Unleashing the Melodic Potential: Music Generation with Char RNNs

Recently, the use of ABC notation in music composition employing recurrent neural networks (RNNs) has gained popularity. As a possible tool for ABC-based music creation, our work investigates char RNN models. Due to its simple and clear representation of musical elements, the ABC notation style is used for training RNN models. This study's main objective is to assess the effectiveness and creativity of Character RNN models for generating music. The technical facets of model implementation are examined in this paper, including network design, training approaches, and text preprocessing methods particular to ABC notation. Furthermore, the model's compositions are evaluated for overall coherence, melodic and rhythmic patterns and harmonic progressions. With the aid of this study, we seek to acquire a deeper grasp of the musical creativity potential of Character RNN models and their capability to generate music.