Using cross-decoder phone coocurrences in phonotactic language recognition

Abstract

Phonotactic language recognizers are based on the ability of phone decoders to produce phone sequences containing acoustic, phonetic and phonological information, which is partially dependent on the language. Input utterances are decoded and then scored by means of models for the target languages. Commonly, various decoders are applied in parallel and fused at the score level. A kind of complementarity effect is expected when fusing scores, since each decoder is assumed to extract different (and complementary) information from the input utterance. This assumption is supported by the performance improvements attained when fusing systems. However, decodings are processed in a fully uncoupled way, their time alignment (and the information that may be extracted from it) being completely lost. In this paper, a simple approach is proposed, which takes into account time alignment information, by considering cross-decoder phone coocurrences at the frame level. To evaluate the approach, a choice of open software (BUT front-end and phone decoders, SRI-LM toolkit, libSVM, FoCal) is used, and experiments are carried out on the NIST LRE2007 database. Adding phone coocurrences to the baseline phonotactic systems provides slight performance improvements, revealing the potential benefit of using cross-decoder dependencies for language modeling.

Improved Modeling of Cross-Decoder Phone Co-Occurrences in SVM-Based Phonotactic Language Recognition

“…In [1], a complete representation of phone co-occurrences was used, so that SVM vectors comprised between 2000 and 3000 unigrams for 2-decoder configurations and more than 124000 unigrams for a 3-decoder configuration. Under such a complete representation, including bigrams and trigrams of phone co-occurrences in SVM vectors was prohibitive.…”

“…Two variants of the approach presented in [1] are proposed. In the first one, SVM vectors consist of counts of up to 3-grams (instead of just unigrams) of 2-phone and 3-phone cooccurrences.…”

“…Recently, a simple approach has been proposed which takes into account cross-decoder phone co-occurrences at the frame level [1]. In that approach, phone segmentation is extracted as side information from 1-best phone decodings, and allows us to consider the co-occurrence of N phone labels (one per decoder) at each frame.…”

“…These decodings are processed and scored in a fully uncoupled way, their time alignment (and the information that may be extracted from it) being completely lost. Recently, a new approach to phonotactic language recognition has been presented [1], which takes into account time alignment information, by considering crossdecoder phone co-occurrences at the frame level, under two language modeling paradigms: smoothed n-grams and Support Vector Machines (SVM). Experiments on the NIST LRE2007 database demonstrated that using phone co-occurrence statistics could improve the performance of baseline phonotactic recognizers.…”

“…As for n-grams, the number of possible kphone co-occurrences increases exponentially with k, so in this work only 2-phone and 3-phone co-occurrences will be considered. In experiments on the NIST LRE2007 database, using Brno University of Technology (BUT) decoders for Czech, Hungarian and Russian [9], it was shown that fusing baseline phonotactic systems with systems based on cross-decoder phone co-occurrences led to improved performance in all the cases (see [1] for details). However, systems based on cross-decoder phone co-occurrences did not outperform the baseline phonotactic systems.…”