This is a skeleton repository indicating the directory structure for the code submission.
Refer to the challenge here - https://www.hackerearth.com/sprints/hack-the-talk/
The submission must be in a tar.gz format. It should have a run.sh in the top directory that will take care of compilation, creating the executable and running your code. Your code must read an input file - input.txt. The Input File will have a list of audio files like:
- one.mp3
- two.mp3
- three.mp3
- four.mp3... and So on.
These files one.mp3, two.mp3 etc will be files in the current directory. Your program would need to output the emotion detected for each file (one.mp3, two.mp3 etc) as happy, unhappy, angry, or neutral in separate lines:
happy
unhappy
neutral
neutral
The above output means one.mp3 was a "happy" conversation, two.mp3 was a "unhappy" conversation etc.
This repo also contains the training data in training_dataset directoy.