Audio & video · checked 2026-10-19

Can I vibecode VEED?

KINDA· weekend project

VEED is two products on one bill. The first is a browser editor for captioned talking-head clips: transcribe, cut the ums and long silences, burn in subtitles, crop for vertical, even out the sound; Whisper and ffmpeg do all of that on your own machine, and an agent builds a working version in a weekend. The second is generative: video from text, avatars, AI voices, dubbing into 50+ languages, and stock footage, which run on models and licensed libraries you can't copy. A full multi-track editor in a browser is a multi-day job, so the prompt stops at cutting by transcript.

$25/moprice per seat

$300you'd save a year

weekendbuild time

—people replaced it

Before you start

What you need on hand.

DifficultyIntermediate · Some system setup or keys.

TimeA weekend; the transcript editor page is the slow part

Claude Code, Codex or CursorAny of them can run the prompt.
Python 3.12The prompt uses FastAPI and faster-whisper.
ffmpeg and ffprobeInstall through your package manager and check that burning subtitles works.
A machine that can run WhisperA CPU is enough for short clips with a small model; a GPU makes long videos practical.
Footage you already ownThe tool cuts and captions; it doesn't create video.

Before you buildVEED Free exports with a watermark at 720p, so it is fine for trying the workflow. If you only need subtitles, auto-editor, Kdenlive and a Whisper model already do it at no cost.

1. Install one agent: Claude Code, Codex, Cursor.
2. Make an empty folder and open it in the agent.
3. Paste the prompt below, press enter, and answer its questions as it builds.

The prompt

Build me a local subtitle and clean-up tool for talking-head videos to replace VEED.
Requirements:

- Python 3.12 + FastAPI + Jinja2: `uvicorn app.main:app --port 4890` serves
  http://localhost:4890. ffmpeg and ffprobe must be on PATH; check at startup and say what is
  missing. Each upload gets a folder ./work/<job-id>/.
- Transcription: ffmpeg extracts 16 kHz mono audio, faster-whisper (model name in .env,
  default "small") returns segments with word timestamps, saved as words.json.
- Editor page: the transcript as clickable words beside an HTML5 video player. Deleting
  words, a filler-word list (um, uh, you know) or every silence longer than 0.7 s (ffmpeg
  silencedetect) builds an edit list of kept time ranges; clicking a word seeks the video.
- Subtitles: group words into lines of at most 42 characters and 2 lines; presets for font,
  size, position and a one-word-highlight style, written as an ASS file; export SRT too.
- Render: ffmpeg cuts and joins the kept ranges, burns the ASS subtitles, normalizes to -16
  LUFS with loudnorm, and encodes H.264 + AAC (CRF 20) as MP4. Output sizes 16:9, 1:1 and
  9:16 by centered crop.
- Storage: SQLite (sqlite3 module) in ./data/jobs.db with a status per job, progress while
  rendering, and a delete button that removes the job folder.
- Local-first: no accounts, no telemetry, nothing uploaded; the only download is the Whisper
  model from Hugging Face on first use.
- Out of scope: multi-track timeline, stock media, AI avatars and voices, translation or
  dubbing, noise removal beyond loudnorm, face tracking, collaboration.
- README: installing ffmpeg, the speed on my machine for a 5-minute clip (measure it and write
  it down), and which Whisper model to pick for my language.

Opens your agent with the prompt filled in, you press enter. Or copy it and paste it into an empty folder.

Share on X

Max MiskowSponsoredAI agents, automations and MVPs, built for your company.Visit site ›

Then keep going

Next prompts, same folder.

The first prompt builds the core. Run these one at a time in the same folder to grow it into your own VEED.

Add subtitle translation to the video tool in this folder. Requirements:

- A target-language field on the editor page and a command `translate <job> <lang>`.
- Translate each subtitle line with an OpenAI-compatible endpoint (base URL, model and key in
  .env), sending the neighbouring lines for context and keeping the lines' timings.
- Keep a translation under 42 characters per line by re-wrapping, and flag lines that don't
  fit in the editor.
- Export SRT and a burned-in MP4 per language. Say in the README exactly what text is sent.

Add speaker tracking to the 9:16 crop of the video tool in this folder. Requirements:

- Detect the largest face every 5th frame with OpenCV (the bundled Haar cascade or the DNN
  face detector), smooth the crop's x-position with a moving average over 1 second, and
  never move more than 8% of the width per second.
- Fall back to the centered crop for stretches with no face.
- A preview image of the first, middle and last frame with the crop box drawn, shown before
  rendering.
- Update the README with the speed cost on a 5-minute clip.

Add a drop-folder mode to the video tool in this folder. Requirements:

- `python -m app.watch ./inbox` watches the folder with watchdog and, for each new
  video, runs transcription, removes silences over 1 second, burns subtitles with the
  default preset and writes the MP4 and SRT to ./outbox.
- Wait until a file's size stops changing before processing it; process one file at a time.
- Move originals to ./inbox/done, and failures to ./inbox/failed with an error text file.
- Update the README with a launchd or systemd example.

What you lose

AI video generation, avatars and AI voices
Translation and dubbing into 50+ languages
The stock audio and video catalog, templates and brand kits
Editing and rendering in the cloud, on any device, with shared workspaces
A multi-track timeline with effects, transitions and 4K export presets

Why people still pay

It is the quick route from raw footage to a captioned, vertical clip with nothing to install. Reviewers single out the automatic subtitles, and the plan bundles stock media and AI tools. Some reviewers on Capterra also describe billing disputes and slow long projects; we haven't verified those individual claims.

Moatproprietary models content & rights execution polish

Do the math

VEED or your own copy?

Seats

Over

Your usage

Count my time12h ×$/h

VEED, 1 yr$300

Your own copy, 1 yr$13running + build session

You keep$288pays for itself in the first month

Estimate · Everything runs on your own computer, so there is no monthly bill beyond electricity and disk. The build is token usage on your agent plan; a long video makes transcription and rendering slow on a CPU.

Leaving VEED

Take your data, then cancel.

1 · Take your data out of VEED

1Finished videos: open the project's Settings tab, choose an export quality under Export Settings and click Done. When rendering finishes, click the Download icon and choose MP4. A project still in draft has to be exported before it can be downloaded.
2Export resolutions follow the plan: 720p on Free, up to 1080p on Creator, up to 4K on Pro, Studio and Enterprise.
3The help article covers MP4 export only. We found nothing on exporting a project file or subtitle files (SRT), so check the subtitle panel for download options and keep your source footage.

2 · Cancel VEED

1In the dashboard, open the workspace menu at the top left and choose Team Billing.
2Click Turn off auto-renewal.
3You keep the premium features until the end of the billing cycle you've paid for.
4Enterprise plans are handled differently; contact your customer success manager.

Export everything you need while you still have a paid plan: a Free export is watermarked, and re-uploading a watermarked file doesn't remove the mark because it is burned in.
AI credits expire at renewal and don't roll over, so spend them before the plan ends.
Some Capterra reviewers describe projects that disappeared when a plan lapsed. We haven't verified this, so download finished videos and keep the originals.
Every invited editor is a seat. A third-party guide says each workspace has its own subscription, so cancel from the workspace you own; remove editors before the renewal date.

Checked 2026-10-19 against VEED's docs: 1, 2, 3

The escape hatch

Don't feel like building it? These are already free.

KdenliveOpen sourceFree GPL video editor with a full timeline, effects and a Speech Editor that cuts by transcript.★ 5.9k · active Oct 2026 · one-click install
macOS, Windows, LinuxVisit ↗ShotcutOpen sourceCross-platform GPLv3 video editor built on Qt and MLT.★ 15k · active Oct 2026 · one-click install
macOS, Windows, LinuxVisit ↗OpenShotOpen sourceFree open-source video editor for Linux, Mac and Windows, with a simple timeline.★ 6.7k · active Oct 2026 · one-click install
macOS, Windows, LinuxVisit ↗

All 5 free alternatives to VEED ›

The numbers

VEED pricing

Plan	Monthly	Annual, per month	What you get
Free	$0	$0	Watermarked exports at 720p, unlimited exports, 10 minutes of subtitles and 50 AI credits a month.
Creator	$25	$12.25	No watermark, exports up to 1080p, unlimited auto-subtitles (fair use 50 hours a month), the full audio and video catalog, 15+ AI tools, 500 AI credits a month.
Pro	$45	$22.08	Everything in Creator plus multiple brand kits, translation into 50+ languages (metered by the minute), up to 12 hours a month of AI voice, exports up to 4K, 2,500 AI credits a month.
Studio	$79	$38.75	Everything in Pro plus custom templates, up to 80 hours a month of AI voice, 15,000 AI credits a month.
Enterprise	custom	—	Custom AI credits, central team and data management, review mode, privacy controls, customer success.

Free tier: Watermark, 720p, 10 subtitle minutes and 50 AI credits a month.
Billing: Billed per editor seat, monthly or yearly; the yearly price is about half of the monthly one. Auto-renewal can be turned off at any time and access lasts to the end of the paid period. Prices show in your local currency. The refund policy is in the Terms of Sale, which we didn't read.
Hidden costs: AI credits don't carry over and expire at renewal; generating video costs 8 credits a second at 720p, so Creator's 500 credits come to about a minute. More credits means a bigger plan variant. Translation minutes are metered per language, and every editor you invite is another seat.
Source: www.veed.io

Also on the list

Descript$24/mo · Audio & videoKINDA

Riverside$29/mo · PodcastingKINDA

HeyGen$29/mo · AI videoNOT REALLY

Questions

VEED, briefly

Not answered here? Tell us and we'll add it.

Ask a question ›

Free adds a watermark to exports and gives 10 minutes of subtitles and 50 AI credits a month.

Kinda. The core is a weekend build with the prompt on this page, but real gaps remain: AI video generation, avatars and AI voices, Translation and dubbing into 50+ languages.

About $25 a month on the Creator plan per seat, $300 a year (checked 2026-10-19).

AI video generation, avatars and AI voices; Translation and dubbing into 50+ languages; The stock audio and video catalog, templates and brand kits; Editing and rendering in the cloud, on any device, with shared workspaces; A multi-track timeline with effects, transitions and 4K export presets. If any of those matter to you, keep paying.