SPEECH EMOTIONAL RECOGNITION

model accuracy
model loss

Inspiration

A lot of things happening on earth but everyone is having their own problem for that we hear music which help better but there is tiny problem the music which we hear is not suggested by our emotion hence we need a person to analyse it but not anymore here is our program to solve this problem

What it does

it uses RAVDESS dataset to create a model using neural network once the model is trained we can classify the live audio and play accordingly the emotion is. we have 0.89p of the training data which is more enough to classify emotions

How we built it

we built with RAVDESS and TESS dataset RAVDESS https://zenodo.org/record/1188976#.XsAXemgzaUk TESS https://tspace.library.utoronto.ca/handle/1807/24487 here it has 8 classes which will give the emotions (0 = neutral, 1 = calm, 2 = happy, 3 = sad, 4 = angry, 5 = fearful, 6 = disgust, 7 = surprised) 03-01-01-01-01-02-05.wav is an example of WRONG prediction other audio is for testing the model

Challenges we ran into

The RAVDESS contains 7356 files. Each file was rated 10 times on emotional validity, intensity, and genuineness. training of huge data takes time and we uses model from the pre trained dataset for time being. the audio processing takes time and also due the system requirement we try to run in colab and replit so it takes lot of time to process it.