Podcast Summarizer

• 

  Main screen of application.

Inspiration

What it does

For a given RSS Feed of a podcast it will:

• Transcribe the podcast using OpenAI's Whisper Speech to Text
• Summarize the podcast

How I built it

Using Opensource models we connected the downloaded mp3 to perform all the things needed.

Challenges we ran into

Setting up the environment for development was challenging. Understanding how to use and link open source models together was also not easy. Most examples show you how to use one model and not several.

Accomplishments that we're proud of

Rapidly building a proof of concept with working code - all built over a weekend. Learned Python, streamlit, Huggingface, Whisper, everything!

What we learned

• Learned about ML Models and how to interact with it
• Learned about pipelines
• Learned about huggingface
• Learned about open source models, datasets
• Learned how to refine models
• Learned about Python <- I did not know python before this.

What's next for Podcast Summarizer

It would be great to take the transcription, build an embedding and then allow people to ask questions like "Does this episode cover _____ ?", "At what point in the episode do they talk about ________?" (and then be able to jump to that point in the podcast.

Built With

• python
• whisper-ai