No abstract

Expressive synthesis from text is a challenging problem. There are two issues. First, read text is often highly expressive to convey the emotion and scenario in the text. Second, since the expressive training speech is not always available for different speakers, it is necessary to develop methods to share the expressive information over speakers. This paper investigates the approach of using very expressive, highly diverse audiobook data from multiple speakers to build an expressive speech synthesis system. Both of two problems are addressed by considering a factorized framework where speaker and emotion are modeled in separate sub-spaces of a cluster adaptive training (CAT) parametric speech synthesis system. The sub-spaces for the expressive state of a speaker and the characteristics of the speaker are jointly trained using a set of audiobooks. In this work, the expressive speech synthesis system works in two distinct modes. In the first mode, the expressive information is given by audio data and the adaptation method is used to extract the expressive information in the audio data. In the second mode, the input of the synthesis system is plain text and a full expressive synthesis system is examined where the expressive state is predicted from the text. In both modes, the expressive information is shared and transplanted over different speakers. Experimental results show that in both modes, the expressive speech synthesis method proposed in this work significantly improves the expressiveness of the synthetic speech for different speakers. Finally, this paper also examines whether it is possible to predict the expressive states from text for multiple speakers using a single model, or whether the prediction process needs to be speaker specific.Index Terms-Audiobook, cluster adaptive training, expressive speech synthesis, factorization, hidden Markov model, neural network.

No abstract

No abstract

In this paper we investigate the use of noise-robust features characterizing the speech excitation signal as complementary features to the usually considered vocal tract based features for Automatic Speech Recognition (ASR). The proposed Excitation-based Features (EBF) are tested in a state-of-theart Deep Neural Network (DNN) based hybrid acoustic model for speech recognition. The suggested excitation features expand the set of periodicity features previously considered for ASR, expecting that these features help in a better discrimination of the broad phonetic classes (e.g., fricatives, nasal, vowels, etc.). Our experiments on the AMI meeting transcription system showed that the proposed EBF yield a relative word error rate reduction of about 5% when combined with conventional PLP features. Further experiments led on Aurora4 confirmed the robustness of the EBF to both additive and convolutive noises, with a relative improvement of 4.3% obtained by combinining them with mel filter banks.

No abstract

No abstract

This work aims at bootstrapping acoustic model training for automatic speech recognition with small amounts of humanlabeled speech data and large amounts of machine-labeled speech data.Semi-supervised learning is investigated to select the machine-transcribed training samples.Two semi-supervised learning methods are proposed: one is the local-global uncertainty based method which introduces both the local uncertainty from the current utterance and the global uncertainty from the whole data pool into the data selection; the other is the margin based data selection, which selects the utterances near to the decision boundary through language model tuning. The experimental results based on a Japanese far-field automatic speech recognition system indicate that the acoustic model trained by automatically transcribed speech data achieve about 17% relative gain when in-domain human annotated data was not available for initialization. While 3.7% relative gain was obtained when the initial acoustic model was trained by small amount of in-domain data.