Inspiration
Our inspiration comes from the lack of resources for movie actors and actresses. Most skills are learnable through watching videos on the internet, but a good grasp of any skill requires thorough practice. This is especially hard for acting
What it does
Starprep will ask the user to select elements that will be included in a generated story. After the user has submitted their story elements, GPT-4o will generate a script for the user to read aloud. We record their microphone audio, then it will send the audio to the flask server which will make a request to Hume. Hume sends back the emotion results and we present it to the user.
How we built it
We first designed a mockup in Figma. We then implemented Starprep using the Flask Python web server, React framework, and Hume's API.
Challenges we ran into
Hume could not read the emotions from ordinary text. It relies heavily on prosody and human speech to identify emotions which makes sense. This changed the original concept of using the generated text as a ground truth for grading the voice actor / actress.
Accomplishments that we're proud of
We are proud of being able to use Hume's API. It has this emotion detection feature that is super cool.
What we learned
We learned about the usefulness and potential of Hume's product.
What's next for Starprep
Improvements to the user interface.
Built With
- flask
- gpt
- hume
- javascript
- python
- react
Log in or sign up for Devpost to join the conversation.