BriefGPT.xyz
Dec, 2016
SampleRNN: 一种无条件端到端的神经音频生成模型
SampleRNN: An Unconditional End-to-End Neural Audio Generation Model
HTML
PDF
Soroush Mehri, Kundan Kumar, Ishaan Gulrajani, Rithesh Kumar, Shubham Jain...
TL;DR
本文提出了一种新的无条件音频生成模型，该模型利用自回归多层感知机和有状态循环神经网络的分层结构来捕捉长时间跨度中时间序列的潜在变化源，并在不同数据集上进行人类评估，结果表明该模型优于竞争模型。同时还展示了模型的各个组件对展示性能的贡献。
Abstract
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 m
→