What it does
Speechify takes in an mp3 and returns one that has fitting music attached to it.
How we built it
The user uploads an mp3 file to our front end. The front end then sends it to our server which then converts it to a base64. We then perform sentiment analysis on it, and using the results, construct a tune that matches the words. Afterwards, the clips are concatenated and sent back to the front end for the user to download.
Challenges we ran into
One of the largest challenges we had was actually using the APIs. For the sentiment analysis, the API requires the audio file in base64. We used Python to run our backend, and there used to be a very nice library that converted mp3 or wav files into base64. However, the package doesn't work with Python3, and the one that does has comparatively limited documentation.
Accomplishments that we're proud of
What we learned
A lot! We figured out how to send files through HTTP requests with a server and front end that we wrote completely from scratch!
What's next for Speechify
1) Speed up the algorithm 2) Run the program with text. Use the google-cloud-speech api to convert the text into an mp3 file, and from there run it through Speechify 3) Design a better/more aesthetic front end