Graphical models for integrating syllabic information

Bartels, Chris; Bilmes, Jeff

doi:10.1016/j.csl.2009.11.001

Cited by 5 publications

(2 citation statements)

References 41 publications

(48 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It has a perplexity of 36.4 on the test set. System WAC HTK; PLP [9] 38.80 GMTK [10]; PLP [9] 40.80 HTK; PLP (our baseline2) 44.26 GMTK; PLP + Fisher-trained, factored AF tandem [9] 46.20 GMTK; PLP + tuning on larger development set [11] 48.10…”

Section: Methodsmentioning

confidence: 99%

“…One may see that our baseline2 result compares favorably among the top three systems that do not use data other than the SVB 500-word corpus. The 4th system made use of articulatory tandem features trained on the Fisher corpus [12], while the 5th system was tuned on a larger development set, of which the details were not described in [11].…”

Section: Experiments 1: Equal Transition Probabilitiesmentioning

confidence: 99%

See 1 more Smart Citation

Transition probabilities are more important than we once thought

Chen

Mak

2012

2012 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

View full text Add to dashboard Cite

It is generally believed that the transition probabilities in a hidden Markov model (HMM) have a limited role in the speech decoding process. In this paper, through a series of recognition experiments on Wall Street Journal (WSJ) read speech and SVitchboard (SVB) conversational telephone speech, we find that the HMM transition probabilities may be more important than we once thought. The experiments include: (1) setting or not setting all outgoing transition probabilities equal; (2) the introduction of word-final triphones and the re-estimation of their transition probabilities; (3) besides grammar factor and insertion penalty, the addition of a third decoding parameter called transition factor to scale the transition probability score during decoding. The results of the above three experiments enable us to improve the the word accuracy of the WSJ and SVB speech recognition task by 0.7% and 5.3% absolute respectively when compared to their baseline model in which all transition probabilities are simply set to 0.5.

show abstract