reci.tech

Helping actors recite scripts expressively w the help of AI

Deployed at https://recitech.herokuapp.com

Devpost: https://devpost.com/software/reci-tech-script-recital-aid

Demo video: https://youtu.be/vwbAvUHnfUo

Video about May Update: https://youtu.be/mVjfRPztskQ

Inspiration

The sentiments in our speech often get misinterpreted, especially during the pandemic when everyone covers their faces. We figured if that sentiment analysis AIs trained on daily sentences resemble how people understand emotions, we may adjust how we voice our sentences to what the AI thinks it is to convey our expressions to the most general audience.

That can also help actors, who need expressiveness the most, to communicate to the most lay-people understandably.

What it does

It takes a line and prompts the user to voice it. It then analyses the sentiments in both the text and the voice, and compares the two.

How we built it

Flask back-end and Bootstrap front-end. The AI is a pre-trained TensorFlow model.

Challenges we ran into

It is hard to coordinate a team of mostly newbie hackers scattered around 4 time-zones, and we have to mostly work on our own in different hours of the day.

Accomplishments that we're proud of

4 of the 5 of us are new to hackathons, but we made it work!

What we learned

Tasks always take pi times more time than expected. We planned to learn and use React on the fly, but realised it is too complex and built a minimum-viable product (MVP) in Bootstrap instead.

What's next for Reci.tech: Script Recital Aid

AIs surely lack artistic nuances, so we may feed it lines spoken by the best actors and show whether the user conveys similar emotional complexity like the best do.

23 May Update

Continued for Mental Health Hack

Completion

1. Deployed to Heroku & an external domain (reci.tech)

   Completion Does the hack work? Did the team achieve everything they wanted? Learning Did the team stretch themselves? Did they try to learn something new? What kind of projects have they worked on before? Design Did the team put thought into the user experience? How well designed is the interface? Technology How technically impressive was the hack? Was the technical problem the team tackled difficult? Did it use a particularly clever technique or did it use many different components? Did the technology involved make you go "Wow"?

Running the app

Ensure Python version is ~3.8

python -V
cat runtime.txt

The outputs should match

Create a virtual environment

python -m venv venv
. venv/bin/activate

Install dependencies

pip install -r requirements.txt

Run flask

export FLASK_APP=run.py
python -m flask run