Music is becoming increasingly easier to consume by way of apps on the internet and songs. Perhaps the most popular feature of music is the musical form. A complex task in the field is grouping music tracks according to their criteria for the structured arrangement of audio files and for the increasing interest in genre grouping in automated songs. Additionally, a critical aspect of the identification and aggregation of music in related genres is the recommended method for song and album generator. In this project, we adapt the transfer learning techniques to train a custom music genre classification system with customized genres and data. The model takes as an input the spectrogram/sonogram of music frames and analyzes the image using a Convolutional Neural Network (CNN) plus a Recurrent Neural Network (RNN). The output of the system is a vector of predicted genres providing maximum accuracy. This system will be useful to predict or analyze the mood or character based on music preferences which can help to cure depression, anxiety and stress.
Key words: Transfer learning, Multi-Framing,CRNN
|
