Smart Editor AI

Smart Editor AI edits a video: the speech is transcribed word by word and shown as captions in the style you choose, with optional text overlays (Apple emojis included), images, B-rolls, background music, sound effects and a voice-over, then everything is exported as an MP4. In the API, a Smart Editor AI video is a project (/v1/captions, ids cap_…). The usual flow is to clean an ad with Smart Remover, then create a project from the cleaned video (job_id).

The exported video looks exactly like the preview of the editor: same fonts, sizes, line breaks, line spacing and boxes.

Price

0.5 credit per second of video (30 per minute), with a minimum of 5 credits. A 30 second video costs 15 credits.

• The credits are taken on the first export of a project (402 insufficient_credits if the balance is too low).
• A failed export gives its credits back.
• Later exports of the same project (after edits) are free, up to 20 per project.
• Videos up to 15 minutes.
• Included from the Growth plan, in the app and through the API (403 captions_not_allowed otherwise).

Lifecycle

Create a project

POST /v1/captions

Give exactly one of job_id, upload_id or video_url. The response (202) is the project, in transcribing status. Follow it with GET /v1/captions/{id} or the caption.succeeded and caption.failed webhooks.

curl https://pixamake.ai/v1/captions \
  -H "Authorization: Bearer $PIXAMAKE_API_KEY" \
  -H "Content-Type: application/json" \
  -H "Idempotency-Key: 6f1d0c2a-3b7e-4f7a-9d52-0f3c9a1e8b44" \
  -d '{
    "job_id": "job_3kT9xQ2vLm8RfZp1Wq7a",
    "style": {"base": "highlight", "highlight_color": "#FDE047", "words_per_line": 2},
    "overlays": [{"text": "-30 % today 🔥", "start": 0, "end": 3, "y": 18, "style": "Brand yellow"}]
  }'

Styles

Built-in styles

Built-in styles never change and cannot be edited: to change a setting, save your own style from one of them (base).

Your styles

Each workspace can save up to 50 named styles, in the editor or through the API, then use them wherever a style is expected, by id (cst_…) or by name.

curl https://pixamake.ai/v1/caption-styles \
  -H "Authorization: Bearer $PIXAMAKE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "Brand yellow", "style": {"base": "highlight", "highlight_color": "#FFC300", "font": "poppins"}}'

A project keeps a copy of the style it was given: changing a saved style does not change videos already set up. Send the style to the project again (PATCH /v1/captions/{id}) to use the new version. Names are unique in a workspace (409 name_taken otherwise).

Style object

Send a reference (built-in id, cst_… id or name), or an object with base (a reference, the project's current style by default) and only the fields to change. As soon as a field other than position differs, preset is custom.

The vertical position is a setting of each video: {"base": "tv", "position": 60} keeps the built-in tv style (and a saved style stays linked the same way), only placed higher.

Overlays

An overlay is a text shown at a given time and place (price, promo, hook). It takes the look of a style: font, colors, outline, banner and padding, shadow, case. Emojis are drawn as Apple emojis, in the app as in the exported video.

In responses, each overlay has style_id (the style it comes from, or custom) and style (the settings it uses).

Images

An image (product photo, logo, sticker) shown at given times, centered across the frame. Transparency is kept.

Up to 20 images per project. In responses, each image also has ratio (its height divided by its width).

Background music

A track from the Pixamake library plays under the sound of the video, looped if needed, lowered automatically while someone speaks, and faded out over the last 1.5 seconds. Music is included in the price.

"music": null removes the music. With auto, the track is picked as soon as the transcript is ready; responses show it in music.track (title, artist, category, mood, tags and a preview link).

"music": { "auto": true, "volume": 25 }

The library

GET /v1/music lists the tracks, the same for every workspace, newest first. Filters: category (hype, upbeat, chill, inspiring, emotional, dramatic, luxury, funny, corporate, lofi), q (text searched in the title, artist, genre, mood and tags) and limit (100 by default, 500 at most).

{
  "object": "music_track",
  "id": "mus_4fT8kQ2vLm9RzXp1Wq7a",
  "title": "Golden Hour",
  "artist": "Studio North",
  "duration_seconds": 142.6,
  "category": "upbeat",
  "mood": "warm, confident",
  "genre": "pop",
  "energy": 4,
  "bpm": 118,
  "tags": ["upbeat", "pop", "claps", "instrumental", "beauty", "fashion"],
  "description": "Bright pop groove with claps and a warm synth bass. Suits beauty and fashion launches.",
  "preview_url": "https://...",
  "is_new": true
}

B-rolls

Videos shown over the video during a time range, above it and below the captions, texts and images (their own sound is cut, the soundtrack goes on). By default a B-roll fills the frame in its own format, centred, on black. Upload the video with POST /v1/uploads first. Up to 20 per project.

In responses, each B-roll has name, length and a preview_url valid one hour.

Voice-over

A voice-over replaces or completes the video's voice, for instance when the video has none. When there is one, the captions are transcribed from it (again, in the background: the project goes back to transcribing).

"voiceover": null removes the voice-over; the captions are transcribed again from the video.

"voiceover": { "audio_url": "https://cdn.example.com/voice-over.mp3", "start": 0.5 },
"original_voice": false

Sound effects

Sound effects played once at a given time, over the soundtrack, chosen from the Pixamake library. Up to 50 per project, included in the price.

In responses, each sound has name and a preview_url valid three hours.

GET /v1/sound-effects lists the library, the same for every workspace, with the same filters as /v1/music: category (whoosh, pop, click, impact, riser, notification, cash, applause, laugh, magic, transition, ambience), q and limit.

"sounds": [{ "sound_effect_id": "mus_9pQ2xT8kLm3RzWv1Yc6b", "start": 0.8 }, { "sound_effect_id": "mus_2hT7kQ9vLm4RzXp8Wc1a", "start": 8.5, "volume": 60 }]

Safe zone

Texts and images never overlap each other or the spoken captions. Around each of them, a margin of 2 % of the frame height stays free while they are shown at the same time. When an element would touch another one, the export moves it to the nearest free height (texts first, in their order, then images); the editor shows the same result, and dragging an element stops at the edge of the others' margins. The y you send is kept as asked: only the export and the preview move the element.

Edit and export again

PATCH /v1/captions/{id} changes words (each { "text", "start", "end" }, plus "line": "start" to start a caption at this word or "line": "join" to keep it in the caption before), style, overlays, images, sounds, brolls (each list replaces the current one), music, voiceover or original_voice. Then POST /v1/captions/{id}/export burns the changes into a new file. With auto_export: false, call the same export endpoint when you are ready.

GET /v1/captions/{id}/srt returns the captions as an SRT file.

The project object

{
  "id": "cap_8LmQ2vT9xK3rWp7aZn4c",
  "object": "caption_project",
  "status": "done",
  "name": "spring-ad.mp4",
  "source_job_id": "job_3kT9xQ2vLm8RfZp1Wq7a",
  "input": { "duration_seconds": 28.4, "width": 1080, "height": 1920 },
  "language": "en",
  "credits": 29,
  "charged": true,
  "exports": 1,
  "words": [{ "text": "This", "start": 0.12, "end": 0.31 }],
  "style": { "preset": "custom", "font": "montserrat", "words_per_line": 2, "highlight": "color", "highlight_color": "#FDE047", "background": null },
  "overlays": [{ "id": "ov_x1", "text": "-30 % today 🔥", "start": 0, "end": 3, "y": 18, "size": 80, "style_id": "cst_8LmQ2vT9xK3rWp7aZn4c", "style": { "font": "poppins", "color": "#FFFFFF" } }],
  "music": { "auto": true, "track_id": "mus_4fT8kQ2vLm9RzXp1Wq7a", "volume": 30, "track": { "id": "mus_4fT8kQ2vLm9RzXp1Wq7a", "title": "Golden Hour", "artist": "Studio North", "duration_seconds": 142.6, "category": "upbeat", "mood": "warm, confident", "tags": ["upbeat", "pop"], "preview_url": "https://..." } },
  "sounds": [{ "id": "snd_x1", "sound_effect_id": "mus_9pQ2xT8kLm3RzWv1Yc6b", "name": "Pop", "start": 0.8, "duration": 0.4, "volume": 80, "preview_url": "https://..." }],
  "brolls": [],
  "voiceover": null,
  "original_voice": true,
  "images": [{ "id": "img_x1", "image_url": "https://...", "start": 4, "end": 7, "y": 30, "size": 40, "ratio": 1 }],
  "result": { "url": "https://...", "expires_at": "2026-10-06T10:00:00.000Z" },
  "srt_url": "https://pixamake.ai/v1/captions/cap_8LmQ2vT9xK3rWp7aZn4c/srt",
  "error": null,
  "created_at": "2026-10-06T08:58:12.000Z",
  "exported_at": "2026-10-06T09:00:41.000Z"
}

result.url expires after one hour: download it, or fetch the project again for a fresh link. error.code is one of transcription_failed, insufficient_credits, no_speech or export_failed.