|
Abstract
|
|
|---|---|
| In this paper we present our results on using Recurrent Neural Networks Language Model scores (RNNLM) trained on different phone-gram orders and using different phonetic ASR recognizers. In order to avoid data sparseness problems and to reduce the vocabulary of all possible n-gram combinations, a K-means clustering procedure was performed using phone vector embeddings as a pre-processing step. We will provide more details on the vocabulary reduction efforts on 2-gram and 3-gram. Additional experiments to optimize the amount of classes, batch-size, hidden neurons, state-unfolding, are also presented. We have worked with the KALAKA-3 database for the plenty closed condition. Thanks to our clustering technique and the combination of high level phone-grams, our phonotactic system performs more than 10% better than the unigram-based RNNLM system. Also, the obtained RNNLM scores are calibrated and fused with other scores from an acoustic-based i-vector system and a traditional PPRLM system. This fusion provides additional improvements showing that they provide complementary information to the LID system. | |
|
International
|
Si |
|
Congress
|
Iberspeech 2016 |
|
|
970 |
|
Place
|
Lisboa - Portugal |
|
Reviewers
|
Si |
|
ISBN/ISSN
|
978-3-319-49169-1 |
|
|
|
|
Start Date
|
23/11/2016 |
|
End Date
|
25/11/2016 |
|
From page
|
109 |
|
To page
|
118 |
|
|
IberSpeech 2016 - Proceedings |