RHYTHMIX – LSTM Based Music Synthesizer

The paper presents an extensive survey of Long Short-Term Memory networks in music generation, covering their application to melody creation, Harmonic progression, rhythm generation,and multi-track music. The latest developments in neural music models are examined. The survey explores difficulties like handling long-term musical dependency and improving the structural coherence of compositions. Possible future directions for combining LSTMs with other neural architectures to advance the quality and complexity of music is discussed.