Search

27 Jul 2020: This model uses a long-range history and exploits topic information in the test text to adjust probabilities of test words. ... The PLSA-based model was found to lower test set perplexity over a traditional wordclass-based 4-gram by 13% (optimistic

27 Jul 2020: The focus is primarily on retrieval studies, and on speech tests directly related to retrieval, not on speech recognition itself. ... The report draws on the many and varied tests done during the project, but also presents a new series of results

27 Jul 2020: set. These factors control the point at which training set performance or test set generalisation become the limiting factors. ... The experimental results provide an empirical verification of the training set/test set performance of frame discriminative

27 Jul 2020: all the test data together or 8.6% when processing the test data shows independently. |

27 Jul 2020: The complete system yields an overall word error rate of 22.0% on the 1996 unpartitioned broadcast news development test data and just 15.8% on the 1997 evaluation test set. |

27 Jul 2020: For high dimensional inputs the performance of the single layer network alone is better on the test data, whereas for low dimensional inputs a two layer adaptive network is required and ... this needs much more training before it can produce a better

27 Jul 2020: It consists of speaker-independent (SI) read material, split into training, development test and evaluation test sets.

27 Jul 2020: An initial series of tests are performed in vitro. The statistical theory to predict the increase in signal to noise ratio is developed and verified empirically. ... Having proven the benefits of spatial compounding in vitro, tests with higher levels of

27 Jul 2020: word bigram "hub" tests and the second lowest error rate on the 20k word trigram "hub" test.

27 Jul 2020: word error rate on the 1998 evaluation test set.

27 Jul 2020: These tests have shown the applicability of nonlinear prediction to speech coding and the improvement in coding performance.

27 Jul 2020: In a number of test cases this algorithm is shown to converge reliably to the correct results. |

27 Jul 2020: The final 2003 CU-HTK CTS system constructed from some of these models is described and its performance on the DARPA/NIST 2003 Rich Transcription (RT-03) evaluation test set is

27 Jul 2020: This mismatch' function is then used to estimate the difference in channel conditions between training and test environments.

27 Jul 2020: On the 1998 test the system produced an average word error rate of 16.1% running in 9.5xRT. |

27 Jul 2020: Given insufficient training data, such a model's test set performance will be lower than that of a suitably biased model.

27 Jul 2020: In this paper we extend the HME to classification and results are reported for three common classification benchmark tests: Exclusive-Or, N-input Parity and Two Spirals.

27 Jul 2020: Results are presented for the 2001 development test set and the 2002 evaluation set.

27 Jul 2020: Abstract for niesler_tr265. Cambridge University Engineering Department Technical Report CUED/F-INFENG/TR265. COMPARATIVE EVALUATION OF WORD- AND CATEGORY-BASED LANGUAGE MODELS. Thomas Niesler and Phil Woodland. July 1996. Conventional n-gram

27 Jul 2020: The results show that the proposed method is very much faster yet performs acceptably compared to conventional systems which depend on keyword-specific training or prior knowledge of the test set