SpecDiff-GAN：一种用于语音和音乐合成的频谱成形噪声扩散GAN
SpecDiff-GAN: A Spectrally-Shaped Noise Diffusion GAN for Speech and Music Synthesis
AI总结：
本文提出SpecDiff-GAN，一种基于HiFi-GAN的神经声码器，通过向判别器输入前添加高斯噪声及频谱成形噪声的扩散过程增强训练稳定性，在语音和音乐合成上实现更优的音频质量与效率。
摘要：
生成对抗网络（GAN）模型能够合成高质量的音频信号，同时确保快速的样本生成。然而，它们难以训练，并且容易出现包括模式崩溃和发散在内的多个问题。在本文中，我们介绍了SpecDiff-GAN，一种基于HiFi-GAN的神经声码器，HiFi-GAN最初是为从梅尔频谱图进行语音合成而设计的。在我们的模型中，通过前向扩散过程增强了训练稳定性，该过程包括在将真实样本和虚假样本输入判别器之前，向两者注入来自高斯分布的噪声。我们进一步通过利用频谱成形的噪声分布来改进模型，旨在使判别器的任务更具挑战性。然后，我们在多个数据集上展示了我们提出的模型在语音和音乐合成方面的优点。我们的实验证实，与几种基线相比，我们的模型在音频质量和效率方面表现更优。
英文摘要：
Generative adversarial network (GAN) models can synthesize highquality audio signals while ensuring fast sample generation. However, they are difficult to train and are prone to several issues including mode collapse and divergence. In this paper, we introduce SpecDiff-GAN, a neural vocoder based on HiFi-GAN, which was initially devised for speech synthesis from mel spectrogram. In our model, the training stability is enhanced by means of a forward diffusion process which consists in injecting noise from a Gaussian distribution to both real and fake samples before inputting them to the discriminator. We further improve the model by exploiting a spectrally-shaped noise distribution with the aim to make the discriminator's task more challenging. We then show the merits of our proposed model for speech and music synthesis on several datasets. Our experiments confirm that our model compares favorably in audio quality and efficiency compared to several baselines.
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。
