Paper Info
Title
On the training aspects of Deep Neural Network (DNN) for parametric TTS synthesis
Abstract
Deep Neural Network (DNN), which can model a long-span, intricate transform compactly with a deep-layered structure, has recently been investigated for parametric TTS synthesis with a fairly large corpus (33,000 utterances) [6]. In this paper, we examine DNN TTS synthesis with a moderate size corpus of 5 hours, which is more commonly used for parametric TTS training. DNN is used to map input text features into output acoustic features (LSP, F0 and V/U). Experimental results show that DNN can outperform the conventional HMM, which is trained in ML first and then refined by MGE. Both objective and subjective measures indicate that DNN can synthesize speech better than HMM-based baseline. The improvement is mainly on the prosody, i.e., the RMSE of natural and generated F0 trajectories by DNN is improved by 2 Hz. This benefit is likely from the key characteristics of DNN, which can exploit feature correlations, e.g., between F0 and spectrum, without using a more restricted, e.g. diagonal Gaussian probability family. Our experimental results also show: the layer-wise BP pre-training can drive weights to a better starting point than random initialization and result in a more effective DNN; state boundary info is important for training DNN to yield better synthesized speech; and a hyperbolic tangent activation function in DNN hidden layers yields faster convergence than a sigmoidal one.
Year
DOI
Venue
2014
10.1109/ICASSP.2014.6854318
ICASSP
Keywords
Field
DocType
subjective measure,deep neural network,objective measure,tts,acoustic features,hmm,backpropagation,layer-wise bp pretraining,speech synthesis,diagonal gaussian probability family,feature extraction,dnn training,parametric tts synthesis,text features,dnn,sigmoidal function,text-to-speech synthesis,hyperbolic tangent activation function,neural nets,hidden markov model,speech processing,speech,acoustics,decision support systems
Convergence (routing),Pattern recognition,Computer science,Activation function,Speech recognition,Parametric statistics,Gaussian,Artificial intelligence,Initialization,Artificial neural network,Hidden Markov model,Sigmoid function
Conference
ISSN
Citations 
PageRank 
1520-6149
48
1.65
References 
Authors
14
4
Name
Order
Citations
PageRank
Qian Yao152751.55
Yuchen Fan233217.14
Wenping Hu3826.77
Frank K. Soong41395268.29