Automatic tone recognition is useful for Mandarin spoken language processing. However, the complex F0 variations from the tone co-articulations and the interplay effects among tonality make it rather difficult to perform tone recognition of Chinese continuous speech. This paper explored the application of Bidirectional Long Short-Term Memory (BLSTM), which had the capability of modeling time series, to Mandarin tone recognition to handle the tone variations in continuous speech. In addition, we introduced attention mechanism to guide the model to select the suitable context information. The experimental results showed that the performance of proposed CNN-BLSTM with attention mechanism was the best and it achieved the tone error rate (TER) of 9.30% with a 17.6% relative error reduction from the DNN baseline system with TER of 11.28%. It demonstrated that our proposed model was more effective to handle the complex F0 variations than other models.
Tonal variations in continuous speech are complicated in nature and it is a challenge to identify the effect of tonal coarticulation given several influencing factors. To address the issue, the present study proposes a scheme for labeling tonal coarticulation in Mandarin continuous speech by applying Hypo-and Hyper-articulation theory. We assume that the bidirectional tonal coarticulation (both carryover and anticipatory effects) as patterns of Hypo-articulation results from the economical articulatory rule. The effects may partially disappear under the influence of specific stress patterns and become unidirectional (carryover or anticipatory). At a prosodic boundary, the effects of tonal coarticulation may completely disappear and lead to the occurrence of patterns of Hyper-articulation. Based on the scheme, we have labeled the data in the Annotated Speech Corpus of Chinese Discourse. It is shown that: three annotators are consistent at a fairly high level (86.2%) on average, and the acoustic parameters of four kinds of tonal coarticulation are significantly different. Therefore, we conclude that the proposal is feasible for investigating tonal coarticulation in Mandarin continuous speech. Though the labeling scheme is language dependent, it may well have cross-linguistic implications.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.