Maximizing Mutual Information for Tacotron

Abstract

End-to-end speech synthesis methods already achieve close-to-human quality performance. However compared to HMM-based and NN-based frame-to-frame regression methods, they are prone to some synthesis errors, such as missing or repeating words and incomplete synthesis. We attribute the comparatively high utterance error rate to the local information preference of conditional autoregressive models, and the ill-posed training objective of the model, which describes mostly the training status of the autoregressive module, but rarely that of the condition module. Inspired by InfoGAN, we propose to maximize the mutual information between the text condition and the predicted acoustic features to strengthen the dependency between them for CAR speech synthesis model, which would alleviate the local information preference issue and reduce the utterance error rate. The training objective of maximizing mutual information can be considered as a metric of the dependency between the autoregressive module and the condition module. Experiment results show that our method can reduce the utterance error rate.

References in corpus (11)

• Auto-Encoding Variational Bayes
• On the difficulty of training Recurrent Neural Networks
• WaveNet: A Generative Model for Raw Audio
• Neural Discrete Representation Learning
• Sequence Transduction with Recurrent Neural Networks
• Style Tokens: Unsupervised Style Modeling, Control and Transfer in End-to-End Speech Synthesis
• InfoVAE: Information Maximizing Variational Autoencoders
• Efficiently Trainable Text-to-Speech System Based on Deep Convolutional Networks with Guided Attention
• Variational Lossy Autoencoder
• PixelVAE: A Latent Variable Model for Natural Images
• A New GAN-based End-to-End TTS Training Algorithm

Cited by in corpus (5)

• A Survey on Neural Speech Synthesis
• 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
• Speaker disentanglement in video-to-speech conversion
• Nana-HDR: A Non-attentive Non-autoregressive Hybrid Model for TTS