arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 1802.08435cs.SDcs.LGeess.AS
收藏

高效的神经音频合成

Efficient Neural Audio Synthesis

DeepMind(深度思维);Google Brain(谷歌大脑)

Nal Kalchbrenner, Erich Elsen, Karen Simonyan, Seb Noury, Norman Casagrande, Edward Lockhart, Florian Stimberg, Aaron van den Oord, Sander Dieleman, Koray Kavukcuoglu

摘要:

序列模型在音频、视觉和文本领域实现了最先进的结果,既可以估计数据分布,又可以生成高质量的样本。然而,这类模型的有效采样仍然是一个难以捉摸的问题。重点放在文本到语音合成,我们描述了一套通用的技术,减少采样时间,同时保持高的输出质量。我们首先描述了一个单层递归神经网络WaveRNN,它有一个双softmax层,与最先进的WaveNet模型的质量相匹配。该网络的紧凑形式使得在GPU上生成24 kHz 16位音频的速度比实时的快4倍。其次,我们应用权重修剪技术来减少WaveRNN中的权重数量。我们发现,对于常数数量的参数,大型稀疏网络比小型密集网络表现更好,这种关系适用于稀疏度超过96%的网络。稀疏波RNN中的少量权重使得在移动CPU上实时的采样高保真音频成为可能。最后,我们提出了一个新的生成方案的基础上subscaling折叠成一批较短的序列,并允许一次生成多个样本的长序列。Subscale WaveRNN每步产生16个样本,而不会损失质量,并提供了一种提高采样效率的正交方法。

英文摘要:

Sequential models achieve state-of-the-art results in audio, visual and textual domains with respect to both estimating the data distribution and generating high-quality samples. Efficient sampling for this class of models has however remained an elusive problem. With a focus on text-to-speech synthesis, we describe a set of general techniques for reducing sampling time while maintaining high output quality. We first describe a single-layer recurrent neural network, the WaveRNN, with a dual softmax layer that matches the quality of the state-of-the-art WaveNet model. The compact form of the network makes it possible to generate 24kHz 16-bit audio 4x faster than real time on a GPU. Second, we apply a weight pruning technique to reduce the number of weights in the WaveRNN. We find that, for a constant number of parameters, large sparse networks perform better than small dense networks and this relationship holds for sparsity levels beyond 96%. The small number of weights in a Sparse WaveRNN makes it possible to sample high-fidelity audio on a mobile CPU in real time. Finally, we propose a new generation scheme based on subscaling that folds a long sequence into a batch of shorter sequences and allows one to generate multiple samples at once. The Subscale WaveRNN produces 16 samples per step without loss of quality and offers an orthogonal method for increasing sampling efficiency.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑