In the last years, the task of Query-by-Example Spoken Term Detection (QbE-STD), which aims to find occurrences of a spoken query in a set of audio documents, has gained the interest of the research community for its versatility in settings where untranscribed, multilingual and acoustically unconstrained spoken resources, or spoken resources in low-resource languages, must be searched. This paper describes and reports experimental results for a QbE-STD system that achieved the best performance in the recent Spoken Web Search (SWS) evaluation, held as part of MediaEval 2013. Though not optimized for speed, the system operates faster than real-time. The system exploits high-performance phone decoders to extract framelevel phone posteriors (a common representation in QbE-STD tasks). Then, given a query and a audio document, a distance matrix is computed between their phone posterior representations, followed by a newly introduced distance normalization technique and an iterative Dynamic Time Warping (DTW) matching procedure with some heuristic prunings. Results show that remarkable performance improvements can be achieved by using multiple examples per query and, specially, through the late (score-level) fusion of different subsystems, each based on a different set of phone posteriors.
El acceso a la versión del editor puede requerir la suscripción del recurso Access to the published version may require subscription
AbstractThis paper evaluates the performance of the twelve primary systems submitted to the evaluation on speaker verification in the context of a mobile environment using the MOBIO database. The mobile environment provides a challenging and realistic test-bed for current state-of-the-art speaker verification techniques. Results in terms of equal error rate (EER), half total error rate (HTER) and detection error trade-off (DET) confirm that the best performing systems are based on total variability modeling, and are the fusion of several sub-systems. Nevertheless, the good old UBM-GMM based systems are still competitive. The results also show that the use of additional data for training as well as gender-dependent features can be helpful.
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.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.