dashboard stays free on every plan.

Delivery metrics

Pace in words per minute, filler-word count, and how much of your script you covered. Measured on the device from the take itself.

Eye contact and framing

How much of the take you spent looking at the lens, your longest look away, and how steady your framing was. When no face is found it reports zero rather than estimating.

A composite you can argue with

One figure, with every input that fed it and its weight shown beside it. Described as a coaching heuristic from what the recording shows — not a measurement of you.