This Project was made during Hack 36 organised by NIT Allahabad.
Our project basically deals with the the creation of an automatically generated playlist based on the facial expressions of the user as music plays a very important role in human daily life. The difficulties in the creation of large playlists and time required to search a song based on the current mood can be overcome here. This Music Website itself selects songs according to the current mood/facial expressions of the user. This project basically allows user to browse songs of a particular type by depicting the facial expressions of the desired type. Since existing methods for automating the playlist generation process are computationally slow, less accurate and additional hardware like EEG and sensors are required to achieve the task, this proposed approach will use machine learning algorithms to get the information about current facial expressions and hence generate a playlist based on the result.
Since facial expressions are best way to express one's feelings, mood and emotions and their computational algorithms have lesser complexity, it can be one of the best way to decide one's current flavour of music. The introduction of spotify API in the traditional music players provided automatically parsing the playlist based on various classes of emotions and moods which was parsed through 'face_classification' API. 'face_classification' API basically categorises a user's mood in one of the five categories 'Happy','Sad','Angry','Neutral' and 'Surprise' and based on that result a playlist of the songs is generated. Moreover, the results are displayed on a webpage which is generated using Python CGI framework. This project also allows the user to play the desired track just on a click.
To analyse the current mood of the user we used an API 'face_classification' which have implemented deep learning(Tensor Flow) in its methodology and this result was fed as input to an another API 'Spotify'. Python CGI is used to host the project in the form of a website. An input to our project is given by feeding an image clicked through webcam.
