Captions

Captions are a video's spoken words on screen, one line at a time, timed to the recording they were heard from. Add them with the Captions tile in the video's Add menu. A caption is one item with its own row on the timeline: its words, which speaker says each line, which words stand out as keywords and which sounds it labels are its settings, and how each speaker's words look is that speaker's text look.

A caption has five tabs: Captions, its words, where they come from and how they look — the list of its speakers at the right end, each name written in that speaker's look, the strip of words beside it with Source and Find in one column at its start, and the strip of looks a press on a speaker opens; Effects and Border, the same as a text item's, over the whole caption; Animate; and Transform, where it sits and how wide it is. A caption made from a sound shows its words wherever that sound plays, so it follows every cut and trim of the footage on its own.

Put the spoken words on screen

Captions on the Add tab puts a caption on the video at the playhead, three seconds long, reading the one word Captions, and opens its settings along the bottom on five tabs: Captions, Effects, Animate, Border and Transform. The Captions tab is where the words come from and how they look: type straight into it and the caption shows what you typed, or fill it from what is said in the video.

The container at the tab's right end lists the caption's speakers — one to begin with — each row reading the name of the look that speaker wears, written in that look, so Bold reads bold; with several speakers, the speaker's own name stands beside it. Source and Find share one narrow container at the start of the strip of words, and scroll along with it. A new caption reads Source: none, and once you type into it, manual. Press Source and that container widens in place, Source lit, with the chooser beside the two words; the word cards stay to its right. Press Find instead and the search stands in the chooser's place; pressing the other word switches between the two, and pressing the lit word closes it. To hear them, pick the Sound to listen to, the Language it is spoken in (or Detect to have it worked out), what to Hear — What the video plays of that sound, or the Whole recording — and your Names, then press Run. To bring in a caption file instead, press Import a file. Listening happens on your device and costs nothing. While it listens, Cancel stands in Run's place and the Source cell shows how far it has got. When it finishes, Source reads the name of the sound the words came from; press Source again, or Escape, to go back to the words. Running it over words you typed replaces them, and it tells you so first.

Names is your own list of names and words that should always be spelled your way — a person, a product, a place — and every caption you make from speech spells them right. Fix the same misheard word to the same spelling three times and the caption offers to remember it: Remember "…"?, with Add to Names.

Beside From audio, the chooser's top row picks the other settings about the words, the same for every speaker: Lines, Keywords (which words are marked), Speakers, Sounds and Translate, each described below.

If the sound changes after its words were heard, Source shows a small mark. Press it and choose Update: the words are heard again, and everything you did to them — fixed words, keyword marks, emoji, speakers, where lines break, and anything you set on one caption — stays wherever the words still line up with the speech.

Captions follow the footage they came from

A caption made from a sound is tied to that sound, so its words stay on the voice through every edit. Cut a clip and the words in the part you cut disappear with it; trim it, slide it, speed it up or slow it down, or reverse it, and the words go along; duplicate it and its words show on both copies. There is nothing to run again afterwards.

That is why a caption made from a sound is not moved or trimmed by dragging its row: its row runs the length of the video, and its words show wherever the sound plays. A caption you typed, or imported without a sound, has its own time instead — drag its row to move it and its ends to trim it, like a drawing. Trimming its start hides its first words rather than shifting them.

To move a caption made from a sound on its own, press Source and choose Remove source. Its words stay exactly where they show now — where the sound plays more than once, where it plays first — and Source reads manual; from then on it moves and trims like a caption you typed, and later edits to the clip no longer carry its words.

A caption holds many short captions, one after another, and its row on the timeline shows each one as its own cell, filled with as many whole words as fit, the last one fading out. A short cell shows a word or two, and a very short one is just a mark where one caption gives way to the next; zoom in to read more. Tap a cell to pick that caption: the playhead jumps to its first frame and the picture shows it; tap it again to go back to the whole caption. Drag a cell's edge to change when it starts or ends — the word at that edge moves with it, and when you drag into the neighbouring caption its boundary word gives way too, down to its shortest — or drag the middle of the picked caption's cell to shift its words in time. Dragging any other cell moves the whole caption: along the timeline when it has its own time, and up or down among the rows.

Fix and rearrange caption words

The Captions tab lays the words out as a strip of cards, one per caption in the order they are said, each showing the caption exactly as the picture draws it — its type, its lines and its width — scaled down, with a + card at the end that starts a new caption after the last. A long video's strip scrolls sideways.

• Fix a word — tap into a card and type, exactly as in a text box. A corrected spelling keeps its timing; when the number of words changes, that caption and its neighbours are matched to the voice again, so a word you add lands where it is said. Undo takes it all back in one step.
• Break a line — press Enter. Enter on an empty line starts a new caption after it, and on a computer Shift+Enter splits the caption where your cursor is. Backspace at the very start of a caption joins it to the one before.
• Move words between captions — drag the blue line between two cards. Drag it left and the earlier caption hands its last words to the next one; drag it right and it takes words back. Drag it past the last word and the two become one. When the last card holds more than one word, a blue line stands after it too: drag it left to split that card.
• Mark a word — press and hold a word (double-click on a computer) for Keyword, Emoji and Fix all. Keyword marks or unmarks it; Emoji picks the emoji that goes with it, or No emoji; Fix all gives every other place the same word was heard the spelling you fixed.
• Find and replace — Find, under Source at the start of the strip, widens that container into a search across the caption, with Replace with and Replace all.

Every card is the same width, and the caption inside is drawn as it shows on the picture, scaled to fit the card. Tapping a card picks that caption: the playhead jumps to its first frame, the picture shows it, and the card shows that caption's own settings without changing size — at its top-right corner the hand, Place this caption on its own, and in its bottom row its Start and End (typed like any number). Press the hand and it lights: the next move you make — dragging the caption on the picture, nudging it, turning it, widening it or typing a position — moves this one caption alone and gives it its own place; press the hand again before moving and nothing changes. Once the caption has its own place the hand reads Placed on its own and stays lit on the card whether or not it is picked, every other caption's place shows on the picture as a plain white box while it is picked, and the target at the card's top-left corner puts it back where every caption shares. The speaker names on the right light the format the picked caption wears; press another to give it that format (New speaker is under Source › Speakers). Pick anything else, or drag a blue line, and the settings step aside.

Speakers and sound labels in captions

Detect speakers, under Speakers in Source, tells the voices apart on your device and gives each speaker a format of their own, starting with a colour. Every caption belongs to one speaker — a new caption starts with Speaker 1 — and shows it only through that speaker's format; to hand a caption to another speaker, pick its card and press that speaker's name at the right; a new speaker is made under Speakers with New speaker. Each speaker has a name and a Side — Left, Center or Right — that their lines keep to, and with Show names on, the speaker's name leads the line whenever the speaker changes. A caption is split wherever the speaker changes. How each speaker's lines look is described under "Style captions" below.

Find sounds, under Sounds in Source, listens to the stretches between the words and suggests labels such as [Laughter], [Applause], [Music] or [Rain]. Each suggestion appears in the strip as a marked card with a tick to add it and a cross to dismiss it, and nothing is added until you say so. Each kind of label has its own switch, so you can turn off every music label at once.

Both work on a caption made from a sound, since they listen to it.

Style captions

Press a row in the container at the right — a speaker, named for the look they wear — to pick them and open a strip of looks in the words' place, its search at the strip's start — Bold, Karaoke, Headline, Sunset, Highlight, Neon and Minimal, the same looks a text item's Text tab offers, Headline and Sunset differing only in the colours they give the spoken word and the keywords — followed by the looks you have starred; the close under the speakers, or Escape, goes back to the words. Every look card shows the sample word Abc in that look rather than the caption's words. Picking one sets the size, weight, capitals, colours, outline, how the spoken word shows and the keyword colour, and leaves your font, alignment, bubble and position alone. The star under the speakers stars the picked speaker's look, and it then appears for captions, text items and tables alike.

Each speaker wears one whole look. The container lists every speaker — one to begin with — each row the name of their look written in it (Bold, Neon, Custom), the speaker's own name beside it once there are several, with nothing picked at first. Whatever you pick or set changes the picked speaker's look alone — their text, their spoken word, their keywords. Under a thin line sit the picked speaker's Remove (back to the first speaker's look, where theirs differs), the star, Recipe and a close. With no speaker picked, a star stands there alone and stars the caption whole. A new speaker starts with a copy of the first speaker's look in a colour of their own, and there is no shared look underneath: each speaker's look is whole.

Recipe opens the picked speaker's look, listed with every setting it holds on the Text look page; opening it takes Source and Find with it. Hover over a setting there, or hold a finger still on it, to read what it does.

The settings about the words themselves live in Source, the same for every speaker: Lines (Words / line, Characters / line, Lines on screen, the Minimum time a caption stays up and Join gaps under, with Re-split to rebuild the captions from these settings — it warns you first); Keywords, which words are marked (Mark automatically, and How many while it is on); Speakers; Sounds; and Translate.

Lines wrap inside the caption's width, up to Lines on screen, and a caption with more words than that room shrinks to fit, so it never runs off a tall video. Captions read in any language's writing — Arabic and Hebrew right to left, Chinese, Japanese, Korean, Hindi, Thai and more — and the letters download the first time a caption needs them.

A single caption has no look of its own: what sets a line apart is its speaker. Give it a speaker of its own — pick its card and press a speaker's name at the right — and that speaker's look is where its look lives.

Effects, animation and placement for captions

Effects and Border are the same menus a text item has, and they apply to the whole caption — every line it shows. A glow round the words is the Edge in Border.

Animate moves the whole caption with its Presets, Path and Automate, like anything else you place. It is also where the spoken word moves: the Text section holds four cards for captions alone — Word pop, Word bounce, Word grow and Word wipe — each moving every word while it is said, and named Pop, Bounce, Grow or Wipe in the Recipe and on the timeline. With one caption picked, Animate offers the presets that are an arrival and an exit, and its Recipe holds only Arrive and Leave, for that caption alone: each time it comes on screen it arrives and leaves around its own words.

Transform holds the same rows as a text item — flip, Perspective, Opacity, Position, Size and rotation — with one difference: Size is the caption's width, and its height follows its words, because every caption has a different number of lines. The height grows away from whichever edge of the frame the caption is nearer to. On the picture, drag the caption to move every line at once, turn it with the turn button, and widen it with the pills on its sides. It can go partly past the edge of the frame like anything else; its box is as tall as its tallest caption and as wide as its lines can run, so a box kept inside the frame keeps every word in view. A picked caption still moves with the whole caption until you press Place this caption on its own on its card; from then on dragging moves that caption alone — the status at the bottom of the display reads Moving this caption only — and the target on its card, Back to the shared place, returns it.

Translate captions

Translate, in the caption card's Recipe, puts the captions into another language for free, on your device. Pick a language under Translate to and press Translate. The first time you use a language, its translator downloads once (Downloading the translator…); then the captions are translated (Translating…). You get a second caption in that language, tied to the same sound and breaking where the original does, placed clear of the original so the two read one above the other. Only the languages listed can be translated; one that cannot be yet says This language can't be translated yet. If you change the original's words afterwards, the translation says The original changed, with Update to translate it again.

Caption files

Export in the File menu lists Video and Captions once a caption holds words. Pick which caption under Captions, when the video has more than one, the Format — SRT or WebVTT — and, for a caption with more than one speaker, whether to include Speaker names, then press Download. Each caption is written where it shows in the video, and the file is made on your device. A caption file carries the words and their timing only, never the look — to keep the look, export the video itself.

To bring a caption file in, press a caption's Source and choose Import a file (SRT, WebVTT or plain text). With a sound chosen, the words are matched to the voice automatically; without one, the file's own timing is kept.

Where you can use captions

• Video — Groups

What you can do with a caption you made

With no speaker picked, the star under the speakers on the Captions tab stars the caption whole, and a starred caption can be put into another video:

Star · Copy

Lines

• Words / line — The most words one caption line holds when the captions are split again.
• Characters / line — The most characters, spaces counted, one caption line holds when the captions are split again.
• Lines on screen — How many rows a caption line is drawn in: a line that would wrap to more is drawn smaller.
  • 1 · 2
• Minimum time — How long a caption line stays on screen at least, up to the next line.
• Join gaps under — A gap between two caption lines shorter than this is closed, so the captions never blink off.

Keywords

• Mark automatically — Whether the caption's keywords are marked automatically as its words change; a word you mark yourself always wins.
• How many — How many of the caption’s words automatic marking marks, as a share of them.

Speakers

• Speaker — Which of the caption's speakers says this line, and so which format it is drawn in.

Sounds

• Laughter — Whether this kind of sound is labelled in the caption.
• Applause — Whether this kind of sound is labelled in the caption.
• Cheering — Whether this kind of sound is labelled in the caption.
• Music — Whether this kind of sound is labelled in the caption.
• Singing — Whether this kind of sound is labelled in the caption.
• Crying — Whether this kind of sound is labelled in the caption.
• Screaming — Whether this kind of sound is labelled in the caption.
• Sigh — Whether this kind of sound is labelled in the caption.
• Cough — Whether this kind of sound is labelled in the caption.
• Footsteps — Whether this kind of sound is labelled in the caption.
• Knocking — Whether this kind of sound is labelled in the caption.
• Door — Whether this kind of sound is labelled in the caption.
• Phone ringing — Whether this kind of sound is labelled in the caption.
• Dog barking — Whether this kind of sound is labelled in the caption.
• Siren — Whether this kind of sound is labelled in the caption.
• Alarm — Whether this kind of sound is labelled in the caption.
• Gunshot — Whether this kind of sound is labelled in the caption.
• Explosion — Whether this kind of sound is labelled in the caption.
• Thunder — Whether this kind of sound is labelled in the caption.
• Rain — Whether this kind of sound is labelled in the caption.