Large-Vocabulary Continuous Speech Recognition of Lhasa Tibetan

Li, Guan Yu; Yu, Hong

doi:10.4028/www.scientific.net/amm.519-520.800

Cited by 3 publications

(3 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, the written characters are unified for all Tibetan dialects. Since Lhasa of Ü-Tsang dialect is Tibetan standard speech, there are much more research works than other dialects on linguistics, speech recognition and corpus [Zhang (2016); Yuan, Guo and Dai (2015); Pei (2009); Li and Meng (2012); Wang, Guo and Xie (2017); Cai and Zhao (2008); Cai (2009); Han and Yu (2010)]. Dialect identification has recently gained substantial interest in the field of language identification.…”

Section: Introductionmentioning

confidence: 99%

Tibetan Multi-dialect Speech and Dialect Identity Recognition

Zhao¹,

Yue²,

Song³

et al. 2019

Computers, Materials &Amp; Continua

View full text Add to dashboard Cite

Tibetan language has very limited resource for conventional automatic speech recognition so far. It lacks of enough data, sub-word unit, lexicons and word inventories for some dialects. And speech content recognition and dialect classification have been treated as two independent tasks and modeled respectively in most prior works. But the two tasks are highly correlated. In this paper, we present a multi-task WaveNet model to perform simultaneous Tibetan multi-dialect speech recognition and dialect identification. It avoids processing the pronunciation dictionary and word segmentation for new dialects, while, in the meantime, allows training speech recognition and dialect identification in a single model. The experimental results show our method can simultaneously recognize speech content for different Tibetan dialects and identify the dialect with high accuracy using a unified model. The dialect information used in output for training can improve multi-dialect speech recognition accuracy, and the low-resource dialects got higher speech content recognition rate and dialect classification accuracy by multi-dialect and multi-task recognition model than task-specific models.

show abstract

Section: Introductionmentioning

confidence: 99%

Tibetan Multi-dialect Speech and Dialect Identity Recognition

Zhao¹,

Yue²,

Song³

et al. 2019

Computers, Materials &Amp; Continua

View full text Add to dashboard Cite

show abstract

“…The pronunciation of Tibetan dialect is very different in different regions, but the written characters based on Tibetan language is unified. Since Lhasa-Ü-Tsang dialect is a Tibetan standard speech, there are much more research works than other dialects on its linguistics, speech recognition and corpus establishment [Zhang (2016); Yuan, Guo and Dai (2015); Li and Meng (2012); Cai and Zhao (2008); Cai (2009)]. End-to-end automatic speech recognition has more advantages for low-resource languages than conventional DNN/HMM systems because it avoids the need for linguistic resources such as dictionaries and phonetic knowledge [Wang, Guo and Xie (2017)].…”

Section: Introductionmentioning

confidence: 99%

Tibetan Multi-Dialect Speech Recognition Using Latent Regression Bayesian Network and End-To-End Mode

Zhao

Yue

Song

et al. 2019

Journal on Internet of Things

View full text Add to dashboard Cite

We proposed a method using latent regression Bayesian network (LRBN) to extract the shared speech feature for the input of end-to-end speech recognition model. The structure of LRBN is compact and its parameter learning is fast. Compared with Convolutional Neural Network, it has a simpler and understood structure and less parameters to learn. Experimental results show that the advantage of hybrid LRBN/Bidirectional Long Short-Term Memory-Connectionist Temporal Classification architecture for Tibetan multi-dialect speech recognition, and demonstrate the LRBN is helpful to differentiate among multiple language speech sets.

show abstract

“…In the study of minority speech recognition of speakerdependent, Li established acoustic model based on phonemes and semi-syllables and trained large-vocabulary continuous speech of Lhasa Tibetan on HTK platform, that the recognition rate is 92.2% [6]. Yao has improved the traditional method of endpoint detection, and the method increased the isolated-word speech recognition rate of Tibetan specific [7].…”

Section: Introductionmentioning

confidence: 99%

The Analysis of Influence Factors and Identification of Speaker-Dependent Primi Speech Recognition

Guo¹,

Bai²,

Su³

et al. 2016

Proceedings of the 2016 International Conference on Artificial Intelligence: Technologies and Applications

View full text Add to dashboard Cite

Abstract-The research object is an endangered minority language of Primi on the southern dialect. This paper takes isolated-word of Primi as the primitive and uses large vocabulary of speech corpora based on HTK. The effect of different quantity of vocabularies, the sample ratio of training and testing, and the data quality on recognition rate of speaker-dependent were investigated. In addition, the identification of speaker-dependent acoustic model were compared. The experimental results show that the high quality data can improve the recognition rate and has little effect on the recognition rate in the vocabulary size.

show abstract

Large-Vocabulary Continuous Speech Recognition of Lhasa Tibetan

Cited by 3 publications

References 0 publications

Tibetan Multi-dialect Speech and Dialect Identity Recognition

Tibetan Multi-dialect Speech and Dialect Identity Recognition

Tibetan Multi-Dialect Speech Recognition Using Latent Regression Bayesian Network and End-To-End Mode

The Analysis of Influence Factors and Identification of Speaker-Dependent Primi Speech Recognition

Contact Info

Product

Resources

About