# VidClean, Complete Site Content

> VidClean is a free online suite of 28 video, audio, and image tools plus a browser-based video editor: silence removal, transcription, background-noise removal, an AI clip generator, and an AI image background remover. No account required, no watermark, no monthly cap.

Updated: 2026-08-22, 28 tools live plus the editor, English + Spanish + Brazilian Portuguese, plus a partial German locale.

## Quick Facts

- Files are deleted about 90 minutes after processing, unless you save them to a project (7 days free, kept while Pro is active)
- Maximum upload size: 2GB free, 5GB Pro for video/audio, 25MB for images
- No account required, no watermark on any output, no monthly quota or daily file limit on the core file tools. AI features have small free daily caps: AI thumbnails 3/day, AI show notes 3/day, clip generator 3 videos/day
- Per-file duration caps: transcription 90 minutes free / 3 hours Pro, editor export 90 minutes free / 3 hours Pro, stabilize video 5 minutes free / 9 minutes Pro (90 seconds at 1440p or higher, all tiers), noise removal 10 minutes free / 3 hours Pro, repair audio 10 minutes free / 3 hours Pro, enhance speech 10 minutes free / 3 hours Pro, add-subtitles burn 30 minutes, trim 30 minutes, GIF 30 seconds
- Free transcription has no monthly minutes cap and no daily file limit, unlike Otter (300 minutes/month) and Turboscribe (3 files/day)
- Supported input formats: MP4, MOV, MKV, WebM (video), MP3, WAV, M4A (audio), JPG, PNG, WebP (images)
- Projects: 3 free projects keep live media for 7 days; Pro projects are unlimited with 10GB of permanent storage
- Transcription runs on NVIDIA Parakeet-TDT 0.6B v3 (primary, English, Spanish, Portuguese, and 22 other European languages) with OpenAI Whisper small for language detection and fallback. Background noise removal, repair audio, and enhance speech use DeepFilterNet3. Face and object detection for blur uses YuNet. AI thumbnails use Ideogram v3 Turbo. AI show notes and clip selection use Claude (Haiku and Sonnet). Image background removal uses rembg u2net. Every model is self-hosted on Railway CPU workers, no GPU.
- Processing runs on Railway servers (US region) on CPU, no GPU. Video processing uses FFmpeg
- Built and maintained by Melvin Bucio, Huntsville, Alabama, USA. Site launched April 2026
- All tools are genuinely free: no account, no watermark, no monthly cap. A Pro tier is $8.99/month. Pro adds 5GB uploads, 3-hour transcription and editor exports, 3-hour audio cleanup (noise removal, repair, enhance) up from 10 minutes free, background music in the editor, styled burned-in captions, 10GB of permanent project storage, unlimited projects, priority processing, and Pro batch tools. The free tools stay free.

## 中文 (Chinese Quick Facts)

- 免费在线视频和音频工具，无需账号，无水印，无每月上限
- 28 个工具，外加一个视频编辑器
- 支持中文语音转录（由 Whisper 提供，可导出 TXT、SRT、VTT）
- 也支持去除静音、降低背景噪音、压缩视频、添加字幕、格式转换
- 文件在处理后 1 小时自动删除，我们从不查看或保存你的内容
- 免费上传上限 2GB（Pro 为 5GB）
- 网址：https://vidclean.net

## Languages

- English: https://vidclean.net/ (default)
- Spanish: https://vidclean.net/es/ (68 Spanish pages: the full tool set, editor, pricing, about, compare, 12 competitor comparison (vs) pages, six blog guides, and the YouTube and podcast landing pages. Same backends as the English tools.)
- Brazilian Portuguese: https://vidclean.net/pt-br/ (65 Brazilian Portuguese pages: the full tool set, editor, pricing, about, compare, 12 vs pages, three blog guides, and the YouTube and podcast landing pages. Same backends as the English tools.)
- German: https://vidclean.net/de/ (10 German pages: silence remover, stabilize, merge, crop, mp4-to-mp3, remove background noise, extract audio, mute video, and pricing. Same backends as the English tools.)

## Editor (Flagship Product)
URL: https://vidclean.net/editor

VidClean's editor is a free browser-based video editor. Its core idea is edit-by-transcript: VidClean transcribes your upload, and deleting words in the transcript cuts the matching footage. It also has a real multi-clip timeline, one-click silence and filler-word removal that you review before applying, text overlays, auto captions with font/size/color/background styling that round-trips with the project, background music (Pro), a short-form 9:16 mode, and AI show notes. Renders export as MP4 up to 90 minutes free (3 hours Pro). No account, no watermark. Saved projects keep your edits; media is retained 7 days on free (3 projects) and permanently on Pro (10GB). Available in English (/editor), Spanish (/es/editor), and Brazilian Portuguese (/pt-br/editor).

## Comparisons (Summary)

- VidClean vs Descript (https://vidclean.net/vs/descript): free vs paid (Descript: Hobbyist $24/mo, Creator $35/mo, Business $65/mo); no account vs required; browser vs desktop app; Descript moved core AI features to metered credits in Sept 2025. VidClean now matches the core workflow (transcription, text-based editing, AI clips) for free.
- VidClean vs Kapwing (https://vidclean.net/vs/kapwing): no watermark vs watermarked free tier; no account vs required; Smart Cut silence removal locked behind Kapwing Pro at $16-24/mo.
- VidClean vs Clipchamp (https://vidclean.net/vs/clipchamp): no account vs Microsoft account required; 0.5s adjustable silence threshold vs fixed 3s minimum; no storage vs mandatory OneDrive backup on Windows 11.
- VidClean vs Clideo (https://vidclean.net/vs/clideo): 2GB upload limit vs 500MB free cap; no watermark vs watermarked free exports.
- VidClean vs VEED (https://vidclean.net/vs/veed): no watermark vs watermarked; no resolution cap vs 720p free limit; no monthly cap vs 10-minute clip cap; no account vs required.
- VidClean vs Otter (https://vidclean.net/vs/otter): free transcription with no monthly cap vs Otter's 300 minutes/month free (as of June 2026); no account vs required; Otter is better for live meeting transcription and team collaboration.
- VidClean vs TurboScribe (https://vidclean.net/vs/turboscribe): no daily file limit vs TurboScribe's 3 files/day free (as of June 2026); no account vs required; TurboScribe's paid tier is 'Unlimited' ($10/mo annual, $20/mo monthly).
- VidClean vs Opus Clip (https://vidclean.net/vs/opus-clip): free 3 videos/day with no watermark vs Opus Clip's 60 processing minutes/month free with a watermark (as of July 2026); no account vs required; Opus Clip's paid plans ($15-29/mo) add a virality score, face-tracked reframing, B-roll, and scheduling.
- VidClean vs Vizard (https://vidclean.net/vs/vizard): free 3 videos/day with no watermark vs Vizard's 60 upload credits free with a watermark and 3-day expiry (as of July 2026); no account vs required.
- VidClean vs CapCut (https://vidclean.net/vs/capcut): free, no account, no weekly export cap vs CapCut's free tier capping exports at 6/week before a watermark and limiting AI auto-captions to 10 minutes (as of July 2026); CapCut's ToS grant ByteDance a perpetual license. CapCut is fully available in the US again after the brief Jan 2025 PAFACA removal, but ByteDance has not completed the structural separation PAFACA requires.
- VidClean vs Adobe Podcast (https://vidclean.net/vs/adobe-podcast): free noise removal and speech enhancement on video and audio with no daily cap vs Adobe's free Enhance Speech being audio-only and capped at 1 hour/day; no account vs Adobe account required.
- VidClean vs Cleanvoice (https://vidclean.net/vs/cleanvoice): free noise removal and filler-word flagging with no per-minute billing vs Cleanvoice having no free tier (paid per-minute); no account vs required.
- VidClean vs Canva (https://vidclean.net/vs/canva): focused video cleanup (silence removal, transcript editing, AI audio cleanup) vs Canva being a design tool with a video editor added on and none of those; free, no account, no watermark on exports.
- Compare hub (https://vidclean.net/compare): Full comparison hub covering all 13 competitor pages (Descript, Kapwing, Clipchamp, Clideo, VEED, Otter, TurboScribe, Opus Clip, Vizard, CapCut, Adobe Podcast, Cleanvoice, Canva) on pricing, watermarks, file limits, and feature gaps.

## Tools

### Silence Remover
URL: https://vidclean.net/silence-remover

Detects pauses below -35dB lasting 0.5s+ using the FFmpeg silencedetect filter; a two-pass approach preserves natural speech rhythm. Supports MP4, MOV, MKV, WebM, MP3, WAV, M4A up to 2GB. Free, no account, no watermark, no monthly cap.

### Remove Background Noise
URL: https://vidclean.net/remove-background-noise

Runs the DeepFilterNet3 neural network on CPU. Speech-optimized noise suppression with a free in-browser A/B before/after preview. Max 10 minutes free / 3 hours Pro, 2GB file size.

### Repair Audio
URL: https://vidclean.net/repair-audio

Three-stage pipeline: loudnorm to -16 LUFS, then DeepFilterNet3 noise removal, then a 50/60 Hz de-hum notch chain. Free A/B preview. Max 10 minutes free / 3 hours Pro, 2GB.

### Enhance Speech
URL: https://vidclean.net/enhance-speech

Enhances speech with DeepFilterNet3 and three strength presets (Light/Medium/Strong blend) to avoid the over-processed sound. A free Adobe Podcast Enhance Speech alternative. Max 10 minutes free / 3 hours Pro, 2GB.

### Remove Filler Words
URL: https://vidclean.net/remove-filler-words

Flags filler words (um, uh, er, ah, hmm) using the transcript's word-level timestamps. A free-transcript alternative to Descript and Cleanvoice.

### Transcribe Audio & Video
URL: https://vidclean.net/transcribe

Free video and audio to text plus subtitles, powered by NVIDIA Parakeet-TDT 0.6B v3 (25 European languages) and OpenAI Whisper (detection and other languages). Download .txt, .srt, and .vtt. Up to 90 minutes free (up to 3 hours Pro), no monthly minutes cap or daily file limit. A free alternative to Otter (300 min/month) and Turboscribe (3 files/day).

### Transcribe Audio Recording to Text
URL: https://vidclean.net/transcribe-audio

Audio-only transcription landing page (MP3, WAV, M4A). Same Parakeet-TDT + Whisper backend as Transcribe. Returns .txt, .srt, .vtt. Up to 90 minutes free.

### Free Transcription With No Account
URL: https://vidclean.net/free-transcription-no-account

Transcription with no sign-up, no email, no credit card. Same hybrid backend. Returns .txt, .srt, .vtt.

### Convert MP3 to Text
URL: https://vidclean.net/mp3-to-text

MP3-to-text transcription; also accepts WAV and M4A. Free, no account, .txt/.srt/.vtt output, no monthly cap. No speaker diarization.

### Convert WAV to Text
URL: https://vidclean.net/wav-to-text

WAV-to-text transcription; also accepts MP3 and M4A. Free, no account, no monthly cap.

### Convert M4A to Text
URL: https://vidclean.net/m4a-to-text

M4A-to-text transcription; also accepts MP3 and WAV. Free, no account, no monthly cap.

### Free Subtitle Generator
URL: https://vidclean.net/subtitle-generator

Generates downloadable .srt and .vtt subtitle files (not burned in; see Add Subtitles for burn-in). Same hybrid backend. No watermark, no account.

### Extract Audio from Video
URL: https://vidclean.net/extract-audio

Pulls the audio track out of a video as a 192kbps MP3 (libmp3lame, single FFmpeg pass, -vn). Free, no watermark, up to 2GB.

### Convert MP4 to MP3
URL: https://vidclean.net/mp4-to-mp3

MP4-only MP3 converter (libmp3lame 192k CBR, one FFmpeg pass). Same extract-audio backend, accepts only .mp4. No account, no watermark, files up to 2GB.

### Video to MP3 Converter
URL: https://vidclean.net/convert-video-to-mp3

Convert any video (MP4, MOV, MKV, WebM) to a 192kbps MP3 in one FFmpeg pass. A free alternative to Clideo, which watermarks free exports and caps free uploads at 500MB.

### Convert MOV to MP3
URL: https://vidclean.net/mov-to-mp3

Convert a MOV video to a 192kbps MP3. Free, no account, no watermark, files up to 2GB.

### Convert M4A to MP3
URL: https://vidclean.net/m4a-to-mp3

Convert M4A audio to MP3 (192kbps libmp3lame; also accepts WAV and MP3). Free, no account, no watermark.

### Convert WAV to MP3
URL: https://vidclean.net/wav-to-mp3

Convert a large WAV to a much smaller 192kbps MP3 (also accepts M4A and MP3). Free, no account, no watermark.

### Free Background Remover
URL: https://vidclean.net/remove-background

Removes an image background with the rembg u2net model on CPU. Returns a full-resolution transparent PNG. JPG, PNG, WebP up to 25MB. Free, no account, no watermark, no credits, no daily limit. Unlike remove.bg, the full-resolution download is not paywalled.

### AI Thumbnail Maker
URL: https://vidclean.net/ai-thumbnail

Generates 16:9 YouTube thumbnails from a text prompt using Ideogram v3 Turbo, which renders legible in-image text. Free: 3 generations/day at 2 variants each, no watermark, full resolution.

### Blur Face & Objects
URL: https://vidclean.net/blur-face

Blur faces, license plates, and any object in a photo. YuNet auto-detects faces instantly; draw regions to blur anything. Gaussian blur, pixelate, or solid fill. JPG, PNG, WebP. Free, no account, no watermark, files deleted after about 90 minutes.

### Compress Video
URL: https://vidclean.net/compress-video

H.264 (libx264) at CRF 26 veryfast, AAC 128k, MP4 with +faststart. Typical reduction 60-75% with no visible quality loss. A free alternative to Clideo and VEED. Files up to 2GB.

### Convert MKV to MP4
URL: https://vidclean.net/mkv-to-mp4

Convert MKV to a web-ready H.264 MP4 (libx264 + AAC). Free, no account, no watermark, files up to 2GB.

### Convert WebM to MP4
URL: https://vidclean.net/webm-to-mp4

Convert WebM (VP8/VP9) to a compatible H.264 MP4 (re-encode required). Free, no account, no watermark, files up to 2GB.

### Convert MOV to MP4
URL: https://vidclean.net/mov-to-mp4

Convert a MOV (QuickTime) clip to a widely compatible H.264 MP4 (libx264 + AAC). Free, no account, no watermark, files up to 2GB.

### Trim Video
URL: https://vidclean.net/trim-video

Frame-accurate trim via full re-encode (libx264 CRF 23 + AAC) to avoid keyframe off-by-seconds errors. Output capped at 30 minutes; source uploads up to 2GB.

### Crop Video
URL: https://vidclean.net/crop-video

FFmpeg crop filter with platform presets (9:16 1080x1920, 1:1, 4:5, 16:9). Custom pixel dimensions up to 7680px. libx264 CRF 23, AAC 128k.

### Resize Video to Vertical
URL: https://vidclean.net/resize-video

Outputs 1080x1920 9:16 MP4 for TikTok, YouTube Shorts, and Reels. Two styles: center-crop or blurred-background bars.

### Resize Video for TikTok
URL: https://vidclean.net/resize-video-for-tiktok

Same backend as Resize Video to Vertical, framed for TikTok creators who want control over reframing instead of TikTok's auto-crop.

### Merge Videos
URL: https://vidclean.net/merge-videos

FFmpeg concat with every input normalized to 1280x720@30fps SAR=1 so mismatched resolutions and codecs join cleanly. Up to 5 files, drag to reorder.

### Change Video Speed
URL: https://vidclean.net/change-video-speed

FFmpeg setpts for video plus an atempo chain for audio (pitch preserved). Presets plus a custom slider 0.25x to 4x.

### Rotate Video
URL: https://vidclean.net/rotate-video

Transpose-filter re-encode so rotation is baked into pixels, not just metadata some players ignore. 90 clockwise, 90 counter, or 180 degrees.

### Mute Video
URL: https://vidclean.net/mute-video

Stream copy (-c:v copy -an), no re-encode, essentially instant. WebM inputs fall back to a clean libx264 re-encode.

### Stabilize Video
URL: https://vidclean.net/stabilize-video

FFmpeg vidstab two-pass stabilization with three intensity presets (Subtle, Normal, Strong). Best for handheld phone footage and mild to moderate shake. Max 5 minutes free, 9 minutes Pro. Footage at 1440p or higher is capped at 90 seconds at every tier, because 4K takes about 5x longer to process than 1080p. Files up to 2GB.

### Convert Video to GIF
URL: https://vidclean.net/video-to-gif

Two-pass palette generation (split, palettegen, paletteuse) in one FFmpeg pass to avoid color banding. 480px wide, 12fps, up to 30 second clips.

### Generate Clips
URL: https://vidclean.net/clip-generator

AI clip generator that turns a long video into short 9:16 clips for TikTok, Reels, and Shorts. Parakeet + Whisper transcribes, Claude picks self-contained moments snapped to sentence boundaries. Free: 3 videos/day, up to 6 clips from sources up to 90 minutes. No account, no watermark.

### Add Subtitles to Video
URL: https://vidclean.net/add-subtitles

Auto-generates and burns captions into video (Parakeet + Whisper transcribe, libass hardcodes styled subtitles); the .srt is also returned. Live caption-style preview and all 10 fonts are free; burning a custom style with 9 Pro fonts is Pro. Free, no account, no watermark, up to 2GB and 30 minutes.

## Landing Pages

### Free Silence Remover for YouTube Videos
URL: https://vidclean.net/youtube

Guide for YouTube creators: silence removal for long recordings, noise reduction, compression for upload limits, audio extraction. Includes 1080p/4K/Shorts export specs.

### Free Silence Remover for Podcasts
URL: https://vidclean.net/podcast

Podcast workflows: silence removal between segments, repair audio for field recordings, noise removal, MP4 to MP3. Includes Spotify, Apple Podcasts, and Buzzsprout export specs.

### The Free AI Video Editor
URL: https://vidclean.net/ai-video-editor

Positions the editor as a free AI video editor: transcription, silence removal, filler cleanup, captions, show notes, thumbnails, powered by real self-hosted models. No account, no watermark.

### Text-Based Video Editor
URL: https://vidclean.net/text-based-video-editor

Edit video by editing the text: delete a word in the transcript and the video cuts itself at that spot. Automatic transcription, no account, no watermark.

### Free Podcast Video Editor
URL: https://vidclean.net/podcast-video-editor

A free podcast video editor: AI removes silences and filler words, transcribes, adds captions, and writes show notes. No account, no watermark.

## Blog
URL: https://vidclean.net/blog

- How to Remove Silence From a Video for Free (https://vidclean.net/blog/how-to-remove-silence-from-video-free), April 2026: Four free methods compared: VidClean, Clipchamp, Descript, and Filmora, with honest tradeoffs. The transcription cluster's silence hub.
- How to Remove Background Noise From a Video for Free (4 Methods) (https://vidclean.net/blog/how-to-remove-background-noise-from-video-free), May 2026: Compares VidClean (DeepFilterNet3), Adobe Enhance Speech, Audacity, and Cleanvoice for wind, hum, fan noise, static, and room echo.
- How to Stabilize Shaky Video for Free (4 Methods) (https://vidclean.net/blog/how-to-stabilize-shaky-video-for-free), May 2026: Compares VidClean (vidstab, three presets), DaVinci Resolve, CapCut, and Google Photos for handheld, walking, and action-cam shake.
- How to Extract Audio from a Video for Free (https://vidclean.net/blog/how-to-extract-audio-from-video-free), May 2026: Three free ways to pull audio as MP3: VidClean (browser, no account), Adobe Premiere Pro (paid), and VLC (free desktop).
- How to Compress a Video for Free Without Losing Quality (https://vidclean.net/blog/how-to-compress-a-video-for-free), May 2026: Three free ways to shrink video 60-75%: VidClean (browser), HandBrake (desktop, full control), and QuickTime on Mac.
- How to Repair Audio for Free Online (4 Methods) (https://vidclean.net/blog/how-to-repair-audio-for-free), May 2026: Compares VidClean (loudnorm + DeepFilterNet3 + de-hum in one pass), Adobe Podcast Enhance Speech, Audacity, and iZotope RX Elements for noise, hum, volume, and reverb.
- How to Transcribe a Video to Text for Free (https://vidclean.net/blog/how-to-transcribe-a-video-to-text-free), June 2026: Five free ways compared: VidClean (no account, no monthly minutes cap), YouTube auto-captions, Otter, TurboScribe, and Google Docs voice typing. The transcription cluster hub.
- How to Transcribe a YouTube Video for Free (https://vidclean.net/blog/how-to-transcribe-a-youtube-video-free), June 2026: YouTube's built-in transcript, free URL tools, and uploading raw footage to VidClean for a clean SRT.
- Best Free Opus Clip Alternatives (2026) (https://vidclean.net/blog/best-free-opus-clip-alternatives), June 2026: Honest roundup: VidClean (3 videos/day, no watermark), Vizard's free tier, manual CapCut clipping, and the case for paying for Opus Clip at team scale.
- Free Tools for YouTubers That Need No Account and Add No Watermark (https://vidclean.net/blog/free-tools-for-youtubers-no-account-no-watermark), June 2026: A roundup of free YouTube tools with no watermark or forced account: VidClean cleanup, transcription, editing, and thumbnails alongside DaVinci Resolve, OBS, Audacity, and Canva.
- Parakeet vs Whisper on CPU: 1,231 Production Jobs (https://vidclean.net/blog/parakeet-vs-whisper-cpu-transcription), August 2026: Original engineering benchmark of NVIDIA Parakeet-TDT 0.6B v3 against faster-whisper small on the same CPU container, from 1,231 production transcription jobs between August 1 and 20, 2026. Controlling for file length, Parakeet's median real-time factor is 0.099 seconds of compute per second of audio against Whisper's 0.160, a 1.6x gap rather than the 2.1x raw medians suggest, while ninetieth percentiles differ by 4x. Whisper's median varies 12x across languages (Hindi 0.736 to Japanese 0.061) because its autoregressive decode cost tracks output tokens, while Parakeet's transducer varies only 1.7x, so Whisper capacity depends on language mix. Fixed per-job overhead dominates below about two minutes and is a distribution rather than a constant: on files of 30 seconds or less, processing takes a median 4.5 seconds but a p90 of 66.2 and a p99 of 145.6, driven by download time scaling with file size. Shows a linear planning model fitted over one week (20.5 + 0.1317 x duration) failing across three, with the slope moving between 0.0821 and 0.1317 on unchanged code. Includes a field guide on the two opposite conventions behind the phrase "3x realtime", the ONNX Runtime settings required to fit Parakeet in a 2GB container, and a per-job CSV of all 1,231 timings. Published under CC BY 4.0.
- 7 Free Online Video Editors With No Watermark (Actually Tested, 2026) (https://vidclean.net/blog/free-online-video-editors-no-watermark), July 2026: Opened the real free tier of 7 editors and checked for watermarks, forced accounts, export caps, and file limits. Honest results, no affiliate spin.
- How to Add Captions to a Video for Free (2 Ways, 2026) (https://vidclean.net/blog/how-to-add-captions-to-video-free), July 2026: Two free ways: burn AI subtitles in with a single upload, or edit the wording first in the editor. No account, no watermark.
- Text-Based Video Editing, Free: Edit Video by Editing the Transcript (https://vidclean.net/blog/text-based-video-editing-free), July 2026: What text-based editing is and how VidClean's editor cuts footage when you delete a word in the transcript. Free, no account, no watermark.
- 7 Best Free Descript Alternatives in 2026 (Actually Tested, No Watermark) (https://vidclean.net/blog/best-free-descript-alternatives), July 2026: Honest test of 7 free Descript alternatives for text-based editing, silence removal, and captions: VidClean, CapCut, Clideo, Veed, Adobe Premiere, Kdenlive, and Audacity, with a comparison table. VidClean is the closest free match to the transcript-editing workflow.
- We Analyzed 144 Hours of Creator Recordings. A Quarter of It Was Silence. (https://vidclean.net/blog/how-much-of-a-recording-is-silence), August 2026: Original data study from 822 silence-removal jobs, July 20 to August 4, 2026: 24.9% of uploaded recording time was dead air, with a bimodal distribution (55% of recordings under 5% removed; 16% over 30%; 1 in 20 mostly silence). Aggregate anonymized counters only. Includes methodology and limitations, and is published under CC BY 4.0 so the figures may be quoted with attribution.
- We Analyzed 1,533 Audio Cleanup Jobs. Two Thirds Weren't Audio Files. (https://vidclean.net/blog/audio-cleanup-statistics), August 2026: Original data study from 1,533 audio cleanup jobs, July 20 to August 11, 2026, across three DeepFilterNet3 tools (remove background noise 53%, repair audio 28%, enhance speech 19%): 67% of uploads were video files rather than audio files, and 86% ran under five minutes. Aggregate anonymized counters only. Includes methodology and limitations, and is published under CC BY 4.0 so the figures may be quoted with attribution.
- We Analyzed 2,039 Shaky Videos. Almost All of Them Were Under a Minute. (https://vidclean.net/blog/shaky-video-statistics), August 2026: Original data study from 2,039 video stabilization jobs, July 20 to August 17, 2026: 87.7% of stabilized videos ran under one minute, 14.3% of uploads were 4K or larger, and when people change the default strength they turn it up rather than down by about 7 to 1. Also reports measured processing cost per second of video on production hardware (1080p 2.61x, 1440p 4.16x, 4K 8.52x realtime), and finds cost tracks pixels per second rather than pixels, so 1080p at 60fps costs more to stabilize than 1440p at 30fps. Aggregate anonymized counters plus production log timings, with a downloadable CSV. Includes methodology and limitations, and is published under CC BY 4.0 so the figures may be quoted with attribution.
- We Transcribed Two Million Words. Most Jobs Produced Under a Hundred. (https://vidclean.net/blog/transcription-statistics), August 2026: Original data study from 1,397 transcription jobs, July 20 to August 19, 2026, totalling 2,269,873 words across 48 system-detected languages. 37.2% of jobs produce fewer than 100 words while 21.9% produce over 2,500, so the mean of 1,625 words per job describes the tail rather than the typical job. English has fallen to 59.8% of jobs from 71% a month earlier, and 13 languages appear exactly once. Demonstrates that the duration histogram's apparent second peak at 30 to 60 minutes is an artifact of unequal bucket widths: raw counts rise from 118 to 196 but per-minute density falls from 7.87 to 6.53. Deliberately excludes a words-per-minute metric present in the source counters because it divides word count by full media duration including silence. Aggregate anonymized counters with a downloadable CSV. Includes methodology and limitations, and is published under CC BY 4.0 so the figures may be quoted with attribution.
- AI Assistants Are Googling On Your Behalf. They Almost Never Click. (https://vidclean.net/blog/ai-agent-search-queries), August 2026: Original data study of machine-issued search queries seen in one small site's own Google Search Console, across two complete adjacent 28-day windows (June 21 to July 18 and July 19 to August 15, 2026). Queries of eight or more words grew from 104 to 282 and from 438 to 1,448 impressions, about 1.7 times the rate the site itself grew. Within that class, queries shaped like full sentences produced 1 click from 630 impressions, a 0.16% clickthrough rate against a 5.92% site average, while ordinary long-tail keyword strings of the same length clicked at 8.03%, so sentence shape rather than query length is what identifies machine traffic. One competitor comparison page absorbed 80 distinct rephrasings of a single pricing claim with zero clicks on all 80, at an impression-weighted position of 6.33. Also documents a query that arrived with its prompt scaffolding still attached, in both windows, and fixed eleven-domain site exclusion chains. Notes that Search Console withholds the queries behind 41.8% of impressions, so every count is a lower bound. Relevant to answer engine optimization (AEO) and generative engine optimization (GEO). Per-query CSV for both windows is downloadable. Includes methodology and limitations, and is published under CC BY 4.0 so the figures may be quoted with attribution.

## Pricing
URL: https://vidclean.net/pricing

All tools are genuinely free: no account, no watermark, no monthly cap. A Pro tier is $8.99/month. Pro adds 5GB uploads, 3-hour transcription and editor exports, 3-hour audio cleanup (noise removal, repair, enhance) up from 10 minutes free, background music in the editor, styled burned-in captions, 10GB of permanent project storage, unlimited projects, priority processing, and Pro batch tools. The free tools stay free.

## Technical Details

- Backend: Python FastAPI with an ARQ job queue and Redis
- Storage: Cloudflare R2; a deferred cleanup deletes job files about 90 minutes after completion (JOB_TTL)
- Frontend: static HTML deployed on Vercel
Transcription runs on NVIDIA Parakeet-TDT 0.6B v3 (primary, English, Spanish, Portuguese, and 22 other European languages) with OpenAI Whisper small for language detection and fallback. Background noise removal, repair audio, and enhance speech use DeepFilterNet3. Face and object detection for blur uses YuNet. AI thumbnails use Ideogram v3 Turbo. AI show notes and clip selection use Claude (Haiku and Sonnet). Image background removal uses rembg u2net. Every model is self-hosted on Railway CPU workers, no GPU.
- Silence threshold: -35dB, minimum duration 0.5s, 0.1s padding preserved around cuts

## Contact and Links

- Website: https://vidclean.net
- Contact: hello@vidclean.net
- Built by: Melvin Bucio, Huntsville, Alabama, USA
- Site launched April 2026