Soundaliser

Inspiration

My phone screen broke so I would have wished I could enjoy memes without seeing them.

What it does

Gets an imagine, identifies what objects it contains through the magic of AI, and then outputs the sound the objects in the imagine would have made at the time the photo was taken.

How we built it

We borrowed a pretrained Convolutional Neural Network and used it to identify the object in the image, thus getting an array of the object. Then we recorded sounds for every category of objects we had. When the program is run, you have to insert an image, and you will get back the set of sounds that image represents.

Challenges we ran into

AI is hard to write. Also we had to manually record the sounds for every single object, which took a lot of time.

Accomplishments that we're proud of

The fact we got it working, and that we have 13 samples of sheep sounds from attendees and volunteers (including Tom Goodman as a sheep)

What we learned

That hackathons are really fun.

What's next for Soundaliser

This can actually become a real application idea, as the concept doesn't have to be applied just to photos, this can also be used to make audio descriptions of movies.

Built With

• python