Human-Mood-Manipulator-using-Speech-Recognition

There are thousands of tracks out there with many different genres and to select a genre close to your current mood is very difficult as a human cannot find a perfect song as per the current mood. This is where our app comes in picture. It will identify the mood of the user using their voice and suggest a good playlist to accompany them in their sad or happy time. We are planning to do is make a music app that will recommend you music based on your mood. Multiple researches proved that music has the power to influence your mood in many ways. Depression is common in this technological advanced age and with music we can reduce or eradicate it completely from a person. CNN model was the best for our classification problem. After training numerous models, we got the best validation accuracy of 77.55% with 18 layers, ReLU activation function, rmsprop activation function, batch size of 16 and 300 epochs. Our model can identify the mood of a human subject with a good accuracy of 77.55% and was able to recommend them playlists from Spotify depending on their predicted moods.