Statistical language modeling using the CMU-cambridge toolkit

We introduce in this paper a hierarchical hybrid statistical language model, represented as a collection of local models plus a general model that binds together the local ones. The model provides a unified framework for modelling language both above and below the word level, and we exemplify with models of both kinds for a large vocabulary task domain. To our knowledge this is the first paper to report an extensive evaluation of the improvements achieved from the use of local models within a hierarchical framework in comparison with a conventional word-based trigram model.

show abstract

“…The baseline model is a word-based trigram model, with Witten-Bell discounting, built with the CMU-SLM Toolkit ( [2]).…”

Section: Resultsmentioning

confidence: 99%

Evaluating hierarchical hybrid statistical language models

Galescu

Allen

2000

6th International Conference on Spoken Language Processing (ICSLP 2000)

View full text Add to dashboard Cite

show abstract

“…In previous works [9] Backing-off smoothing technique was chosen because the involved recursive scheme has been well integrated in the finite state formalism. Within the backing-off formalism the Witten-Bell discounting was applied [10]. For the evaluation of the k-TSS LM, a task-oriented Spanish corpus [11], consisting of 82.000 words and a vocabulary of 1.208 words, was used.…”

Section: 1-the Syntactic Language Modelmentioning

confidence: 99%

Integrating different acoustic and syntactic language models in a continuous speech recognition system

Varona¹,

Torres²,

Ipiña³

et al. 2000

6th International Conference on Spoken Language Processing (ICSLP 2000)

View full text Add to dashboard Cite

Continuous Speech Recognition (CSR) systems require acoustic models to represent the characteristics of the acoustic signal and Language Models (LM) to represent the syntactic constraints of the language. Both acoustic and LM probability distributions are usually independently obtained and evaluated. Then, the respective "best" models are selected to be integrated in the CSR systems. But, in this paper it was proved that the use of more accurate acoustic models (for example the use of semicontinuous models instead of discrete ones or the use of a bigger number of then representing a more complete set of sublexical units) didn't always mean a better performance of the integrated system because the acoustic improvements were softened when the LM probabilities were applied. This experimental evaluation was carried out over a Spanish speech application task.

show abstract

“…It incorporates many search techniques to provide high level search performance, and can deal with various types models to be used for their cross evaluation. Standard formats that other popular modeling tools [1][2] use is adopted. On a 20k-word dictation task with word 3-gram and triphone HMM, it realizes almost real-time processing on most current PCs.…”

Section: Introductionmentioning

confidence: 99%

Julius --- an open source real-time large vocabulary recognition engine

Lee¹,

Kawahara²,

Shikano³

2001

7th European Conference on Speech Communication and Technology (Eurospeech 2001)

328

View full text Add to dashboard Cite

Julius is a high-performance, two-pass LVCSR decoder for researchers and developers. Based on word 3-gram and context-dependent HMM, it can perform almost realtime decoding on most current PCs in 20k word dictation task. Major search techniques are fully incorporated such as tree lexicon, N-gram factoring, cross-word context dependency handling, enveloped beam search, Gaussian pruning, Gaussian selection, etc. Besides search efficiency, it is also modularized carefully to be independent from model structures, and various HMM types are supported such as shared-state triphones and tiedmixture models, with any number of mixtures, states, or phones. Standard formats are adopted to cope with other free modeling toolkit. The main platform is Linux and other Unix workstations, and partially works on Windows. Julius is distributed with open license together with source codes, and has been used by many researchers and developers in Japan.

show abstract

Statistical language modeling using the CMU-cambridge toolkit

Cited by 306 publications

References 0 publications

Evaluating hierarchical hybrid statistical language models

Evaluating hierarchical hybrid statistical language models

Integrating different acoustic and syntactic language models in a continuous speech recognition system

Julius --- an open source real-time large vocabulary recognition engine

Contact Info

Product

Resources

About