SampleRNN: An Unconditional End-to-End Neural Audio Generation Model
This addresses audio generation for applications like music or speech synthesis, but it appears incremental as it builds on existing neural methods.
The paper tackles unconditional audio generation by generating one sample at a time, using a hierarchical model combining autoregressive multilayer perceptrons and recurrent neural networks to capture long-term variations, and human evaluation shows it is preferred over competing models.
In this paper we propose a novel model for unconditional audio generation based on generating one audio sample at a time. We show that our model, which profits from combining memory-less modules, namely autoregressive multilayer perceptrons, and stateful recurrent neural networks in a hierarchical structure is able to capture underlying sources of variations in the temporal sequences over very long time spans, on three datasets of different nature. Human evaluation on the generated samples indicate that our model is preferred over competing models. We also show how each component of the model contributes to the exhibited performance.