Rhythm

• 

  Interface for interaction

Inspiration

The inspiration came from the fact that all of the teammates took public speaking courses in their first semester and felt enormous benefit. There, in terms of self improvement we wanted to bring what we learned in the class to people through AI.

What it does

Rhythm allows a user to record a speech and after they submit, Rhythm will judge it on various parameters like pitch, volume, pace, misspoken words and stuttering.

How we built it

We worked and are using 6 AI models (Wav2Vec, classification models). Separate models for each parameter and functionality. We used react as the front end. Python was used in the backend in order to finetune models. We have used Google, ChatGPT and 11ElevenLab's APIs. Flask was used in order to bring together react and python models.

Challenges we ran into

We had little to no knowledge of AI. It was overwhelming to train so many models and making them execute in synchronization was quite difficult.

Accomplishments that we're proud of

Getting done with so many models in such a short time with such a limited knowledge. Having an appealing UI/UX while also having great functionality.

What we learned

Try to plan how AI models with interact with each other. We learned a lot about cors. 3D vector graphics for the UX. ## What's next for Rhythm Factoring in non verbal communication like body language and eye contact. Also, we wish to break down the speech in detail and increase our model's efficiency on where tone and pitch should be changed.

Built With

• flask
• google-web-speech-api
• javascript
• python
• react