desktop v0.4.18-alpha · mini 2026.05.28.1-alpha · open · local-first · live

Clean My Video.

Drag in any clip. Drop out a clean copy. CMVideo finds every swear and slur with on-device speech recognition, then silences, beeps, or replaces them with a TTS overlay - whichever you pick.

Download the full app choose your OS below View on GitHub

Desktop app: 100% local — no accounts, no clip uploads. Audio never leaves your machine. This web page may use optional analytics/ads with your consent.

CMVideo Mini
Clean My Video mini
Mini web version · paste a URL or drop a file

Most sites work — YouTube, X, TikTok, Reddit, Twitch and 1,800+ others.

Word filters, file size, transcript & subtitles

Mini caps: download/extract ≤1 hr / size cap scales with quality (200 MB at 144p → 4 GB at 4K) · video 144p–4K · audio in MP3 / M4A / AAC / OGG / OPUS / WAV / FLAC · 20 jobs/hour. Multi-extractor stack (yt-dlp → gallery-dl → Cobalt → streamlink) covers most mainstream sites; a few will still need the desktop app. TTS, batching, and fuzzy/leetspeak matching live in the full app ↓

YouTube works best in the desktop app Mini supports YouTube MP4 / MP3 download (best-effort) — censoring and 100% reliability live in the full app.
How does YouTube work here?

YouTube is best in the desktop app. The mini can be blocked by YouTube's anti-bot systems on shared cloud IPs, so YouTube reliability is not guaranteed here.

Censoring on the mini: Silence and Beep modes work on URLs and uploaded files (≤8 min). YouTube censoring uses the same Whisper path as other sites — reliability varies on YouTube URLs.

Every other source works as a normal download here — Vimeo, TikTok, Reddit, X, Twitch, SoundCloud, Bandcamp and ~1,800 other sites that yt-dlp supports, with the same residential-proxy boost on the sites that need it.

The desktop app is still the most reliable option for YouTube — it runs on your machine, on your IP, as a personal-use tool, with no proxy in the middle and no anti-bot surprises.

●

Local & private

Whisper-class speech-to-text runs on your machine. Audio is never uploaded; the only network calls are when you ask CMVideo to download a video from a URL.

▲

Drag-drop or URL

Drop one or many files, or paste a YouTube / yt-dlp-supported URL. Pick the output format (MP4, MOV, MP3, WAV, OGG) and quality. Batches run unattended.

◆

Silence, Beep, or TTS

Replace flagged words with crisp silence, a classic 1 kHz censor beep, or a TTS overlay - sixteen selectable voices including six livestream-style ElevenLabs voices (Brian / Adam / Sam / ...) when you supply your own API key. Add the "Retro audio" colour for a lo-fi crunch, or the File-size presets to ship a smaller MP4 / MP3.

Get CMVideo 0.4.18-alpha

Linux gets a single-file AppImage; Windows gets a single .exe — download and double-click. No installer, no Python. macOS stays source-first for now (untested).

Linux
.AppImage · chmod +x → double-click no install
Windows
.exe · double-click to run no install
macOS
.tar.gz · works via the Linux instructions untested

Linux quickstart: chmod +x CMVideo-0.4.18-alpha-x86_64.AppImage && ./CMVideo-0.4.18-alpha-x86_64.AppImage · Windows: if SmartScreen appears, choose “More info” → “Run anyway” (unsigned prerelease build). · checksums & release notes

Looking for an older build? All previous releases stay available on GitHub (0.4.0-alpha and onwards).

Three steps

  1. 1

    Drop a file or paste a URL

    MP4, MOV, MP3, WAV, OGG locally - or any YouTube / 1,800+ site URL supported by yt-dlp.

  2. 2

    Pick what to remove

    Swears, slurs, or both. Choose Silence, Beep, or TTS (sixteen voices). Add Retro audio or shrink the output via File-size presets. Optionally save a full transcript.

  3. 3

    Hit Censor

    CMVideo transcribes, finds every match (including phonetic / leet-speak variants), and writes the cleaned file next to the original.