Log inSign up
●16 reads●MIT License

Effortless AI Summaries from Videos to Notion

Table of contents

YouTube to Notion Tool

This tool converts YouTube videos into text using Whisper for speech-to-text and extracts summaries with Llama, automatically saving the results to a Notion database.
Demo Video

Project Structure

/STTnotion
|-- config.py
|-- your_ideal_model.gguf
|-- downloaded_audio
|   `-- yt_vid_you_download.wav
|-- main.py
|-- notion_handler.py
|-- requirements.txt
|-- whisper_handler.py
`-- youtube_handler.py

Requirements

To run this tool, you need to:

  1. Install the necessary libraries: streamlit, youtube_dl, whisper, notion-client
  2. Obtain a Notion API key and database ID
  3. Install FFmpeg (for audio processing)

Setting Up the Environment

Create a .env file to store your Notion API key and database ID:

NOTION_TOKEN=your_notion_api_key_here
NOTION_DATABASE_ID=your_notion_database_id_here

Running the Program

  1. Install the required libraries:

    pip install -r requirements.txt

  2. Ensure you have FFmpeg installed.

  3. Set up your .env file with the Notion API key and database ID.

  4. Download your ideal language model(.gguf file), like LLama. Put the model file under STTnotion folder.

  5. Edit MODEL_PATH in llama_handler.py.

  6. Run the program:

    streamlit run main.py

Troubleshooting

FFmpeg Installation

To check if FFmpeg is installed, open your command line (terminal) and type:

ffmpeg -version

If FFmpeg is installed, you will see version information. If not, you’ll receive an error message.

  • Installing FFmpeg:

    • Windows:

      • Use Chocolatey (package manager):

        choco install ffmpeg

      • Or download from the official website.

    • macOS:

      • Use Homebrew:

        brew install ffmpeg

    • Linux (Ubuntu/Debian):

      sudo apt update
sudo apt install ffmpeg

After installation, run ffmpeg -version again to confirm success.

YouTube-DL Issues

If you encounter an error such as Unable to extract uploader id, you may need to update yt-dlp:

python3 -m pip install --force-reinstall <https://github.com/yt-dlp/yt-dlp/archive/master.tar.gz>

Then, run:

yt-dlp URL

Ensure you import it correctly in your project:

import yt_dlp as youtube_dl

Notion API Setup

  1. Create a Notion integration to obtain the API key.
    image

  2. Add the integration to your Notion database to retrieve the database ID.
    image1

  3. To find your Notion database ID, locate the URL of your database:

    https://www.notion.so/<long_hash_1>?v=<long_hash_2>

    <long_hash_1> is your database ID.

Property Settings in Notion

If you encounter errors such as “Title is not a property that exists” or “Content is not a property that exists”, ensure the properties are correctly defined in your Notion database. Update the notion_handler.py with the correct property names:
image2

Original Code:

def create_notion_page(text):
    new_page = notion.pages.create(
        parent={"database_id": NOTION_DATABASE_ID},
        properties={
            "Title": {"title": [{"text": {"content": "YouTube Transcription"}}]},
            "Content": {"rich_text": [{"text": {"content": text[:2000]}}]},  # Notion API limit
        },
    )

Updated Code:

page_properties = {
    "Name": {
        "title": [
            {
                "text": {
                    "content": file_name
                }
            }
        ]
    },
    "Tags": {
        "multi_select": [
            {"name": "content"}
        ]
    }
}
new_page = notion.pages.create(
    parent={"database_id": NOTION_DATABASE_ID},
    properties=page_properties
)

This refined version includes clear instructions, troubleshooting tips, and code examples that will help users understand and effectively use your tool.

Models

Datasets