Inspiration
I make the audio part because I made a text generation project before and I think it would be interesting to use similar methods to generate music. For the image part, it is because I made an image super-resolution before, so I also think a similar method could be used in image Coloring
What it does
It can generate music and color black-and white-image
How we built it
We used TensorFlow and data collected from the internet to train our audio and image model.
Challenges we ran into
We tried to make the front-end for the audio visualization but because we are not familiar with the javascript, we didn't make it according to time. And I ran into a typical issue in TensorFlow because of the model shaping and resizing.
Accomplishments that we're proud of
Most of our team members are first time doing Hackathon, and many of them are just at the beginning of programming. But we still tried our best to make this project.
What we learned
We learned that we could use the midi file for audio training instead of mp3 or wav. This file is easier to read and process.
What's next for Computer AI Audio Composing and Image Coloring
The audio part could be used to generate unlimited music with the same style as background music. The image part could be used to color old movies.
Built With
- google-colab
- python
- tensorflow
Log in or sign up for Devpost to join the conversation.