An unsupervised joint prosody labeling and modeling method for Mandarin speech is proposed, a new scheme intended to construct statistical prosodic models and to label prosodic tags consistently for Mandarin speech. Two types of prosodic tags are determined by four prosodic models designed to illustrate the hierarchy of Mandarin prosody: the break of a syllable juncture to demarcate prosodic constituents and the prosodic state to represent any prosodic domain's pitch-level variation resulting from its upper-layered prosodic constituents' influences. The performance of the proposed method was evaluated using an unlabeled read-speech corpus articulated by an experienced female announcer. Experimental results showed that the estimated parameters of the four prosodic models were able to explore and describe the structures and patterns of Mandarin prosody. Besides, certain corresponding relationships between the break indices labeled and the associated words were found, and manifested the connections between prosodic and linguistic parameters, a finding further verifying the capability of the method presented. Finally, a quantitative comparison in labeling results between the proposed method and human labelers indicated that the former was more consistent and discriminative than the latter in prosodic feature distributions, a merit of the method developed here on the applications of prosody modeling.

No abstract

In this paper, a simple recurrent neural network (SRNN) is employed to model the prosody of continuous Mandarin speech to assist tone recognition. For each syllable in continuous speech, several acoustic features carrying prosodic information are extracted and taken as inputs to the SRNN. If proper linguistic features extracted from the context of the syllable are set as output targets, the SRNN can learn to represent the prosodic state of the utterance at the syllable using its hidden nodes. Outputs of the hidden nodes then serve as additional recognition features to assist recognition of the tone of the syllable. The performance of the proposed tone recognition approach was examined by simulation on a multilayer perception (MLP)-based speaker-dependent tone recognition task. The recognition rate was improved from 91.38% to 93.10%. The SRNN prosodic model is further analyzed to exploit the linguistic meaning of prosodic states. By vector quantizing the outputs of the hidden nodes of the SRNN, a finite-state automata that roughly represents the mechanism of human prosody pronunciation can be obtained.

A new RNN-based prosodic information synthesizer for Mandarin Chinese text-to-speech (TTS) is proposed in this paper. Its four-layer recurrent neural network (RNN) generates prosodic information such as syllable pitch contours, syllable energy levels, syllable initial and final durations, as well as intersyllable pause durations. The input layer and first hidden layer operate with a word-synchronized clock to represent currentword phonologic states within the prosodic structure of text to be synthesized. The second hidden layer and output layer operate on a syllable-synchronized clock and use outputs from the preceding layers, along with additional syllable-level inputs fed directly to the second hidden layer, to generate desired prosodic parameters. The RNN was trained on a large set of actual utterances accompanied by associated texts, and can automatically learn many human-prosody phonologic rules, including the wellknown Sandhi Tone 3 F0-change rule. Experimental results show that all synthesized prosodic parameter sequences matched quite well with their original counterparts, and a pitch-synchronousoverlap-add-based (PSOLA-based) Mandarin TTS system was also used for testing of our approach. While subjective tests are difficult to perform and remain to be done in the future, we have carried out informal listening tests by a significant number of native Chinese speakers and the results confirmed that all synthesized speech sounded quite natural.

In order to detect Chinese spelling errors, especially for essays written by foreign learners, a word vector/conditional random field (CRF)based detector is proposed in this paper. The main idea is to project each word in a test sentence into a high dimensional vector space in order to reveal and examine their relationships by using a CRF. The results are then utilized to constrain the time-consuming language model rescoring procedure. Official SIGHAN-2015 evaluation results show that our system did achieve reasonable performance with about 0.601/0.564 accuracies and 0.457/0.375 F1 scores in the detection/correction levels.