This paper demonstrates how Long Short-term Memory (LSTM) recurrent neural networks can be used to generate complex sequences with long-range structure by predicting one data point at a time. The approach is applied to text and online handwriting, and extended to handwriting synthesis by conditioning predictions on a text sequence. The resulting system can generate highly realistic cursive handwriting in various styles. The paper includes detailed descriptions of the LSTM architecture, training methods, and experimental results on the Penn Treebank and Wikipedia datasets, as well as the IAM Online Handwriting Database. The generated samples showcase the network's ability to model long-range dependencies and learn complex structures from the data.This paper demonstrates how Long Short-term Memory (LSTM) recurrent neural networks can be used to generate complex sequences with long-range structure by predicting one data point at a time. The approach is applied to text and online handwriting, and extended to handwriting synthesis by conditioning predictions on a text sequence. The resulting system can generate highly realistic cursive handwriting in various styles. The paper includes detailed descriptions of the LSTM architecture, training methods, and experimental results on the Penn Treebank and Wikipedia datasets, as well as the IAM Online Handwriting Database. The generated samples showcase the network's ability to model long-range dependencies and learn complex structures from the data.