Efficiently Trainable Text-to-Speech System Based on Deep Convolutional Networks with Guided Attention

Abstract

This paper describes a novel text-to-speech (TTS) technique based on deep convolutional neural networks (CNN), without use of any recurrent units. Recurrent neural networks (RNN) have become a standard technique to model sequential data recently, and this technique has been used in some cutting-edge neural TTS techniques. However, training RNN components often requires a very powerful computer, or a very long time, typically several days or weeks. Recent other studies, on the other hand, have shown that CNN-based sequence synthesis can be much faster than RNN-based techniques, because of high parallelizability. The objective of this paper is to show that an alternative neural TTS based only on CNN alleviate these economic costs of training. In our experiment, the proposed Deep Convolutional TTS was sufficiently trained overnight (15 hours), using an ordinary gaming PC equipped with two GPUs, while the quality of the synthesized speech was almost acceptable.

arXiv comment5 pages, 3figures, IEEE ICASSP 2018

References in corpus (13)

• Adam: A Method for Stochastic Optimization
• Neural Machine Translation by Jointly Learning to Align and Translate
• Sequence to Sequence Learning with Neural Networks
• WaveNet: A Generative Model for Raw Audio
• Character-level Convolutional Networks for Text Classification
• Convolutional Sequence to Sequence Learning
• A Neural Conversational Model
• Training Very Deep Networks
• Language Modeling with Gated Convolutional Networks
• Deep Voice: Real-time Neural Text-to-Speech
• Quasi-Recurrent Neural Networks
• Neural Machine Translation in Linear Time
• Deep Voice 2: Multi-Speaker Neural Text-to-Speech

Cited by in corpus (57)

• A Comparative Study on Transformer vs RNN in Speech Applications
• MoGlow: Probabilistic and controllable motion synthesis using normalising flows
• Towards Automatic Face-to-Face Translation
• Sequence-to-Sequence Acoustic Modeling for Voice Conversion
• A Comprehensive Survey on Deep Music Generation: Multi-level Representations, Algorithms, Evaluations, and Future Directions
• Voice Transformer Network: Sequence-to-Sequence Voice Conversion Using Transformer with Text-to-Speech Pretraining
• Listen to Dance: Music-driven choreography generation using Autoregressive Encoder-Decoder Network
• MultiSpeech: Multi-Speaker Text to Speech with Transformer
• ESPnet2-TTS: Extending the Edge of TTS Research
• An Overview of Voice Conversion and its Challenges: From Statistical Modeling to Deep Learning
• High Quality Streaming Speech Synthesis with Low, Sentence-Length-Independent Latency
• FastWave: Accelerating Autoregressive Convolutional Neural Networks on FPGA
• Deep Long Audio Inpainting
• ConvS2S-VC: Fully convolutional sequence-to-sequence voice conversion
• Review of end-to-end speech synthesis technology based on deep learning
• JDI-T: Jointly trained Duration Informed Transformer for Text-To-Speech without Explicit Alignment
• TTS-Portuguese Corpus: a corpus for speech synthesis in Brazilian Portuguese
• VARA-TTS: Non-Autoregressive Text-to-Speech Synthesis based on Very Deep VAE with Residual Attention
• Investigation of Japanese PnG BERT language model in text-to-speech synthesis for pitch accent language
• Online Trajectory Planning Through Combined Trajectory Optimization and Function Approximation: Application to the Exoskeleton Atalante
• Maximizing Mutual Information for Tacotron
• ESPnet-TTS: Unified, Reproducible, and Integratable Open Source End-to-End Text-to-Speech Toolkit
• Learning to Maximize Speech Quality Directly Using MOS Prediction for Neural Text-to-Speech
• Adversarial Feature Learning and Unsupervised Clustering based Speech Synthesis for Found Data with Acoustic and Textual Noise
• SNAC: Speaker-normalized affine coupling layer in flow-based architecture for zero-shot multi-speaker text-to-speech
• Robust Sequence-to-Sequence Acoustic Modeling with Stepwise Monotonic Attention for Neural TTS
• A study of latent monotonic attention variants
• EfficientTTS: An Efficient and High-Quality Text-to-Speech Architecture
• Mitigating Unauthorized Speech Synthesis for Voice Protection
• Hierarchical Sequence to Sequence Voice Conversion with Limited Data
• Initial investigation of an encoder-decoder end-to-end TTS framework using marginalization of monotonic hard latent alignments
• Sequential Skip Prediction with Few-shot in Streamed Music Contents
• Exploring Machine Speech Chain for Domain Adaptation and Few-Shot Speaker Adaptation
• DeepSinger: Singing Voice Synthesis with Data Mined From the Web
• AttS2S-VC: Sequence-to-Sequence Voice Conversion with Attention and Context Preservation Mechanisms
• Neural Sequence-to-Sequence Speech Synthesis Using a Hidden Semi-Markov Model Based Structured Attention Mechanism
• Predicting pairwise preferences between TTS audio stimuli using parallel ratings data and anti-symmetric twin neural networks
• ICE-Talk: an Interface for a Controllable Expressive Talking Machine
• UFANS: U-shaped Fully-Parallel Acoustic Neural Structure For Statistical Parametric Speech Synthesis With 20X Faster
• Karaoker: Alignment-free singing voice synthesis with speech training data
• Exploring Aligned Lyrics-Informed Singing Voice Separation
• Teacher-Student Training for Robust Tacotron-based TTS
• Fast DCTTS: Efficient Deep Convolutional Text-to-Speech
• Accent Estimation of Japanese Words from Their Surfaces and Romanizations for Building Large Vocabulary Accent Dictionaries
• Speaking Speed Control of End-to-End Speech Synthesis using Sentence-Level Conditioning
• Byakto Speech: Real-time long speech synthesis with convolutional neural network: Transfer learning from English to Bangla
• Speaker disentanglement in video-to-speech conversion
• Generic Indic Text-to-speech Synthesisers with Rapid Adaptation in an End-to-end Framework
• FPETS : Fully Parallel End-to-End Text-to-Speech System
• Exploring Teacher-Student Learning Approach for Multi-lingual Speech-to-Intent Classification
• A Neural Text-to-Speech Model Utilizing Broadcast Data Mixed with Background Music
• N-Singer: A Non-Autoregressive Korean Singing Voice Synthesis System for Pronunciation Enhancement
• Improving Time Series Classification Algorithms Using Octave-Convolutional Layers
• Jejueo Datasets for Machine Translation and Speech Synthesis
• Investigating context features hidden in End-to-End TTS
• A Fully Time-domain Neural Model for Subband-based Speech Synthesizer
• Spoofing Speaker Verification Systems with Deep Multi-speaker Text-to-speech Synthesis