[2010.04301] Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modelingopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

This paper presents Non-Attentive Tacotron based on the Tacotron 2 text-to-speech model, replacing the attention mechanism with an explicit duration predictor. This improves robustness significantly as measured by unaligned duration ratio and word deletion rate, two metrics introduced in this paper for large-scale robustness evaluation using a pre-trained speech recognition model. With the use of Gaussian upsampling, Non-Attentive Tacotron achieves a 5-scale mean opinion score for naturalness of 4.41, slightly outperforming Tacotron 2. The duration predictor enables both utterance-wide and per-phoneme control of duration at inference time. When accurate target durations are scarce or unavailable in the training data, we propose a method using a fine-grained variational auto-encoder to train the duration predictor in a semi-supervised or unsupervised manner, with results almost as good as supervised training.

3 mentions: @heiga_zen@ak92501
Date: 2020/10/14 00:53

Referring Tweets

@heiga_zen A paper from my team: Jonathan Shen, Ye Jia, Mike Chrzanowski, Yu Zhang, Isaac Elias, Heiga Zen, Yonghui Wu Non-Attentive Tacotron: Robust & Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling paper: t.co/DicrCmFZRq audio: t.co/QsoDYuFvPm t.co/HjmtsDonAV
@ak92501 Non-Attentive Tacotron: Robust and Controllable Neural TTS Synthesis Including Unsupervised Duration Modeling pdf: t.co/UNlkaBdWIa abs: t.co/9P2hJ2NINZ samples: t.co/LAAYQCwBOH t.co/9pcFKXw0Nc

Related Entries

Read more [2001.07044] JVS-MuSiC: Japanese multispeaker singing-voice corpuscontact arXivarXiv Twitter
0 users, 2 mentions 2020/01/22 02:21
Read more GitHub - lattas/AvatarMe: Public repository for the CVPR paper AvatarMe
0 users, 4 mentions 2020/04/23 18:52
Read more GitHub - mit-han-lab/data-efficient-gans: Differentiable Augmentation for Data-Efficient GAN Trainin...
0 users, 2 mentions 2020/09/23 06:53
Read more [2010.01679] Learning Complete 3D Morphable Face Models from Images and Videosopen searchopen naviga...
0 users, 2 mentions 2020/10/09 14:22
Read more Neural Sparse Voxel Fields
0 users, 3 mentions 2020/10/16 00:53