Non-Autoregressive Neural Text-to-Speech
arXiv:1905.08459
Abstract
In this work, we propose ParaNet, a non-autoregressive seq2seq model that converts text to spectrogram. It is fully convolutional and brings 46.7 times speed-up over the lightweight Deep Voice 3 at synthesis, while obtaining reasonably good speech quality. ParaNet also produces stable alignment between text and speech on the challenging test sentences by iteratively improving the attention in a layer-by-layer manner. Furthermore, we build the parallel text-to-speech system and test various parallel neural vocoders, which can synthesize speech from text through a single feed-forward pass. We also explore a novel VAE-based approach to train the inverse autoregressive flow (IAF) based parallel vocoder from scratch, which avoids the need for distillation from a separately trained WaveNet as previous work.
arXiv commentPublished at ICML 2020. (v3 changed paper title)
Cited by in corpus (14)
- Glow-TTS: A Generative Flow for Text-to-Speech via Monotonic Alignment Search
- A Survey on Neural Speech Synthesis
- DiffWave: A Versatile Diffusion Model for Audio Synthesis
- AdaSpeech: Adaptive Text to Speech for Custom Voice
- NU-Wave: A Diffusion Probabilistic Model for Neural Audio Upsampling
- WaveGrad: Estimating Gradients for Waveform Generation
- Text-Conditioned Transformer for Automatic Pronunciation Error Detection
- Diff-TTS: A Denoising Diffusion Model for Text-to-Speech
- PortaSpeech: Portable and High-Quality Generative Text-to-Speech
- STYLER: Style Factor Modeling with Rapidity and Robustness via Speech Decomposition for Expressive and Controllable Neural Text to Speech
- MelGlow: Efficient Waveform Generative Network Based on Location-Variable Convolution
- NAST: A Non-Autoregressive Generator with Word Alignment for Unsupervised Text Style Transfer
- VAENAR-TTS: Variational Auto-Encoder based Non-AutoRegressive Text-to-Speech Synthesis
- An Improved Single Step Non-autoregressive Transformer for Automatic Speech Recognition