Semantic Search for YouTube Videos using Ollama nomic-embed-text and Link to Video Location

Описание: In this project, we transcribe a YouTube video using OpenAI's Whisper, use Ollama nomic-embed-text, and use cosine similarity to perform a semantic search on the Video, and link to the video location.


00:58 - Ollama pull nomic-embed-text
03:37 - Use the Whisper base model to transcribe the YouTube audio
05:20 - Use Ollama nomic-embed-text to embed the YouTube transcript
05:35 - Apply lambda embed function to all rows of the dataframe
07:09 - Embed the search term using embed_query of OllamaEmbeddings
07:45 - Apply cosine similarity to all rows of the dataframe
08:08 - Sort the dataframe base on the cosine similarity descending
09:15 - Jump to the Video Location of the YouTube semantic search result


Github:
https://github.com/business24ai/yt_re...

YouTube:
Talk to AutoGen Agents using Whisper and Gradio
   • Talk to AutoGen Agents using Whisper and G...  

Talk to AutoGen Agents using Whisper and Gradio
   • Talk to AutoGen Agents using Whisper and G...  

Transcribe YouTube Videos using Whisper and create links to Video Locations in Obsidian and Excel
   • Transcribe YouTube Videos using Whisper an...  

The Swedish Investor
   / @theswedishinvestor