An Innovative Prosody Modeling Method for Chinese Speech Recognition

Peng, Gang; Wang, William Yang

doi:10.1023/b:ijst.0000017013.70486.51

Cited by 5 publications

(1 citation statement)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the contrary, in the explicit tone recognition, tone patterns are independently modeled and recognized in parallel to phonetic recognition. In combination with the spectral scores, the tone recognition scores can be directly integrated in the first pass search process [7], or used to rescore the N-best recognition results [5] or word lattices [6] of the phonetic recognizers. Tonal information also provides phonological and lexical constraints, and therefore the N-best output of tone recognition can be utilized to expand a toneless syllable lattice to a tonal syllable lattice [2].…”

Section: Introductionmentioning

confidence: 99%

Exploiting prosodic and lexical features for tone modeling in a conditional random field framework

Wei

Wang

et al. 2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

Tonal cues play an important role in distinguishing ambiguous words in Mandarin speech recognition. This paper explores an innovative tone modeling framework using prosodic and lexical features, as well as syllable context information. A discriminative model, namely a Conditional Random Field (CRF), is adopted, which is sufficiently flexible to handle multiple interacting features and long-range dependencies of observations. After the first pass search of a recognition system, the CRF based tone models are employed to rerank N-best hypotheses according to the tonal scores which can represent the correctness of the tone sequence given each candidate hypothesis and the observed speech signal. Experiments results show that the tonal cues help to achieve 7.8% and 8.6% relative reductions of character error rate on two widely used Mandarin speech recognition tasks, Hub-4 test and 863 test.

show abstract

Section: Introductionmentioning

confidence: 99%

Exploiting prosodic and lexical features for tone modeling in a conditional random field framework

Wei

Wang

et al. 2008

2008 IEEE International Conference on Acoustics, Speech and Signal Processing

View full text Add to dashboard Cite

show abstract

Improving Mandarin Tone Recognition Based on DNN by Combining Acoustic and Articulatory Features Using Extended Recognition Networks

Lin

Gao

et al. 2018

J Sign Process Syst

View full text Add to dashboard Cite

The networks of syllables and characters in Chinese∗

Peng

Minett

Wang

2008

Journal of Quantitative Linguistics

Self Cite

View full text Add to dashboard Cite

We develop networks using the syllables (both base syllables and tonal syllables) and characters of Chinese. The nodes (vertices) of the networks represent the syllables of the syllable network and the characters of the character network respectively. The links (edges) are established between any two syllables (or two characters) that form part of one or more words. We use two dictionaries to perform the analysis: a Putonghua 1 dictionary and a Cantonese dictionary. All networks here show low distances and high clustering coefficients compared with ER random networks. The degree distributions all follow a power-law; however, the exponents for the base syllable, tonal syllable and Chinese character networks differ considerably. These differences may account for the different cognitive processes used when constructing new Chinese words. The networks are compared to the syllabic networks of Portuguese in terms of the magnitude of the power-law exponent. The Chinese character network is found to be the most similar to the Portuguese syllabic network (g % 1.4).

show abstract

An Innovative Prosody Modeling Method for Chinese Speech Recognition

Cited by 5 publications

References 22 publications

Exploiting prosodic and lexical features for tone modeling in a conditional random field framework

Exploiting prosodic and lexical features for tone modeling in a conditional random field framework

Improving Mandarin Tone Recognition Based on DNN by Combining Acoustic and Articulatory Features Using Extended Recognition Networks

The networks of syllables and characters in Chinese∗

Contact Info

Product

Resources

About