I always wanted to deduct human pyschology considering his/her daily activities and provide insights to him.

As we are completely dependent on the internet furled by data, it is highly controlling us.Therefore provide correct insights can make us not to fall prey to the tech that we build ourselves.

What it does

This project is about hosting an endpoint that takes lyrics and conversations as input and classifies emotions. This can be integrated with Spotify to read users recent played songs and let user know the overall

How we built it

It's built on a NLP model that takes the lyrics and tokenises the words, feeds to ML models pipelines ,fit and transform and classifies the input.(mostly lyrics)

Challenges we ran into

Took time to tune and customize as it was worked upon an NLP ML jupyter notebook .

Accomplishments that we're proud of

The model predictions were mostly correct with capabilities of multi label classifications and the development was done at very low cost(0$ USD) on top of AWS Graviton2 ARM Instacnes.

What we learned

We learning how we can train a model and artifact a model and serve it as an endpoint and find way to enchance it to other input formats.

What's next for psychoReader

Next up would be the sentimental and mood analysis of blogs, articles and to the next of it would be image and video format sentimental analysis.

Share this project: