---
title: "Short-form video: captions + logo + hook"
method: POST
path: "/v1/studio/short"
tags: ["Shorts"]
---

# Short-form video: captions + logo + hook

`POST /v1/studio/short`

## Request

- Base URL: `https://apihub.co.il`
- URL: `https://apihub.co.il/v1/studio/short`
- Auth: HTTP bearer

The Reels / TikTok / Shorts edit in one call: jump cuts (silences and fillers; with `remove_retakes` also the retakes, false starts and stutters — the best take of each stays), animated captions with full control of font, weight, size, colours, outline, box, position and animation, your logo in a top corner, and a hook headline in the first seconds. Also B-roll overlays (full screen, picture-in-picture or split screen), zoom / punch-in, colour grade, music and sound effects. 9:16, 1080p, best quality by default.

**Billing:** Per second of output (≤720p / 1080p tiers; audio and images cheaper) + transcription seconds when captions or silence removal transcribe automatically + the voice clean-up per second of output when asked for (`audio.denoise` / `enhance`, off by default) and only when it found noise to reduce or enhanced the voice (`auto` on a clean recording is free) + effect seconds for a heavy engine effect in `effects` (its seconds on screen × its measured weight — GET /v1/effects). Failed renders are not charged. · **Rate limit:** 30/min · **Asynchronous**

## Headers

- `Idempotency-Key` string
- `X-Tag` string
- `x-wait-ms` integer

## Request body

- object
  - `hook` object
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `text` string, required: The hook, e.g. “3 טעויות שכולם עושים בשיווק”
    - `start` number: seconds (default 0) — or anchor it with `at` / `during`
    - `style` union: Preset id (GET /v1/studio/hook-styles) or overrides. In the text, *word* = accent colour, ~word~ = struck through, \n = a smaller second line
      - 'box_red' | 'yellow' | 'black' | 'outline' | 'clean' | 'type' | 'punch' | 'glitch' | 'marker' | 'words' | 'box' | 'counter' | 'shake' | 'rise' | 'strike' | 'blur' | 'flash' | 'look-clean' | 'look-bold-lime' | 'look-night' | 'look-warm'
      - object
        - `x` number
        - `y` number
        - `font` string
        - `size` number
        - `color` string
        - `margin` number
        - `preset` 'box_red' | 'yellow' | 'black' | 'outline' | 'clean' | 'type' | 'punch' | 'glitch' | 'marker' | 'words' | 'box' | 'counter' | 'shake' | 'rise' | 'strike' | 'blur' | 'flash' | 'look-clean' | 'look-bold-lime' | 'look-night' | 'look-warm'
        - `rotate` number
        - `shadow` number
        - `weight` integer
        - `outline` number
        - `position` 'top' | 'center' | 'bottom'
        - `animation` 'pop' | 'slide' | 'fade' | 'typewriter' | 'none' | 'punch' | 'glitch' | 'marker' | 'words' | 'box' | 'counter' | 'shake' | 'rise' | 'strike' | 'blur' | 'flash'
        - `uppercase` boolean
        - `background` string, nullable
        - `side_margin` number
        - `accent_color` string
        - `shadow_color` string
        - `outline_color` string
        - `letter_spacing` number
        - `max_chars_per_line` integer
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `subtext` string: Optional smaller second line
    - `duration` number: seconds on screen (default 3.5) — not with `during`
  - `logo` object
    - `url` string, uri, required: Logo image — PNG with transparency works best
    - `size` number: Width, % of the video width
    - `margin` number: % of the short side
    - `opacity` number
    - `position` 'top-left' | 'top-right' | 'bottom-left' | 'bottom-right'
  - `zoom` object[]
    - `x` number: focus point, 0 = left … 1 = right
    - `y` number: focus point, 0 = top … 1 = bottom (faces are usually ~0.35)
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `to` number: zoom factor reached at the end (punch: immediately)
    - `end` number: seconds — the zoom holds `to` until here (with `at`: give `duration` instead)
    - `ease` 'linear' | 'smooth' | 'punch': smooth = slow push-in, punch = instant jump cut-in
    - `from` number: zoom factor at start (1 = full frame)
    - `start` number: seconds, on the edited timeline — or anchor it with `at` / `during`
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `duration` number: seconds — with `at`
  - `audio` object: Voice clean-up, opt in: `denoise` (default off; `auto` reduces background noise only when there is some) and `enhance`. Reported in the render's `voice`
    - `denoise` 'off' | 'auto' | 'light' | 'strong': Background-noise reduction (DeepFilterNet 3) — default `off` on edit / short / remove-silence, `auto` on clean-audio: `auto` measures the noise and reduces it only when there is some (a clean recording, a music bed or digital silence stays as it is) · `light` a gentle fixed reduction · `strong` as much as the model removes · `off` the sound as recorded. Mains hum (50/60 Hz) is notched out whenever it is found
    - `enhance` boolean: Gentle voice enhancement after the noise: a high-pass at 75 Hz, a de-esser, a 3:1 compressor and a small presence lift — each only when the measurement calls for it (a clean studio voice stays as it is)
  - `color` object
    - `gamma` number: 1 = unchanged; >1 lifts the shadows
    - `preset` 'none' | 'warm' | 'cool' | 'vivid' | 'cinematic' | 'teal-orange' | 'bw' | 'vintage' | 'matte': A ready grade; the fields below fine-tune it
    - `sharpen` number
    - `contrast` number: 1 = unchanged
    - `vignette` number: Darkened corners, 0…1
    - `brightness` number: −1…1 (0 = unchanged; 0.05 is a subtle lift)
    - `saturation` number: 1 = unchanged, 0 = black & white
    - `temperature` number: −1 cool (blue) … +1 warm (orange)
  - `fonts` object[]: Your brand fonts: up to 4 fonts from your brand kit, each { family, url } (the url as the brand kit gives it — your own files only). A caption, hook or text style may then name `family` as its `font`. Text a font cannot draw (Hebrew words in a Latin-only font) is set in the nearest built-in font, and the render's `warnings` say so
    - `url` string, uri, required
    - `latin` boolean
    - `arabic` boolean
    - `family` string, required
    - `hebrew` boolean
    - `category` 'sans' | 'serif' | 'display' | 'handwriting' | 'rounded'
    - `cyrillic` boolean
  - `music` object
    - `url` string, uri, required: Music file (mp3, wav, m4a)
    - `duck` boolean: Lower the music automatically while someone speaks
    - `loop` boolean
    - `volume` number
    - `duck_db` number: With `duck`: how far the music dips under speech, dB — −6 (default) = half as loud under the voice, −3 a gentle dip, −18 a deep one. A fixed depth whatever the speaker's level; attack 40 ms, release 700 ms
    - `start_ms` number: Start the music this far into the video
    - `normalize` boolean: Bring the track to a standard loudness first, so `volume` means the same for every file (a quiet file is not lost under the speech)
    - `fade_in_ms` number
    - `fade_out_ms` number
  - `texts` object[]
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `exit` 'none' | 'fade' | 'smear' | 'blur'
    - `text` string, required: *word* = accent colour · ~word~ = struck through · \n = a smaller second line
    - `start` number: seconds, on the edited timeline — or anchor the element to a word with `at` / `during`
    - `style` object: Overrides on top of the preset
      - `x` number
      - `y` number
      - `font` string
      - `size` number
      - `align` 'center' | 'start' | 'end'
      - `color` string
      - `hollow` boolean
      - `italic` boolean
      - `rotate` number
      - `shadow` number
      - `weight` integer
      - `outline` number
      - `position` 'top' | 'center' | 'bottom'
      - `uppercase` boolean
      - `background` string, nullable
      - `side_margin` number
      - `accent_color` string
      - `shadow_color` string
      - `outline_color` string
      - `letter_spacing` number
      - `max_chars_per_line` integer
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `effect` 'none' | 'fade' | 'pop' | 'blur_in' | 'slide_up' | 'typewriter' | 'highlight' | 'punch' | 'glitch' | 'marker' | 'words' | 'box' | 'counter' | 'shake' | 'rise' | 'strike' | 'flash'
    - `preset` 'hero' | 'mega' | 'accent' | 'pill' | 'pill-accent' | 'marker' | 'outline' | 'small' | 'script' | 'type': GET /v1/studio/text-styles
    - `duration` number: seconds (default 1.6) — not with `during`
  - `words` object[]: Your own timed words (ms) instead of a transcript
    - `id` string: Stable id (w0001…), as GET /v1/speech/transcripts/{id} returns it — anchors and cues name words by it
    - `end` number, required: ms
    - `keep` boolean: Never drop this word from the captions: a cut that leaves only a sliver of it shows it at the join (a word of your own script)
    - `text` string, required
    - `start` number, required: ms
    - `spoken` string: What was said, when the displayed text differs ("3,000" vs "three thousand")
    - `speaker` string, nullable
  - `output` object
    - `fit` 'crop' | 'blur' | 'pad': When the aspect ratio changes: crop the edges, blurred copy behind, or solid bars
    - `fps` number
    - `format` 'mp4' | 'webm' | 'mov' | 'gif' | 'mp3' | 'wav' | 'm4a' | 'jpg' | 'png'
    - `quality` 'draft' | 'standard' | 'high': high (default) = best picture; draft renders fastest for previews
    - `background` string: Bar colour for fit=pad
    - `resolution` '480' | '720' | '1080' | 'source': Short side in pixels (1080 = Full HD). `source` keeps the input size, max 1080
    - `aspect_ratio` 'original' | '9:16' | '16:9' | '1:1' | '4:5' | '4:3': 9:16 for Reels/TikTok/Shorts, 1:1 or 4:5 for feed, 16:9 for YouTube
  - `panels` object[]
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `rect` object: Part of the frame (% ) — omit for the whole frame
      - `x` number, required
      - `y` number, required
      - `width` number, required
      - `height` number, required
    - `color` string: A white or black card turns the frame into a typography scene
    - `start` number: seconds, on the edited timeline — or anchor it with `at` / `during`
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `fade_ms` number
    - `opacity` number
    - `duration` number: seconds (0.1–0.2 = a flash; not with `during`)
  - `sounds` object[]
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `url` string, uri, required: Sound effect (e.g. from /v1/ai/sound-effects)
    - `start` number: seconds, on the edited timeline — or anchor it with `at`
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `volume` number: 1 sits just under the speech; 0.5 is a soft accent
    - `normalize` boolean: Level the effect first, so `volume` means the same for every file
  - `effects` object[]: Effects from the library — GET /v1/effects: [{ id, params, start + duration | at | during }]
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string, required: The effect's id — GET /v1/effects
    - `ref` string: Your name for it; the render report uses it
    - `start` number: seconds on the edited timeline — or anchor it to words with `at` / `during`
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `params` object: The effect's own params (GET /v1/effects/{id}); every field has a default
    - `duration` number: seconds (not with `during`)
  - `inserts` object[]: Clips placed INTO the video in the same render, each from its own in-point and with its sound — see POST /v1/studio/edit `inserts`
    - `x` number: pip: exact placement, % of the width (the box's centre)
    - `y` number: pip: exact placement, % of the height
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `fit` 'crop' | 'blur' | 'pad': cut / overlay: how a clip of another shape fills the frame — crop the edges (default), a blurred copy behind it, or bars
    - `url` string, uri, required: The clip (mp4, mov, webm…): b-roll, a stock clip, an excerpt of a video you are allowed to use (your own, CC BY with its credit line on screen as a `texts[]` element, licensed) — e.g. the url of POST /v1/youtube/videos/{id}/file. Its licence is the record its own call kept (the file call, an upload's `license`, a stock save) — GET /v1/usage?tag=<job> lists the job's sources
    - `from` number: Where inside the clip to start, seconds — the in-point of an excerpt
    - `mode` 'cut' | 'overlay' | 'pip': cut = the clip replaces the picture for its span (a cut-away; with `audio: "keep"` it replaces the sound too) · overlay = full frame over the video, dissolving in and out, the video's sound carries on · pip = in a box (`position` / `x`,`y`, `size`), the video's sound carries on
    - `size` number: pip: the box's width, % of the video's width (the clip's shape is kept)
    - `audio` 'keep' | 'duck' | 'mute': The clip's OWN sound. keep = heard at `volume` — in `cut` mode instead of the video's sound (the captions of the words it covers are left out: nobody hears them), otherwise mixed over it · duck = heard under the speech, dipping `duck_db` whenever the speaker talks · mute = silent (b-roll; the default)
    - `start` number: seconds, on the edited (output) timeline — or anchor it with `at` (a word) / `during` (a stretch of words)
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `volume` number: The clip's sound level (1 = as recorded)
    - `duck_db` number: audio duck: how far the clip's sound dips while the speaker talks, dB
    - `duration` number: Seconds on screen. Default: the rest of the clip from `from` (a `during` span sets it instead). Never runs past the clip or the video — a longer ask is shortened and reported in `warnings`
    - `position` 'top-left' | 'top-right' | 'bottom-left' | 'bottom-right' | 'center': pip: the corner
    - `transition` object: Default: a hard cut for `cut`, a 240 ms crossfade for `overlay` and `pip`
      - `ms` number: Length of the transition, ms (at most a third of the clip on screen)
      - `type` 'cut' | 'crossfade' | 'dip', required: cut = a hard cut in and out · crossfade = the clip dissolves in over its first `ms` and out over its last (picture and sound) · dip = through black: the video fades down over the `ms / 2` before the clip, the clip up over its first `ms / 2` (and the same on the way out)
  - `brand_id` string: A brand of yours (GET /v1/brands): its logo (at its corner), colours, fonts and caption style fill what this request leaves out — anything you set here wins. `default` = your default brand
  - `captions` union: Caption style (preset id or full style object), or false for no captions
    - false
    - object
      - `style` union: A preset id (GET /v1/studio/caption-styles) or a style object; `preset` + fields overrides a preset. A retired preset still works: it renders as the default with a warning
        - 'classic' | 'hormozi' | 'tiktok' | 'karaoke' | 'one-word' | 'marker' | 'news' | 'neon' | 'quiet' | 'pop' | 'typewriter' | 'bounce' | 'underline' | 'elegant' | 'handwritten' | 'comic' | 'sticker' | 'drama' | 'cinema' | 'shadow' | 'lime' | 'retro' | 'scale' | 'bubble' | 'sport' | 'minimal' | 'slide' | 'volt' | 'neon-glow' | 'marker-tape' | 'keyword-pop' | 'keyword-box' | 'editorial' | 'poster' | 'broadcast' | 'soft' | 'cards' | 'cards-bold' | 'cards-neon' | 'kinetic-pop' | 'fire' | 'gold' | 'glass' | 'vhs' | 'candy' | 'look-clean' | 'look-bold-lime' | 'look-night' | 'look-warm'
        - object
          - `x` number
          - `y` number
          - `font` string
          - `glow` number
          - `mode` 'line' | 'cards'
          - `size` number
          - `skew` number
          - `color` string
          - `fonts` string[]
          - `anchor` 'top' | 'center' | 'bottom'
          - `hollow` boolean
          - `italic` boolean
          - `margin` number
          - `preset` 'classic' | 'hormozi' | 'tiktok' | 'karaoke' | 'one-word' | 'marker' | 'news' | 'neon' | 'quiet' | 'pop' | 'typewriter' | 'bounce' | 'underline' | 'elegant' | 'handwritten' | 'comic' | 'sticker' | 'drama' | 'cinema' | 'shadow' | 'lime' | 'retro' | 'scale' | 'bubble' | 'sport' | 'minimal' | 'slide' | 'volt' | 'neon-glow' | 'marker-tape' | 'keyword-pop' | 'keyword-box' | 'editorial' | 'poster' | 'broadcast' | 'soft' | 'cards' | 'cards-bold' | 'cards-neon' | 'kinetic-pop' | 'fire' | 'gold' | 'glass' | 'vhs' | 'candy' | 'look-clean' | 'look-bold-lime' | 'look-night' | 'look-warm'
          - `shadow` number
          - `weight` integer
          - `outline` number
          - `palette` object[]
            - `color` string, required: Text colour of the card
            - `background` string, nullable: Box behind the whole card (null = none)
          - `grouping` integer
          - `keywords` string[]
          - `position` 'bottom' | 'center' | 'top'
          - `animation` 'none' | 'fade' | 'highlight' | 'karaoke' | 'pop' | 'bounce' | 'typewriter' | 'slide' | 'word'
          - `max_lines` integer
          - `max_words` integer
          - `uppercase` boolean
          - `background` string, nullable
          - `past_color` string, nullable
          - `side_margin` number
          - `word_radius` number
          - `active_color` string
          - `active_scale` number
          - `shadow_color` string
          - `active_rotate` number
          - `keyword_color` string
          - `keyword_scale` number
          - `outline_color` string
          - `future_opacity` number
          - `letter_spacing` number
          - `underline_color` string
          - `word_background` string, nullable
          - `active_underline` number
          - `active_background` string, nullable
          - `background_radius` number
          - `emphasize_numbers` boolean
          - `keyword_background` string, nullable
          - `max_chars_per_line` integer
          - `keyword_active_color` string
      - `lead_ms` number: The highlight of a word switches this many ms before the word's frame (default: one output frame)
      - `captions` object[]: Edited captions (ms). Get them from GET /v1/speech/transcripts/{id}/captions, fix the text, send them back
        - `end` number, required: ms
        - `text` string, required
        - `start` number, required: ms
        - `words` object[]
          - `id` string: Stable id (w0001…), as GET /v1/speech/transcripts/{id} returns it — anchors and cues name words by it
          - `end` number, required: ms
          - `keep` boolean: Never drop this word from the captions: a cut that leaves only a sliver of it shows it at the join (a word of your own script)
          - `text` string, required
          - `start` number, required: ms
          - `spoken` string: What was said, when the displayed text differs ("3,000" vs "three thousand")
          - `speaker` string, nullable
      - `language` string: Write the captions in this language (two letters, e.g. `en`, `he`, `ru`) when the video is spoken in another: the speech is translated sentence by sentence, each line keeps its sentence's time span (one small text-model call on your account, billed at the POST /v1/ai/text rate). Omit = the spoken language. Ignored when you send `captions` of your own or the spoken language is the same. Word-by-word karaoke becomes an even sweep over each translated line
      - `hide_words` string[]: Word ids never shown in the captions (a number a counter already shows)
      - `placements` object[]: Where the captions sit for a stretch: y (% of the height), scale, colour, a chip behind each line, or hidden — e.g. under a presenter's circle on a plate, and in the lower third (clear of the face) elsewhere. A caption never straddles two placements
        - `y` number: Centre of the captions, % of the height
        - `at` object: Start at a word (instead of a time)
          - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
          - `word` string, required: The word's id (w0001…)
          - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
        - `id` string: Your name for the element; the render report uses it
        - `chip` string: Cards: each line on a box of this colour (a label chip)
        - `hide` boolean: No captions at all in this stretch
        - `color` string: Every word in this colour (cards: the palette's rotation off; a dark ink drops the outline — for a white background)
        - `scale` number: Their size × this
        - `start` number: seconds, on the edited timeline — or anchor it with `at` / `during`
        - `during` object: Start and end with a stretch of words (instead of a time and a duration)
          - `to` string, required: Id of the last word
          - `from` string, required: Id of the first word
          - `pad_end_ms` number: End this much later (default 0)
          - `pad_start_ms` number: Start this much earlier (default 0)
        - `duration` number: seconds (not with `during`)
  - `keyterms` string[]: Names and terms to recognise correctly (brands, people)
  - `overlays` object[]: B-roll images / clips over the video
    - `x` number: Exact placement, % of the width (centre of the overlay) — pip only
    - `y` number: Exact placement, % of the height — pip only
    - `at` object: Start at a word (instead of a time)
      - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
      - `word` string, required: The word's id (w0001…)
      - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
    - `id` string: Your name for the element; the render report uses it
    - `url` string, uri, required: Image or video clip (e.g. from /v1/ai/image or /v1/ai/video)
    - `mode` 'full' | 'pip' | 'split': full = cut-away over the whole frame, pip = small window, split = split screen (visual on one part, the speaker on the other)
    - `size` number: pip width, % of the video width
    - `type` 'image' | 'video': Detected from the file name when omitted
    - `float` number: Gentle up/down drift, % of the height
    - `start` number: seconds, on the edited (output) timeline — or anchor it with `at` / `during`
    - `during` object: Start and end with a stretch of words (instead of a time and a duration)
      - `to` string, required: Id of the last word
      - `from` string, required: Id of the first word
      - `pad_end_ms` number: End this much later (default 0)
      - `pad_start_ms` number: Start this much earlier (default 0)
    - `rotate` number: Tilt in degrees (stacked-card look) — pip only
    - `shadow` boolean: Drop shadow under a pip / card
    - `fade_ms` number
    - `duration` number: seconds (not with `during`)
    - `position` 'top-left' | 'top-right' | 'bottom-left' | 'bottom-right' | 'center'
    - `split_side` 'top' | 'bottom': split: where the visual goes
    - `split_ratio` number: split: share of the height the visual takes
    - `speaker_focus` number: split: which band of the speaker stays visible (0 top … 1 bottom; faces ≈ 0.35)
    - `float_period_ms` number
  - `cut_sound` object: A sound effect on every jump cut (a whoosh); reported in `elements` as cut1…
    - `max` integer: At most this many cut sounds — the first cuts get them
    - `url` string, uri, required: The sound effect (e.g. a whoosh from /v1/ai/sound-effects)
    - `volume` number: 1 sits just under the speech; 0.5 is a soft accent
    - `normalize` boolean: Level the effect first, so `volume` means the same for every file
    - `offset_ms` number: Shift from the cut, ms (negative = the sound starts before the join, so a whoosh peaks on it)
  - `video_url` string, uri, required: Public URL of the source video (mp4, mov, webm, mkv…) — or the `url` of a previous render
  - `voiceover` union
    - object
      - `at` object: Start at a word (instead of a time)
        - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
        - `word` string, required: The word's id (w0001…)
        - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
      - `id` string: Your name for the element; the render report uses it
      - `url` string, uri, required: Narration audio (e.g. from /v1/ai/voice)
      - `start` number: seconds (default 0) — or anchor it with `at`
      - `during` object: Start and end with a stretch of words (instead of a time and a duration)
        - `to` string, required: Id of the last word
        - `from` string, required: Id of the first word
        - `pad_end_ms` number: End this much later (default 0)
        - `pad_start_ms` number: Start this much earlier (default 0)
      - `volume` number
      - `duck_original` boolean: Lower the original sound while the narration plays
    - object[]
      - `at` object: Start at a word (instead of a time)
        - `edge` 'start' | 'end', required: The moment the word starts, or the moment it ends
        - `word` string, required: The word's id (w0001…)
        - `offset_ms` number: Shift in ms, applied after the cuts (default 0)
      - `id` string: Your name for the element; the render report uses it
      - `url` string, uri, required: The line's audio
      - `start` number: seconds, on the edited timeline — when the original line starts; or anchor it to the line's first word with `at`
      - `during` object: Start and end with a stretch of words (instead of a time and a duration)
        - `to` string, required: Id of the last word
        - `from` string, required: Id of the first word
        - `pad_end_ms` number: End this much later (default 0)
        - `pad_start_ms` number: Start this much earlier (default 0)
      - `volume` number
  - `transition` object: How the kept spans are joined (a crossfade or a dip to black at every jump cut). Omit for plain cuts
    - `ms` number: Length of the transition, ms — rounded to whole frames, never more than half of the shorter span; every join shortens the output by this much
    - `type` 'cut' | 'crossfade' | 'dip': cut = a plain jump cut (default); crossfade = the spans dissolve into each other, picture and sound; dip = the picture dips to black between them
  - `language_code` string: Spoken language for automatic transcription, e.g. `he`, `en`. Omit to detect
  - `speech_models` string[]
  - `transcript_id` string: Reuse a transcript from /v1/speech/transcripts (no transcription charge). Omit to transcribe automatically
  - `remove_retakes` union: Cut retakes, false starts and stutters found in the words, keeping the best take of each: `true`, or { mode, judge, window_s }
    - boolean
    - object
      - `mode` 'auto' | 'last' | 'best': auto (default) = the last complete take, unless an earlier take of the same words is clearly cleaner · last = always the last complete take (standard practice) · best = the take with the highest score (complete, fluent — fewest fillers, stumbles and hesitations —, confident, steady pace)
      - `judge` boolean: When two complete takes score too close to call, a model watches just those takes and chooses — a separate POST /v1/ai/text call on the cheapest model, charged as its own line in your history (none when no group is ambiguous)
      - `window_s` number: How long after a take its retake is looked for, seconds (default 30). Several recordings of one script joined with POST /v1/studio/merge: their length
  - `remove_silence` union: Jump cuts on by default; false to keep every pause
    - false
    - object
      - `fillers` string[]
      - `padding_ms` number: Breathing room kept around speech
      - `min_pause_ms` number: Silences longer than this are cut (detected in the audio)
      - `threshold_db` number: Silence level in dB (default: adapted to the video’s loudness)
      - `remove_fillers` boolean: Also cut um / uh / אה / אממ…
  - `normalize_audio` boolean

## Response `200`

Completed (or 202 while processing)

- object
  - `id` string: Execution id — stable across internal retries; use it with GET /v1/executions/{id}
  - `status` 'pending' | 'processing' | 'completed' | 'partial' | 'failed' | 'cancelled' | 'timeout'
  - `operation` string
  - `tag` string, nullable: Your X-Tag, if sent
  - `created_at` string, date-time
  - `completed_at` string, date-time, nullable
  - `output` object: Operation result (see each operation). While a Studio render is processing: `stage` (queued · transcribing · downloading · preparing · cleaning_voice · mixing_audio · rendering · finishing), `phase` (the engine's: download · prepare · voice · audio · encode · qa · upload, or null), `progress` (a percent, 0–100) and `progress_fraction` (0–1), `eta_seconds` (seconds left, never below 0) and `estimate_seconds`
  - `error` string, nullable
  - `usage` object
    - `credits_charged` number, nullable
    - `credits_reserved` number
    - `estimated_credits` number
    - `cost_source` 'provider_reported' | 'calculated' | 'estimated' | 'null', nullable
    - `provider_usage` object, nullable: Normalised consumption, e.g. { VIDEO_SECOND: 22.4 }
  - `balance` object
    - `available` number
    - `reserved` number

## Other responses

- `202`: Processing — poll GET /v1/executions/{id}
- `400`: Invalid input
- `402`: Insufficient credits / project budget
- `404`: Resource not found (or not yours)
- `429`: Rate limited
- `503`: Service temporarily paused

---

[API](https://skmtc.dev/apihub/apis/apihub-one-api-for-chat-image-video-ai.md) · [All operations](https://skmtc.dev/apihub/apis/apihub-one-api-for-chat-image-video-ai/llms.txt) · [OpenAPI document](https://skmtc.dev/apihub/apis/apihub-one-api-for-chat-image-video-ai/revisions/a8caa42835ce?raw)