vid-mediakit 0.1.0

Download, transcribe, and convert videos from YouTube, Loom, Vimeo, and more

mediakit

Download, transcribe, and convert videos from YouTube, Loom, Vimeo, and more — as a Python library.

from mediakit import download_video, get_transcription, convert

path  = download_video("https://www.loom.com/share/abc123")
text  = get_transcription(path)            # requires OPENAI_API_KEY
mp3   = convert(path, output_format="mp3", reencode=True)

---

Features

Module	What it does
download_video()	Download from YouTube, Vimeo, Loom, Wistia, and any yt-dlp supported site
get_transcription() / VideoTranscriber	Transcribe audio/video via OpenAI Whisper, auto-chunks files > 25 MB
VideoScriptSummarizer	Summarize a transcript with GPT-4
convert()	Convert any video/audio format using ffmpeg (mp4→mp3, mkv→mp4, etc.)

---

Requirements

• Python ≥ 3.9
• ffmpeg (for conversion and audio extraction) — install instructions below
• OpenAI API key (for transcription only)

---

Installation

From PyPI (once published)

pip install mediakit

Local / development install

git clone https://github.com/yourusername/mediakit
cd mediakit
pip install -e .

With FastAPI server support

pip install "mediakit[server]"

---

Setup

Copy .env.example to .env and add your OpenAI key:

cp .env.example .env
# edit .env and set OPENAI_API_KEY=sk-...

---

Usage

Download a video

from mediakit import download_video

# Basic download → saves to ./downloads/
path = download_video("https://youtu.be/dQw4w9WgXcQ")

# Custom output directory and quality
path = download_video("https://youtu.be/dQw4w9WgXcQ", output_dir="my_videos", ytdlp_format="720p")

Supported format presets: default, best, worst, 360p, 480p, 720p, 1080p, 1440p, 4k

---

Transcribe a video or audio file

from mediakit import get_transcription, VideoTranscriber

# One-liner
text = get_transcription("video.mp4")

# Full control
transcriber = VideoTranscriber()
text = transcriber.transcribe_video("video.mp4")  # handles large files automatically

Files larger than 25 MB are automatically split into chunks before sending to Whisper.

---

Summarize a transcript

from mediakit import VideoScriptSummarizer

summarizer = VideoScriptSummarizer()
summary = summarizer.generate_summary(transcript_text)

---

Convert video/audio formats

from mediakit import convert

# mkv → mp4 (stream-copy, fast, no re-encode)
convert("recording.mkv", output_format="mp4")

# mp4 → mp3 (re-encode audio)
convert("video.mp4", output_format="mp3", reencode=True)

# Explicit output path
convert("video.mp4", output_path="/tmp/audio.wav", reencode=True)

# Overwrite if exists
convert("video.mp4", output_format="mp3", reencode=True, overwrite=True)

Supported formats: anything ffmpeg handles — mp4, mp3, mkv, wav, m4a, ogg, flac, opus, webm, etc.

---

Bulk download from Excel

python -m video_downloader.excel_bulk urls.xlsx

The Excel file needs a column named url (or URL / link). A filename column is optional.

---

ffmpeg

ffmpeg is required for conversion and audio extraction.

# Ubuntu / Debian
sudo apt install ffmpeg

# macOS
brew install ffmpeg

# Windows
# Download from https://ffmpeg.org/download.html and add to PATH

---

Architecture

┌─────────────────────────────────────────────────────┐
│                    mediakit                          │
│  ┌──────────────┐  ┌────────────────┐  ┌──────────┐ │
│  │  downloader  │  │  transcription │  │converter │ │
│  │  (yt-dlp +   │  │  (Whisper API  │  │ (ffmpeg  │ │
│  │   requests)  │  │   + GPT-4)     │  │subprocess│ │
│  └──────────────┘  └────────────────┘  └──────────┘ │
└─────────────────────────────────────────────────────┘
         │                   │                  │
   YouTube/Vimeo         OpenAI API         local ffmpeg
   Loom/Wistia           Whisper + GPT      (system dep)

---

Publishing to PyPI

Here's the full flow to publish your own version:

1. Register at https://pypi.org
2. Enable 2FA on your account
3. Generate an API token (Account Settings → API tokens)

4. Build the package:
   pip install build twine
   python -m build          # creates dist/*.whl and dist/*.tar.gz

5. Test on TestPyPI first (optional but recommended):
   twine upload --repository testpypi dist/*
   pip install --index-url https://test.pypi.org/simple/ mediakit

6. Publish to production PyPI:
   twine upload dist/*
   # enter: __token__ as username, your API token as password

7. Anyone can now install it:
   pip install mediakit

Tip: Store your token in ~/.pypirc or as env var TWINE_PASSWORD so you don't paste it every time.

---

Running Tests

pip install pytest
pytest tests/

---

License

MIT