[Feature request]: Add support for time-stamped YouTube transcript extraction

[Lumberj3ck]

What do you need?

Add support for time-stamped YouTube transcript extraction

Problem

When processing large YouTube videos through ollama, we can encounter token limit exceptions. Currently, the system extracts the entire transcript, which can exceed tokens capacity for lengthy videos.

Proposed Solution

Add functionality to extract YouTube video transcripts from a specific timestamp, allowing users to process segments of videos rather than the entire content. This would be particularly useful when:

• Working with long videos where full transcripts exceed token limits
• Only specific segments of the video are relevant for analysis
• Processing needs to start from a particular point in the video

Implementation Options

I propose two alternative approaches for implementing this feature. The team can decide which approach better fits the project's needs:

Option 1: URL Parameter Parsing

Parse the timestamp parameter ('t') directly from YouTube URLs when present

• Example: https://youtu.be/4YLtvNeRIrg?t=2934
• No additional CLI flags needed

Option 2: CLI Flag Approach

Add a new CLI flag for specifying start time:

--start-time <seconds>  # Specify start time for transcript extraction

Benefits

• Reduces token usage by processing only relevant video segments
• Prevents token limit exceptions
• Provides more precise control over content processing