Hey everyone!
Big update just shipped v0.4.0, our largest release yet. Six new AI features, and every one of them runs locally. Your footage still never leaves your machine.
What's new
- AI dubbing, fully local a new NLLB-200 translation service batch-translates transcripts on-device (Whisper translate voice clone render). Optional Sarvam fallback only for Indian language pairs, if you configure it.
- Video background removal per-frame matting with the result dropped in as a non-destructive alpha layer above your clip. Original stays untouched.
- Auto B-roll from your own footage on-device CLIP search matches each transcript segment to clips you already have and inserts the best match. No stock libraries, no uploads.
- Multilingual captions one-click translation into multiple languages with per-track SRT/VTT export.
- Edit by speaker diarization labels who spoke when; remove, tighten, or isolate a single speaker's segments like text.
- Automatic multicam sync client-side audio cross-correlation aligns camera angles without timecode. Preview offsets with confidence scores, apply in one undoable step.
Also in this release
- Visual footage search powered by local CLIP embeddings (shipped earlier this spring)
- Thumbnail A/B testing and YouTube virality scoring
- New model support: Sarvam AI, Smallest AI, Kimi K2, Google Gemma
- Hardened upload endpoints, stabilized VPS deployment, and a repaired build/CI pipeline community contributions from @MikyJ05 and @interfluve-wav landed in this release. Open source works.
Why it matters
Every other "AI video editor" asks you to upload raw footage to the cloud. For unreleased material, client work, or anything sensitive, that's a non-starter. We keep wrapping more intelligence around the editor while keeping the architecture boring and private: Docker services on your own hardware, no telemetry, no account required.
Full changelog: /changelog/0.4.0 on the site.
What should we build next and what would you dub first?
Free AI Video Editor OpenCutAI
Hey Product Hunt!
We're launching OpenCut AI - a fully open-source, local-first AI video editor.
We took [OpenCut](https://github.com/ekaanth/openc...), an open-source video editor, and built an entire AI suite on top of it:
- Edit by text - transcribe your video, then edit it like a Google Doc. Delete a sentence, and the video cuts itself.
- Voice cloning - clone any voice from a 6-second sample and generate voiceovers.
- AI image generation - create images from text and drop them on the timeline.
- Filler word removal - one-click removal of "um", "uh", "like", and "you know."
- Natural language commands - tell the editor "remove the intro" or "speed up the middle."
- Smart subtitles - karaoke, pill, and classic styles, generated instantly.
- Audio denoising - clean up background noise automatically.
- Podcast Clip Generator - AI finds the most viral-worthy 30–60 second moments from long podcasts.
- Word-Pop Karaoke Subs - Hormozi-style subtitles where each word pops up when spoken. 4 preset styles.
- Multi-speaker detection - auto-detect who's talking with pyannote AI. Video auto-cuts at speaker boundaries.
- Auto-Reframe 9:16 - face-tracking crop for TikTok, Reels, and Shorts.
- Brand Kit - define your brand once and apply intro/outro cards, lower thirds, watermarks in one click.
- Emotion detection - SpeechBrain AI detects emotional peaks to find the most impactful moments.
Everything runs locally. No cloud. No API keys. No subscriptions. No per-minute billing. Your videos never leave your machine.
Under the hood, we're standing on the shoulders of incredible open-source models like Whisper for transcription, XTTS v2 for voice cloning, Stable Diffusion for image generation, Llama 3.2 via Ollama for natural language commands, Pyannote for speaker diarization, and MediaPipe for face detection. All of it packaged in a single `docker compose up -d`.
We built this because video editing tools are either expensive, cloud-dependent, or closed-source and we wanted one that was none of those things.
Self-host it for as low as $20/mo, or just run it on your laptop.
Would love your feedback — what AI features would you want in a video editor?
Free AI Video Editor OpenCutAI
We just shipped multilingual Indian language support [see the thread](https://x.com/humblefool/status/2036125305908371528). OpenCut AI now supports 30+ languages including 22 Indian regional languages (Hindi, Tamil, Telugu, Kannada, Bengali, Malayalam, Marathi, Gujarati, Punjabi, Odia, and more) powered by [Sarvam AI](https://sarvam.ai). Transcribe, translate, and generate voiceovers with 37+ Indian voice speakers — all from within the editor. Add your Sarvam API key in Settings and you're good to go.
Would love your feedback — what AI features would you want in a video editor?