EN
Download

Glossary

Short-form video editing and AI glossary

The terms you will run into when editing Reels, TikToks and Shorts, or using an AI video editor, explained simply and with real numbers.

AI voiceover
An AI voiceover is narration produced by text-to-speech from a written script. It is used for tutorials, product videos, faceless content and translated versions. Narration typically runs at 130 to 160 words per minute, so a 30-second voiceover needs roughly 70 to 80 words. AI voiceover
Aspect ratio 9:16
The 9:16 aspect ratio is the full-screen vertical phone format: 9 units wide by 16 tall, or 1080×1920 pixels in Full HD. It is the format of TikTok, Instagram Reels, YouTube Shorts and Stories, as opposed to horizontal 16:9 (1920×1080) used by TV and regular YouTube. Aspect ratio 9:16
Auto reframing
Auto reframing adapts a video to a different aspect ratio, usually 16:9 to 9:16, by detecting the face or person speaking and moving a virtual camera to keep them in frame. Good reframing smooths its movements so the frame does not jitter with every gesture. Auto reframing
B-roll
B-roll is supplementary footage cut over the main shot, which is often a person talking to camera: product close-ups, hands at work, locations, screen recordings. It shows what the voice describes, hides cuts and keeps attention up. In short-form video, a B-roll shot usually lasts 1 to 3 seconds.
Background removal
Background removal separates a person or object from its surroundings so the background can be replaced with a color, a blur, an image or another video. Once reserved for green screens, it is now done by AI on ordinary footage, though edges can be rough around hair and fast motion.
Bitrate
Bitrate is the amount of data used per second of video, measured in megabits per second (Mbps). Higher bitrate means more detail and bigger files. For example, YouTube recommends about 8 Mbps for 1080p uploads at 24 to 30 fps and 12 Mbps at 48 to 60 fps.
Burned-in captions
Burned-in captions, also called open captions, are part of the video image and cannot be turned off. They allow custom styles and animation, look the same everywhere and are standard on TikTok and Reels, where many people watch without sound. They are the opposite of a separate SRT file. Burned-in captions
Codec
A codec is the method used to compress and decompress video. The most common are H.264 (AVC), which plays almost everywhere, and H.265 (HEVC), used by iPhones, which produces files up to about half the size at similar quality. A codec is not the same as the container, the file extension such as .mp4 or .mov.
Diarization
Diarization is the part of automatic transcription that works out who is speaking and when. It assigns each sentence to a speaker (Speaker 1, Speaker 2) in an interview or podcast. It is used to color captions by person, reframe on the active speaker or quickly find a guest's answers.
Dubbing
Dubbing replaces a video's voice track with speech in another language. With AI, it can be produced from a translated transcript and a synthetic voice, sometimes cloned from the original speaker. It differs from translated captions, which keep the original voice and show the text on screen. Dubbing
Ducking
Ducking automatically lowers music volume while someone speaks and brings it back up in the gaps. It keeps speech intelligible without cutting the music. In practice, music drops roughly 15 to 20 dB below the voice, with a smooth fade lasting a few tenths of a second.
False start
A false start is a sentence that is begun, abandoned and restarted ("So today I... Today I'm going to show you"). Automatic editing tools detect false starts from the word-level transcript and keep only the last complete take, which makes simply restarting a fluffed line the fastest way to shoot. False start
Filler words
Filler words are sounds and words that add no meaning: "um", "uh", "like", "you know", "basically" used as padding. Automatic editors find them in the transcript and cut them. It is worth reviewing the result, since words such as "like" are sometimes part of a real sentence. Filler words
Frame rate (fps)
Frame rate, measured in frames per second (fps), is how many images are shown each second. 24 fps gives a film look, 25 fps is the European broadcast standard, 30 fps is common on phones and 60 fps gives very smooth motion for sports or gaming. Keep the same frame rate from shoot to export.
Hook
The hook is the opening of a short video, its first 1 to 3 seconds, designed to stop viewers from scrolling. It promises a benefit, asks a question or shows the end result. A strong hook fits in one sentence of 6 to 10 words, reinforced by on-screen text. Hook
Jump cut
A jump cut is a cut between two moments of the same shot that makes the subject appear to jump slightly on screen. It happens when you remove a pause or stumble from a talking-head video. Now a style in its own right on social media, it is often softened with a small zoom on every other cut. Jump cut
Karaoke captions
Karaoke captions highlight each word at the exact moment it is spoken, like song lyrics on a karaoke screen. They require a word-level timestamped transcript. The style helps viewers follow along with the sound off and makes the rhythm of speech visible, which suits talking-head videos. Karaoke captions
Keyframe
A keyframe is a point on the timeline that sets the value of a property at a given moment: position, scale, opacity or volume. The editor calculates the change between two keyframes automatically. Keyframes are used to animate a slow zoom, move a title or fade a logo in.
LUFS
LUFS (Loudness Units relative to Full Scale) measures perceived loudness over the whole duration of a sound, not just its peaks. Platforms adjust playback volume to even things out; aiming for around -14 LUFS is common practice for online video and keeps a clip from sounding too quiet or squashed.
LUT
A LUT (Look-Up Table) is a file that maps a video's colors to new values. It converts flat log footage into normal, contrasty images, or applies the same color look to every video a brand publishes. A LUT is applied in one click, then dialed up or down in intensity.
MCP
MCP (Model Context Protocol) is an open protocol that lets an AI assistant such as Claude use external tools and apps. In video editing, an MCP server lets you give plain-language instructions, for example "remove the silences and export in 9:16", which the assistant carries out in the editing software. MCP
Punch-in
A punch-in is an instant digital zoom, usually 5 to 15%, applied at a cut. It suggests a second camera angle and hides the jump in a jump cut. A stronger punch-in of 20 to 40% emphasizes a key line and works best once or twice a minute at most. Punch-in
Raw footage
Raw footage is video straight out of the camera or phone, before any editing. A finished video is assembled from one or more raw clips. Logging means watching them to mark the good parts; AI tools speed this step up by working from the transcript.
Reel
A Reel is a short vertical video posted on Instagram or Facebook, with optional music, effects and captions. Reels appear in a dedicated feed and in the main feed, including for people who do not follow the account. The maximum length has changed several times, so check it in the app before posting. Reel
Safe zone
The safe zone is the part of the frame that stays visible and readable once the app interface is drawn on top. In 9:16, a cautious rule is to keep roughly the top 14%, the bottom 35% and 6% on each side clear, about 270, 670 and 65 pixels on a 1080×1920 frame. Safe zone
Short (YouTube Short)
A YouTube Short is a vertical or square video shown in YouTube's Shorts feed. According to YouTube's official help, square or vertical videos uploaded since 8 December 2025 with a length of up to three minutes are classified as Shorts. Short (YouTube Short)
SRT file
An SRT (SubRip) file is a plain-text file holding a video's captions, numbered and timestamped in hours:minutes:seconds,milliseconds format. It is uploaded separately to YouTube or a player, and viewers can switch it on or off. Unlike burned-in captions, it does not support animation or custom styling. SRT file
Timeline
The timeline is the area of an editing app where clips, audio, text and effects are lined up in time order across one or more tracks. In an automatic editor, it lets you see every cut the AI made and change or undo it.
Transcription
Transcription converts the speech in a video into written text. Editing tools use word-level timestamped transcripts, where every word is tied to its exact moment. This is the basis for captions, silence and filler word removal, translation and text-based editing. Transcription
UGC
UGC (user-generated content) is content made by users or customers rather than by a brand: filmed reviews, unboxings, phone demos. By extension, a UGC video is an ad in that spontaneous style, shot to camera by a paid creator. UGC
Voice cloning
Voice cloning uses AI to create a synthetic voice that imitates a real person from a recorded sample. It should only be done with that person's consent. In the European Union, the AI Act has required since 2 August 2026 that content realistically imitating an existing person be disclosed. Voice cloning
Watermark
A watermark is a logo or text permanently burned into an exported video, usually in a corner. Editing apps add one to free-plan exports and remove it on paid plans. Another app's watermark on a brand's video can look unprofessional and distract from the message.