Automatic word recognition based on second-order hidden Markov models

Mari, Jean‐François; Haton, Jean Paul

doi:10.21437/icslp.1994-65

Cited by 13 publications

(15 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…This viewpoint is confirmed by Mari et al (1994Mari et al ( , 1997 in an experiment on connected digit recognition. Although in a direct comparison, the second-order HMMs improve on the results from first-order HMMs, this is no longer the case when the first-order model employs durational postprocessing.…”

Section: Prior Work On High-order Hidden Markov Modelsmentioning

confidence: 54%

“…In a series of papers, second-order HMMs were next applied to continuous speech recognition (Mari & Haton, 1994;Mari, Fohr & Junqua;Mari, Haton & Kriouile, 1997). They focus on the enhanced duration modelling capabilities of these models as primary factor in their enhanced performance, pointing out similarities 4…”

Section: Prior Work On High-order Hidden Markov Modelsmentioning

confidence: 99%

“…To summarise, previous work (He, 1988;Kriouile et al, 1990;Mari et al, 1994Mari et al, , 1996Mari et al, , 1997 focused on extending algorithms for processing second-order HMMs.…”

Section: Prior Work On High-order Hidden Markov Modelsmentioning

confidence: 99%

“…Although concerns were expressed about the effect the increase in the number of states might have (Mari et. al, 1994(Mari et.…”

Section: The Order Reducing (Ored) Algorithmmentioning

confidence: 99%

“…Although concerns were expressed about the effect the increase in the number of states might have (Mari et. al, 1994(Mari et. al, , 1997, we show in Section 2.6 that it does not contribute in any way to additional computational requirements.…”

Section: The Order Reducing (Ored) Algorithmmentioning

confidence: 99%

See 4 more Smart Citations

Efficient high-order hidden Markov modelling

Preez¹,

Weber²

1998

5th International Conference on Spoken Language Processing (ICSLP 1998)

View full text Add to dashboard Cite

Currently, first-order hidden Markov models (HMMs) form the backbone around which most automatic speech processing applications are built. Their higher-order extensions are known to be more powerful, but, due to their complexity and computational demands, they are seldomly used. It is the purpose of this work to advance their application In this work we unify HMMs of all orders by deriving and proving the ORder rEDucing (ORED) algorithm. This algorithm will reduce any higher-order HMM (also mixed-order) to an equivalent first-order representation. This makes it possible to process any higher-order HMM using known first-order algorithms, thereby making unnecessary the current approach of extending specific HMM algorithms to specific higher orders. It also provides an alternative theoretical basis to reason about high-order HMMs. From this perspective high-order transition probabilities are simply powerful mathematical specifications of first-order topology. We use this insight to explain old topologies and to design new ones.We address computational concerns by developing the Fast Incremental Training (FIT) algorithm. This algorithm avoids training redundant high-order probabilities by noting which lower-order transition probabilities are zero. This considerably reduces the memory and processor requirements during training. In addition, the resultant models have far fewer parameters and generalise better on previously unseen data.To show the practical applicability of our methodology we apply it to automatic language recognition. We find that it compares well with systems that require expensive transcribed databases (our system does not require this). ContentsCHAPTER 1 PURPOSE, OBJECTIVES AND CONTRIBUTION .

show abstract

Section: Prior Work On High-order Hidden Markov Modelsmentioning

confidence: 54%

Section: Prior Work On High-order Hidden Markov Modelsmentioning

confidence: 99%

“…To summarise, previous work (He, 1988;Kriouile et al, 1990;Mari et al, 1994Mari et al, , 1996Mari et al, , 1997 focused on extending algorithms for processing second-order HMMs.…”

Section: Prior Work On High-order Hidden Markov Modelsmentioning

confidence: 99%

“…Although concerns were expressed about the effect the increase in the number of states might have (Mari et. al, 1994(Mari et.…”

Section: The Order Reducing (Ored) Algorithmmentioning

confidence: 99%

Section: The Order Reducing (Ored) Algorithmmentioning

confidence: 99%

See 3 more Smart Citations