Music and Personalized AI assistant bots
It can generate user's content by going through their music playlists, likes and downloads from different cloud based platforms like spotify and youtube music. Llama index should be a good start for building the project on jupytr notebook with python, I am still assembling a team for the project right now. I can start with creating a custom model with custom data to analyze and fine tune into the desired output. In real time, this will require tagging all the music genres, artists, movies, tv shows and albums into their respective emotions and mood they reflect. Moreover, it would also depend on individual choices and how they deal with certain emotions or moods in their lives. So it could take a lot of time customizing and fine tuning chunks of data then implementing RAG over it to come into production. This would require a lot of developers working on different sets of data together and the data from certain companies in compliance with data privacy. The biggest challenge would be data mining and extraction where a lot of policies must be applied in order to get the data needed for this project (compliant with data privacy rules and regulations) The biggest accomplishments from this project would be to be clear of the user's data without actually revealing their information anywhere just the associated email address receives the desirable output for their personalized content. This could potentially be a new feature for amazon alexa. Learning about data science and creating/using a large language model to generate results based on user's search index on different platforms. Moreover, fine tuning the model for more accurate results and applying RAG concepts for pin point accuracy. Suggest new/undiscovered content concurrent with user's preferences. Allowing speech recognition to generate results directly and/or speech to text models. The link provides a generalized scope of the idea in play.
Log in or sign up for Devpost to join the conversation.