CoolFace
Apppublic

dancesee/docu-genie-app

sourceHugging Faceupdated 8mo agoView on Hugging Face
1likes
App README

๐Ÿงž Docu-Genie

Transform videos into professional documents using AI.

Upload any video tutorial, training session, or walkthrough โ€” and Docu-Genie will automatically transcribe it, polish it with AI, and generate a beautiful PDF document complete with screenshots.

๐ŸŽฏ What It Does

  1. 1.Transcribes your video using OpenAI Whisper (supports 90+ languages)
  2. 2.Humanizes the transcript with Mistral AI โ€” converting raw speech into professional "Step 1, Step 2" format
  3. 3.Captures screenshots at key moments (10%, 50%, 90% of video)
  4. 4.Generates a PDF with cover page, headers, footers, and styled sections

โœจ Features

FeatureDescription
๐ŸŽฅ Video UploadDrag & drop or click to upload (MP4, MOV, AVI)
๐Ÿ—ฃ๏ธ 90+ LanguagesEnglish, Hindi, Spanish, French, German, Chinese, Arabic, and more
๐Ÿค– AI TranscriptionPowered by OpenAI Whisper (tiny model for speed)
โœ๏ธ AI HumanizationMistral-7B polishes raw speech into professional SOP format
๏ฟฝ Auto ScreenshotsCaptures key frames from your video
๐Ÿ“„ Professional PDFCover page, headers, footers, styled sections
โšก Fast ProcessingOptimized for CPU deployment

๐Ÿ› ๏ธ Tech Stack

LayerTechnology
BackendPython, Flask, Gunicorn
AI TranscriptionOpenAI Whisper
AI HumanizationMistral-7B (via Hugging Face Inference API)
Video ProcessingMoviePy, FFmpeg
PDF GenerationFPDF, Pillow
DatabaseSQLite
FrontendHTML, TailwindCSS, JavaScript
DeploymentDocker, Hugging Face Spaces

๐Ÿ“– How to Use

  1. 1.Upload your video โ€” Drag & drop or click the upload zone
  2. 2.Wait for processing โ€” The AI transcribes, humanizes, and generates your PDF
  3. 3.Download your SOP โ€” Click the download button to get your professional PDF

Processing Time (Free CPU)

Video LengthApproximate Time
1 minute~1-2 minutes
5 minutes~3-5 minutes
10 minutes~5-8 minutes

๐Ÿš€ Deploy Your Own

Hugging Face Spaces (Recommended)

  1. 1.Fork this Space or clone the repo
  2. 2.Go to Settings โ†’ Variables and Secrets
  3. 3.Add secret: HF_TOKEN = Your Hugging Face access token
  4. 4.Restart the Space

Local Development

bash
git clone https://github.com/kiteflies-ok/docu-genie-app.git
cd docu-genie-app
pip install -r requirements.txt
python app.py

Open http://localhost:7860

Docker

bash
docker build -t docu-genie .
docker run -p 7860:7860 docu-genie

๐Ÿ”ง Configuration

Environment VariableDescription
HF_TOKENHugging Face access token (required for AI humanization)

๐ŸŽฏ Supported Formats

  • โ€”Video: MP4, MOV, AVI, MKV, WebM
  • โ€”Max Size: 1GB
  • โ€”Languages: 90+ (auto-detected)

๐Ÿ“„ PDF Output Includes

  • โ€”โœ… Professional cover page with title and date
  • โ€”โœ… "Docu-Genie SOP" header on every page
  • โ€”โœ… Page numbers (e.g., "Page 1/3")
  • โ€”โœ… Screenshots at key video timestamps
  • โ€”โœ… AI-polished SOP with "Step 1, Step 2" format
  • โ€”โœ… Clean, readable formatting

๐Ÿ› Troubleshooting

IssueSolution
Timeout errorsUse shorter videos (<15 min) or upgrade to GPU Space
AI humanization failsCheck that HF_TOKEN secret is set correctly
Poor transcriptionEnsure clear audio in source video

๐Ÿ“ License

MIT License


Made with โœจ by Docu-Genie | Powered by Whisper + Mistral AI