Youtube_Extract_Clip_and_Generate_Content_Ideas

[Dr0p42]

User journey to be defined @jravenel

[FlorentLvr]

Spec:

INPUT

command: Youtube_Extract_Clip_and_Generate_Content_Ideas

params:

• "URL": "https://www.youtube.com/watch?v=-DVyjdw4t9I",
• "START": "00:00:00",
• "END": "10:22:00",
• "NAME":"Guido van Rossum: Code Readbilty, Indentations and Bugs"
• "PROMPT": By default, generate 30 contents ideas

OUTPUT

Engine

• Full video: content_generation/Youtube_Extract_Clip_and_Generate_Content_Ideas/Guido_van_Rossum_Code_Readbilty_Indentations_and_Bug/Guido van Rossum: Python and the Future of Programming | Lex Fridman Podcast #341.csv (SOURCE)
• Clip: content_generation/Youtube_Extract_Clip_and_Generate_Content_Ideas/Guido_van_Rossum_Code_Readbilty_Indentations_and_Bug/Clip_Guido van Rossum: Code Readbilty, Indentations and Bugs.mp4
• CSV/parquet Log: content_generation/Youtube_Extract_Clip_and_Generate_Content_Ideas/Guido_van_Rossum_Code_Readbilty_Indentations_and_Bug/Clip_Guido van Rossum: Code Readbilty, Indentations and Bugs.csv
• CSV/result: content_generation/Youtube_Extract_Clip_and_Generate_Content_Ideas/Guido_van_Rossum_Code_Readbilty_Indentations_and_Bug/CSV_Guido van Rossum: Code Readbilty, Indentations and Bugs_Result.csv

Chat

• Key idea, Hook, Content, Asset
• Key idea, Hook, Content, Asset

[Dr0p42]

Title: Implement New Workflow for Content Draft Generation from YouTube Clips

Potential blockers:

• Create Drafts in Buffer:
  • It might not be possible or reliable to do it, we need to investigate this part.

Description

User Story:

As a user, I want to be able to execute a command with the following parameters:

• "URL": "https://www.youtube.com/watch?v=-DVyjdw4t9I"
• "START": "00:00:00"
• "END": "10:22:00"
• "NAME":"Guido van Rossum: Code Readability, Indentations and Bugs"
• "PROMPT": By default, generate 30 content ideas

The command should automatically create 30 content drafts in Buffer. I expect to receive the 30 drafts in my chat and a link to Buffer for easy access.

As a technical user, I should be able to access the outputs directory in my JupyterLab, to view the assets or files generated for this specific command execution, as I may want to make updates to the command.

---

Requirements:

We need to create a new workflow in the models/Youtube_Extract_Clip_and_Generate_Content_Ideas/ directory. Initially, we can have one notebook to facilitate a quick launch.

Input:

• Command parameters
• Buffer cookie
• Output directory
• Workspace chat_id
• Workspace message_id
• Workspace auth token

Model:

• Download youtube video and save in output_dir/out.mp4 (if we can download mp3 directly it's better)
• Install ffmpeg
• Convert mp4 to mp3  output_dir/out.mp4 save to  output_dir/out.mp3
• Cut mp3 based on command parameters (start/end) output_dir/out_cut.mp3
• Use OpenAI Whisper to get transcript from mp3.
• Use GPT with prompt from command + transcript to generate draft contents. Ensure that the generated drafts are displayed in a markdown table or other easily accessible data structure.
• Use Markdown Lexer to get the table content and load drafts in a dataframe.

Output:

• Save dataframe in output_dir/drafts.csv
• Create Drafts in buffer.
• Generate message to send back to naas chat.
• Use chat_id, message_id and auth token, to update the message using PUT naas.ai/api/chat//message/ (Authorization: Bearer ).

[Dr0p42]

On my side I need to create issues in workspace to be able to update a message and create a new message type to handle command execution/jobs.

[FlorentLvr]

@jravenel, this template is integrated in: abi/models/content-engine/YouTube_Extract_Clip_and_Generate_Content_Ideas.ipynb As the webhook feature is not working yet, you can copy/paste the message in the input bar of your chat. Let me know if you have any questions!

[github-actions[bot]]

:tada: This issue has been resolved in version 1.0.0 :tada:

The release is available on GitHub release

Your semantic-release bot :package::rocket: