Integrating High and Low Smoothed LMs in a CSR System

Varona, Amparo; Torres, M. Inés

doi:10.1007/978-3-540-24586-5_28

Cited by 2 publications

(2 citation statements)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The gap among accumulated probabilities is therefore usually bigger than the gap among LM probabilities (see Figure 1). The immediate consequence is that LM probabilities are irrelevant in most situations when choosing the best, most probable, partial path [4]. The scarce contribution of the LM probabilities leads to low CSR system performances.…”

Section: Scaling Factors Over Acoustic and Lm Probability Distributionsmentioning

confidence: 99%

See 1 more Smart Citation

Scaling Acoustic and Language Model Probabilities in a CSR System

Varona

Torres

2004

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

It is well known that a direct integration of acoustic and language models (LM) into a Continuous Speech Recognition (CSR) system leads to low performances. This problem has been analyzed in this work as a practical numerical problem. There are two ways to get optimum system performances: scaling acoustic or language model probabilities. Both approaches have been analyzed from a numerical point of view. They have also been experimentally tested on a CSR system over two Spanish databases. These experiments show similar reductions in word recognition rates but very different computational cost behaviors. They also show that the values of scaling factors required to get optimum CSR systems performances are closely related to other heuristic parameters in the system like the beam search width.

show abstract

Section: Scaling Factors Over Acoustic and Lm Probability Distributionsmentioning

confidence: 99%

“…In addition, they appear more times. As a consequence, the contribution of LM probabilities could not be relevant [4] to obtain the most probable word sequenceΩ. Therefore, low CSR system performances are usually obtained.…”

Section: Introductionmentioning

confidence: 99%

Scaling Acoustic and Language Model Probabilities in a CSR System

Varona

Torres

2004

Lecture Notes in Computer Science

Self Cite

View full text Add to dashboard Cite

show abstract

Scaling Smoothed Language Models

Varona¹,

Torres²

2005

Int J Speech Technol

View full text Add to dashboard Cite

In Continuous Speech Recognition (CSR) systems a Language Model (LM) is required to represent the syntactic constraints of the language. Then a smoothing technique needs to be applied to avoid null LM probabilities. Each smoothing technique leads to a different LM probability distribution. Test set perplexity is usually used to evaluate smoothing techniques but the relationship with acoustic models is not taken into account. In fact, it is well-known that to obtain optimum CSR performances a scaling exponential parameter must be applied over LMs in the Bayes' rule. This scaling factor implies a new redistribution of smoothed LM probabilities. The shape of the final probability distribution is due to both the smoothing technique used when designing the language model and the scaling factor required to get the optimum system performance when integrating the LM into the CSR system. The main object of this work is to study the relationship between the two factors, which result in dependent effects. Experimental evaluation is carried out over two Spanish speech application tasks. Classical smoothing techniques representing very different degrees of smoothing are compared. A new proposal, Delimited discounting, is also considered. The results of the experiments showed a strong dependence between the amount of smoothing given by the smoothing technique and the way that the LM probabilities need to be scaled to get the best system performance, which is perplexity independent in many cases. This relationship is not independent of the task and available training data.

show abstract

Integrating High and Low Smoothed LMs in a CSR System

Cited by 2 publications

References 9 publications

Scaling Acoustic and Language Model Probabilities in a CSR System

Scaling Acoustic and Language Model Probabilities in a CSR System

Scaling Smoothed Language Models

Contact Info

Product

Resources

About