AI-Based Twitter Podcast Generator
Budget: $30 – $250 USD
A general description of the desired functionality is below. It involves three primary parts
- single page website
- twitter api integration
- hyperfeeder plugin implementation
I will only consider applications that have reviewed the hyperfeeder repo and and provide an overview of how they will develop the hyperfeeder plugin.
https://github.com/normand1/HyperFeeder
A new user will visit a URL.
The URL is a single-page website with a simple and responsive design.
They will click "connect my twitter account”
From here, they will oauth (probably) their twitter account - or whatever we need them to do in order to access the API for/as them
User should input a name for their podcast and get back an RSS feed URL that is generated for this user’s “podcast”. This will basically be for a directory where we store their episodes.
After this, they system should contact the twitter api and request the last 10 bookmarks. (if this is a returning users, It should compare to a list of bookmarks that have already been used to create podcast stories in previous runs. It should only use the most recent 10 unused bookmarks)
The system will then gather the top 50 comments for each bookmark.
1 bookmark, its comments, and all metadata associated with these posts will be fed into hyperfeeder, where it is summarized by an LLM, in order to create a hyper feeder “story” (some prompt engineering will be needed here)
This will be done for each of the 10 bookmarks.
When this is complete, the hyper feeder pipeline should run, to create an audio file for the stories.
Following this, the outputted audio file should be saved as a podcast episode with metadata to the user’s podcast directory.
- single page website
- twitter api integration
- hyperfeeder plugin implementation
I will only consider applications that have reviewed the hyperfeeder repo and and provide an overview of how they will develop the hyperfeeder plugin.
https://github.com/normand1/HyperFeeder
A new user will visit a URL.
The URL is a single-page website with a simple and responsive design.
They will click "connect my twitter account”
From here, they will oauth (probably) their twitter account - or whatever we need them to do in order to access the API for/as them
User should input a name for their podcast and get back an RSS feed URL that is generated for this user’s “podcast”. This will basically be for a directory where we store their episodes.
After this, they system should contact the twitter api and request the last 10 bookmarks. (if this is a returning users, It should compare to a list of bookmarks that have already been used to create podcast stories in previous runs. It should only use the most recent 10 unused bookmarks)
The system will then gather the top 50 comments for each bookmark.
1 bookmark, its comments, and all metadata associated with these posts will be fed into hyperfeeder, where it is summarized by an LLM, in order to create a hyper feeder “story” (some prompt engineering will be needed here)
This will be done for each of the 10 bookmarks.
When this is complete, the hyper feeder pipeline should run, to create an audio file for the stories.
Following this, the outputted audio file should be saved as a podcast episode with metadata to the user’s podcast directory.