Contains fulltext : 296407.pdf (Publisher’s version ) (Open Access)

\n Contains fulltext :\n 238372.pdf (Publisher’s version ) (Open Access)\n \n Contains fulltext :\n 238372pre.pdf (Author’s version preprint ) (Open Access)\n

\n Contains fulltext :\n 238373.pdf (Publisher’s version ) (Open Access)\n

\n Contains fulltext :\n 238370.pdf (Publisher’s version ) (Open Access)\n

\n Contains fulltext :\n 228101.pdf (Publisher’s version ) (Open Access)\n

Reaction times (RTs) are used widely in psychological and\npsycholinguistic research as inexpensive measures of underlying\ncognitive processes. However, inferring cognitive processes\nfrom RTs is hampered by the fact that actual responses are the\nresult of multiple factors, many of which may not be related to\nthe process of interest. In lexical decision experiments, the use\nof RTs is further complicated by the fact that the response to\nsome stimuli is missing, and the fact that part of the responses\nare ’incorrect’.\nIn this paper we investigate the distribution of missing and incorrect\nresponses in the RT sequences of two large lexical decision\nexperiments. It appears that a substantial part of incorrect\nresponses cluster together. Then, we investigate the effect of\nclusters of incorrect responses on surrounding RTs.\nAlso, we extend previous research on methods for discovering\nand removing so-called local speed effects from RT sequences.\nFor this purpose, we show that a recently introduced graph based\nRT analysis method can help to better understand and\nanalyze RT sequences.

The neural processing of speech leads to specific patterns in\nthe brain which can be measured as, e.g., EEG signals. When\nproperly aligned with the speech input and averaged over many\ntokens, the Event Related Potential (ERP) signal is able to\ndifferentiate specific contrasts between speech signals. Well known\neffects relate to the difference between expected and\nunexpected words, in particular in the N400, while effects in\nN100 and P200 are related to attention and acoustic onset effects.\nMost EEG studies deal with the amplitude of EEG signals\nover time, sidestepping the effect of phase and phase synchronization.\nThis paper investigates the relation between phase in\nthe EEG signals measured in an auditory lexical decision task\nby Dutch participants listening to full and reduced English word\nforms. We show that phase synchronization takes place across\nstimulus conditions, and that the so-called circular variance is\nnarrowly related to the type of contrast between stimuli.

The recent advent of deep learning techniques in speech tech-nology and in particular in automatic speech recognition hasyielded substantial performance improvements. This suggeststhat deep neural networks (DNNs) are able to capture structurein speech data that older methods for acoustic modeling, suchas Gaussian Mixture Models and shallow neural networks failto uncover. In image recognition it is possible to link repre-sentations on the first couple of layers in DNNs to structuralproperties of images, and to representations on early layers inthe visual cortex. This raises the question whether it is possi-ble to accomplish a similar feat with representations on DNNlayers when processing speech input. In this paper we presentthree different experiments in which we attempt to untanglehow DNNs encode speech signals, and to relate these repre-sentations to phonetic knowledge, with the aim to advance con-ventional phonetic concepts and to choose the topology of aDNNs more efficiently. Two experiments investigate represen-tations formed by auto-encoders. A third experiment investi-gates representations on convolutional layers that treat speechspectrograms as if they were images. The results lay the basisfor future experiments with recursive networks.

Sequences of reaction times (RT) produced by participants in an experiment are not only influenced by the stimuli, but by many other factors as well, including fatigue, attention, experience, IQ, handedness, etc. These confounding factors result in longterm effects (such as a participant’s overall reaction capability) and in short- and medium-time fluctuations in RTs (often referred to as ‘local speed effects’). Because stimuli are usually presented in a random sequence different for each participant, local speed effects affect the underlying ‘true’ RTs of specific trials in different ways across participants. To be able to focus statistical analysis on the effects of the cognitive process under study, it is necessary to reduce the effect of confounding factors as much as possible. In this paper we propose and compare techniques and criteria for doing so, with focus on reducing (‘filtering’) the local speed effects. We show that filtering matters substantially for the significance analyses of predictors in linear mixed effect regression models. The performance of filtering is assessed by the average between-participant correlation between filtered RT sequences and by Akaike’s Information Criterion, an important measure of the goodness-of-fit of linear mixed effect regression models.

Analyzing EEG signals recorded while participants are listening to continuous speech with the purpose of testing linguistic hypotheses is complicated by the fact that the signals simultaneously reflect exogenous acoustic excitation and endogenous linguistic processing. This makes it difficult to trace subtle differences that occur in mid-sentence position. We apply an analysis based on multivariate temporal response functions to uncover subtle mid-sentence effects. This approach is based on a per-stimulus estimate of the response of the neural system to speech input. Analyzing EEG signals predicted on the basis of the response functions might then bring to light conditionspecific differences in the filtered signals. We validate this approach by means of an analysis of EEG signals recorded with isolated word stimuli. Then, we apply the validated method to the analysis of the responses to the same words in the middle of meaningful sentences.

This paper investigates the processes in comprehending spoken noun-noun compounds, using data from the BALDEY database. BALDEY contains lexicality judgments and reaction times (RTs) for Dutch stimuli for which also linguistic information is included. Two different approaches are combined. The first is based on regression by Dynamic Survival Analysis, which models decisions and RTs as a consequence of the fact that a cumulative density function exceeds some threshold. The parameters of that function are estimated from the observed RT data. The second approach is based on DIANA, a process-oriented computational model of human word comprehension, which simulates the comprehension process with the acoustic stimulus as input. DIANA gives the identity and the number of the word candidates that are activated at each 10 ms time step. Both approaches show how the processes involved in comprehending compounds change during a stimulus. Survival Analysis shows that the impact of word duration varies during the course of a stimulus. The density of word and non-word hypotheses in DIANA shows a corresponding pattern with different regimes. We show how the approaches complement each other, and discuss additional ways in which data and process models can be combined.

This paper combines two different approaches to modeling reaction time data from lexical decision experiments, viz. a dataoriented statistical analysis by means of a linear mixed effects model, and a process-oriented computational model of human speech comprehension. The linear mixed effect model is implemented by lmer in R. As computational model we apply DIANA, an end-to-end computational model which aims at modeling the cognitive processes underlying speech comprehension. DIANA takes as input the speech signal, and provides as output the orthographic transcription of the stimulus, a word/non-word judgment and the associated reaction time. Previous studies have shown that DIANA shows good results for large-scale lexical decision experiments in Dutch and North-American English. We investigate whether predictors that appear significant in an lmer analysis and processes implemented in DIANA can be related and inform both approaches. Predictors such as ‘previous reaction time’ can be related to a process description; other predictors, such as ‘lexical neighborhood’ are hard-coded in lmer and emergent in DIANA. The analysis focuses on the interaction between subject variables and task variables in lmer, and the ways in which these interactions can be implemented in DIANA.

Event-Related Potential (ERP) signals obtained from EEG recordings are widely used for studying cognitive processes in spoken language processing.The computation of ERPs involves averaging over multiple participants and multiple stimuli.Especially with speech stimuli, which also evoke substantial exogenous excitation, even averaging within conditions results in pooling many sources of variance.This raises questions about the statistical processing needed to uncover reliable differences between conditions.In this study we investigate differences between ERPs when participants listened to full and reduced pronunciations of verb forms in Dutch, in isolation and in mid-sentence position.Conventional statistical analysis uncovers some (but not all) differences between full and reduced forms in isolation, but not in mid-sentence position.In this paper, we show that linear mixed models (lmer) and generalized additive models (gam), which are able to account for participant-and stimulus-related variance may uncover more effects than conventional statistical models.However, depending on the complexity of the data, lmer and gam models may not be able to fit the data closely enough to warrant blind interpretation of the summary output.We discuss opportunities and threats of these approaches to analyzing ERP signals.

Most state-of-the-art phone classifiers use the same features and decision criteria for all phones, despite the fact that different broad classes are characterized by different manners and place of articulation that result in different acoustic features. This paper uses manifold learning to address structure in the acoustic space. Previous approaches to dimensionality reduction based on manifold learning assumed that the acoustic space can be characterized by a uniform manifold structure. In this paper we relax this assumption by learning different manifold structures for broad phonetic classes. Because all known classifiers make confusions between broad classes, we designed a two-level classifier in which the top level consists of a number of partially overlapping broad classes. Since the resulting classifiers are not statistically independent, we propose a new method for fusing the classifiers. Experimental results show that our two-level classifier obtained slightly better results when broad-class specific manifolds were learned, compared to a uniform manifold. However, the accuracy is still considerably lower than what could be obtained with oracle knowledge about broad class membership. From this we infer that phones do not form compact clusters in acoustic space.

This paper investigates a computational model that combines a frontend based on an auditory model with an exemplar-based sparse coding procedure for estimating the posterior probabilities of sub-word units when processing noisified speech. Envelope modulation spectrogram (EMS) features are extracted using an auditory model which decomposes the envelopes of the outputs of a bank of gammatone filters into one lowpass and multiple bandpass components. Through a systematic analysis of the configuration of the modulation filterbank , we investigate how and why different configurations affect the posterior probabilities of sub-word units by measuring the recognition accuracy on a semantics-free speech recognition task. Our main finding is that representing speech signal dynamics by means of multiple bandpass filters typically improves recognition accuracy. This effect is particularly noticeable in very noisy conditions. In addition we find that to have maximum noise robustness, the bandpass filters should focus on low modulation frequencies . This reenforces our intuition that noise robustness can be increased by exploiting redundancy in those frequency channels which have long enough integration time not to suffer from envelope modulations that are solely due to noise. The ASR system we design based on these findings behaves more similar to human recognition of noisified digit strings than conventional ASR systems. Thanks to the relation between the modulation filterbank and procedures for computing dynamic acoustic features in conventional ASR systems, the finding can be used for improving the frontends in those systems.

DIANA is an end-to-end computational model of speech processing, which takes as input the speech signal, and provides as output the orthographic transcription of the stimulus, a word/non-word judgment and the associated estimated reaction time. So far, the model has only been tested for Dutch. In this paper, we extend DIANA such that it can also process North American English. The model is tested by having it simulate human participants in a large scale North American English lexical decision experiment. The simulations show that DIANA can adequately approximate the reaction times of an average participant (r = 0.45). In addition, they indicate that DIANA does not yet adequately model the cognitive processes that take place after stimulus offset.

This paper addresses the question how to compare reaction times computed by a computational model of speech comprehension with observed reaction times by participants. The question is based on the observation that reaction time sequences substantially differ per participant, which raises the issue of how exactly the model is to be assessed. Part of the variation in reaction time sequences is caused by the so-called local speed: the current reaction time correlates to some extent with a number of previous reaction times, due to slowly varying variations in attention, fatigue etc. This paper proposes a method, based on time series analysis, to filter the observed reaction times in order to separate the local speed effects. Results show that after such filtering the between-participant correlations increase as well as the average correlation between participant and model increases. The presented technique provides insights into relevant aspects that are to be taken into account when comparing reaction time sequences

This paper describes a computational model of speech comprehension that takes the acoustic signal as input and predicts reaction times as observed in an auditory lexical decision task. By doing so, we explore a new generation of end-to-end computational models that are able to simulate the behaviour of human subjects participating in a psycholinguistic experiment. So far, nearly all computational models of speech comprehension do not start from the speech signal itself, but from abstract representations of the speech signal, while the few existing models that do start from the acoustic signal cannot directly model reaction times as obtained in comprehension experiments. The main functional components in our model are the perception stage, which is compatible with the psycholinguistic model Shortlist B and is implemented with techniques from automatic speech recognition, and the decision stage, which is based on the linear ballistic accumulation decision model. We successfully tested our model against data from 20 participants performing a largescale auditory lexical decision experiment. Analyses show that the model is a good predictor for the average judgment and reaction time for each word.

Retrieving information from the ever-increasing amount of unannotated audio and video recordings requires techniques such as unsupervised pattern discovery or query-by-example. In this paper we focus on queries that are specified in the form of an audio snippet containing the desired word or expression excised from the target recordings. The task is to retrieve all-and-only the instances whose match score with the query meet an absolute criterion. For this purpose we introduce a distance measure between two acoustic vectors that can be calibrated in a completely unsupervised manner. The use of that measure also allows the use of a fast matching approach, which makes it possible to skip more than 97 % of full-fledged DTW with-out affecting performance in terms of precision and recall. We demonstrate the effectiveness of the proposals with query-by-example experiments conducted on a read speech corpus for English and a spontaneous speech corpus for Dutch.

The use of higher-order polynomial acoustic features can improve the performance of automatic speech recognition.However, the dimensionality of the polynomial representation can be prohibitively large, making the training of acoustic models using polynomial features for large vocabulary ASR systems infeasible.This paper presents an iterative polynomial training framework for acoustic modeling, which recursively expands the current acoustic features into their second-order polynomial feature space.In each recursion the dimensionality is reduced by a linear projection, such that increasingly higher order polynomial information is incorporated while keeping the dimensionality of the acoustic models constant.Experimental results obtained for a large-vocabulary continuous speech recognition task show that the proposed method outperforms conventional mixture models.

With the increasing rate of patent application filings, automated patent classification is of rising economic importance. This article investigates how patent classification can be improved by using different representations of the patent documents. Using the Linguistic Classification System (LCS), we compare the impact of adding statistical phrases (in the form of bigrams) and linguistic phrases (in two different dependency formats) to the standard bag-of-words text representation on a subset of 532,264 English abstracts from the CLEF-IP 2010 corpus. In contrast to previous findings on classification with phrases in the Reuters-21578 data set, for patent classification the addition of phrases results in significant improvements over the unigram baseline. The best results were achieved by combining all four representations, and the second best by combining unigrams and lemmatized bigrams. This article includes extensive analyses of the class models (a.k.a. class profiles) created by the classifiers in the LCS framework, to examine which types of phrases are most informative for patent classification. It appears that bigrams contribute most to improvements in classification accuracy. Similar experiments were performed on subsets of French and German abstracts to investigate the generalizability of these findings.

Modeling the second-order statistics of articulatory trajectories is likely to improve the performance in classifying phone segments compared to using only linear combinations of MFCCs. Nevertheless, the extremely high dimensionality of the feature space spanned by a combination of monomials of degree-1 and degree-2 makes it difficult to effectively exploit the discriminative information in the full covariance matrix. This paper proposes a novel algorithm, dubbed Knowledge-based Quadratic Discriminant Analysis (KnQDA), for reducing the number of dimensions of the space spanned by degree-1 and degree-2 monomials by using phonetic knowledge for selecting the set of degree-2 monomials that are most likely to improve classification. KnQDA seeks a trade-off between overfitting and undertraining, which further improves the learnability. Binary classifications on all pairs of phones in TIMIT show the effectiveness of the proposed method, especially on those phone pairs that overlap strongly in the linear feature space.

Contains fulltext : 102118.pdf (author's version ) (Open Access) Contains fulltext : 102118.pdf (Publisher’s version ) (Open Access)

Contains fulltext : 101559.pdf (Publisher’s version ) (Open Access)

In this work we propose the use of Functional Data Analysis (FDA) as a powerful methodology to tackle problems where multiple continuous speech parameters have to be analyzed jointly. A production study on contrastive focus placement in Neapolitan Italian is used as illustration. Two features are analyzed, viz. f0and relative speech rate, both expressed as continuous functions of time. The results show that known facts about the prosody of Neapolitan Italian emerge from the data, but also other interesting local or cross-feature relationships between contour traits appear. Thus, FDA results can be used as guidance in the exploration of speech feature contour shapes, an operation that used to be carried out manually in previous speech research. The capability of jointly analyzing multiple continuous features provides a valuable improvement not only for speech analysis but also for speech re-synthesis.

In the present paper we investigate the effect of categorising raw behavioural data or computational model responses. In addition, the effect of averaging over stimuli from potentially different populations is assessed. To this end, we replicate studies on word learning and generalisation abilities using the ACORNS models. Our results show that discrete categories may obscure interesting phenomena in the continuous responses. For example, the finding that learning in the model saturates very early at a uniform high recognition accuracy only holds for categorical representations. Additionally, a large difference in the accuracy for individual words is obscured by averaging over all stimuli. Because different words behaved differently for different speakers, we could not identify a phonetic basis for the differences. Implications and new predictions for infant behaviour are discussed.

Contains fulltext : 94442.pdf (author's version ) (Open Access)

Contains fulltext : 94477.pdf (Publisher’s version ) (Open Access)

This paper investigates the effects of novel words on a cogni-tively plausible computational model of word learning. The model is first familiarized with a set of words, achieving high recognition scores and subsequently offered novel words for training. We show that the model is able to recognize the novel words as different from the previously seen words, based on a measure of novelty that we introduce. We then propose a pro-cedure analogous to novelty preference in infants. Results from simulations of word learning show that adding this procedure to our model speeds up training and helps the model attain higher recognition rates. Index Terms: language acquisition, word learning, computa-tional modelling

In this paper, we evaluate a number of machine learning techniques for the task of ranking answers to why-questions. We use TF-IDF together with a set of 36 linguistically motivated features that characterize questions and answers. We experiment with a number of machine learning techniques (among which several classifiers and regression techniques, Ranking SVM and SVM map ) in various settings. The purpose of the experiments is to assess how the different machine learning approaches can cope with our highly imbalanced binary relevance data, with and without hyperparameter tuning. We find that with all machine learning techniques, we can obtain an MRR score that is significantly above the TF-IDF baseline of 0.25 and not significantly lower than the best score of 0.35. We provide an in-depth analysis of the effect of data imbalance and hyperparameter tuning, and we relate our findings to previous research on learning to rank for Information Retrieval.

In the present paper we show that a general-purpose word learning model can simulate several important findings from recent experiments in language acquisition. Both the addition of background noise and varying the speaker have been found to influence infants' performance during word recognition experiments. We were able to replicate this behaviour in our artificial word learning agent. We use the results to discuss both advantages and limitations of computational models of language acquisition.

Contains fulltext : 86189.pdf (author's version ) (Open Access)

This paper presents an analysis of phoneme durations of emotional speech in two languages: Dutch and Korean. The analyzed corpus of emotional speech has been specifically developed for the purpose of cross-linguistic comparison, and is more balanced than any similar corpus available so far: a) it contains expressions by both Dutch and Korean actors and is based on judgments by both Dutch and Korean listeners; b) the same elicitation technique and recording procedure were used for recordings of both languages; and c) the phonetics of the carrier phrase were constructed to be permissible in both languages. The carefully controlled phonetic content of the carrier phrase allows for analysis of the role of specific phonetic features, such as phoneme duration, in emotional expression in Dutch and Korean. In this study the mutual effect of language and emotion on phoneme duration is presented.

While developing an approach to why-QA, we extended a passage retrieval system that uses off-the-shelf retrieval technology with a re-ranking step incorporating structural information. We get significantly higher scores in terms of MRR@150 (from 0.25 to 0.34) and success@10. The 23% improvement that we reach in terms of MRR is comparable to the improvement reached on different QA tasks by other researchers in the field, although our re-ranking approach is based on relatively lightweight overlap measures incorporating syntactic constituents, cue words, and document structure.

This paper combines acoustic features with a high temporal and a high frequency resolution to reliably classify articulatory events of short duration, such as bursts in plosives. SVM classification experiments on TIMIT and SV Articulatory showed that articulatory-acoustic features (AFs) based on a combination of MFCCs derived from a long window of 25 ms and a short window of 5 ms that are both shifted with 2.5 ms steps (Both) outperform standard MFCCs derived with a window of 25 ms and a shift of 10 ms (Baseline). Finally, comparison of the TIMIT and SV Articulatory results showed that for classifiers trained on data that allows for asynchronously changing AFs (SV Articulatory) the improvement from Baseline to Both is larger than for classifiers trained on data where AFs change simultaneously with the phone boundaries (TIMIT).

In this paper we compare three different implementations of language learning to investigate the issue of speaker-dependent initial representations and subsequent generalization. These implementations are used in a comprehensive model of lan-guage acquisition under development in the FP6 FET project ACORNS. All algorithms are embedded in a cognitively and ecologically plausible framework, and perform the task of de-tecting word-like units without any lexical, phonetic, or phono-logical information. The results show that the computational approaches differ with respect to the extent they deal with un-seen speakers, and how generalization depends on the variation observed during training. Index Terms: Language acquisition, Computational modeling 1.

This paper introduces the EU-FP7 project CLARIN, a joint effort of over 150 institutions in Europe, aimed at the creation of a sustainable language resources and technology infrastructure for the humanities and social sciences research community. The paper briefly introduces the vision behind the project and how it relates to speech research with a focus on the contributions that CLARIN can and will make to research in spoken language processing. Index Terms: speech research, resources, standardization

In this paper we introduce Functional Data Analysis (FDA) as a tool for analyzing dynamic transitions in speech signals. FDA makes it possible to perform statistical analyses of sets of mathematical functions in the same way as classical multivariate analysis treats scalar measurement data. We illustrate the use of FDA with a reduction phenomenon affecting the French word c'était /setε/ 'it was', which can be reduced to [stε] in conversational speech. FDA reveals that the dynamics of the transition from [s] to [t] in fully reduced cases may still be different from the dynamics of [s] - [t] transitions in underlying /st/ clusters such as in the word stage.

In this paper, we discuss a computational model that is able to detect and build word-like representations on the basis of sensory input. The model is designed and tested with a further aim to investigate how infants may learn to communicate by means of spoken language. The computational model makes use of a memory, a perception module, and the concept of 'learning drive'. Learning takes place within a communicative loop between a 'caregiver' and the 'learner'. Experiments carried out on three European languages with different genetic background (Finnish, Swedish, and Dutch) show that a robust word representation can be learned in using less than 100 acoustic tokens (examples) of that word. The model is inspired by the memory structure that is assumed functional for human cognitive processing.

Young infants learn words by detecting patterns in the speech signal and by associating these patterns to stimuli presented by non-speech modalities (e.g vision). In this paper, we model this behaviour by designing and testing a computational model of word discovery. The model is able to build word-like representations on the basis of multimodal input data. The discovery of words (and word-like entities) takes place within a communicative loop between two protagonists, a ’carer ’ and the ’learner’. Experiments carried out on three different European languages (Finnish, Swedish, and Dutch) show that a robust word representation can be learned in using about 50 acoustic tokens (examples) of that word. The model is inspired by the memory structure that is assumed functional for human speech processing. Index Terms: language acquisition, unsupervised word detection, computational modelling 1.

Subjective speech evaluation is the gold standard to assess speech quality of head and neck cancer patients. This study investigates if conventional acoustic-phonetic and novel feature analysis contribute to the development of a multidimensional speech assessment protocol. Speech recordings of 51 patients 6 months post-treatment and of 18 control speakers were subjectively evaluated for intelligibility, nasal resonance and articulation. Self-evaluation of speech problems was assessed by the EORTC QLQ-H&N35 speech subscale. Feature analysis was performed to assess objectively nasality in vowels /a,i,u/. Results revealed that size of the vowel triangle, pressure release of /k/ and nasality in /i/ predict best intelligibility, articulation and nasal resonance and differentiated best between patients and controls. Within patients, /k/ and /x/ differentiated tumour site and tumour classification. Various objective variables were related to speech problems as reported by patients.

In this paper we explore the feasibility of the MemoryPrediction Theory, implemented in the form of a Hierarchical Temporal Memory (HTM), for automatic speech recognition. Up to nowHTMs have almost exclusively been applied to image processing. However, the underlying theory can also be used as an approach to active perception of audio signals. Using the software platform under development by NUMENTA we implemented a system for isolated digit recognition, the speech recognition task that can be most easily cast in a form similar to image recognition. Our results show that the HTM approach holds promises for speech recognition. At the same time it is clear that the present implementation is not ideally suited for processing signals that encode information mainly in dynamic changes.

This paper presents the steps needed to make a corpus of Dutch spontaneous dialogues accessible for automatic phonetic research aimed at increasing our understanding of reduction phenomena and the role of fine phonetic detail. Since the corpus was not created with automatic processing in mind, it needed to be reshaped. The first part of this paper describes the actions needed for this reshaping in some detail. The second part reports the results of a preliminary analysis of the reduction phenomena in the corpus. For this purpose a phonemic transcription of the corpus was created by means of a forced alignment, first with a lexicon of canonical pronunciations and then with multiple pronunciation variants per word. In this study pronunciation variants were generated by applying a large set of phonetic processes that have been implicated in reduction to the canonical pronunciations of the words. This relatively straightforward procedure allows us to produce plausible pronunciation variants and to verify and extend the results of previous reduction studies reported in the literature.

Recent research suggests that it is more appropriate to model pronunciation variation with syllable-length acoustic models than with triphones. Due to the large number of factors contributing to pronunciation variation at the syllable level, the creation of multi-path model topologies appears necessary. In this paper, we construct multi-path models using phonetic knowledge to initialise the parallel paths, and a data-driven solution for their reestimation. When applied to 94 frequent syllables in a Dutch read speech recognition task, the approach leads to improved recognition performance when compared with a much more complex triphone recogniser. A detailed analysis of the pronunciation variation captured by the parallel paths pinpoints the deficiencies of the approach, and provides insights into how these may be overcome.

In this paper, we construct multi-path syllable models using phonetic knowledge for initialising the parallel paths, and a data-driven solution for their re-estimation. We hypothesise that the richer topology of multi-path syllable models would be better at accounting for pronunciation variation than context-dependent phone models that can only account for the effects of left and right neighbours. We show that parallel paths that are initialised with phonetic knowledge and then reestimated do indeed result in different trajectories in feature space. Yet, this does not result in better recognition performance. We suggest explanations for this finding, and provide the reader with important insights into the issues playing a role in pronunciation variation modelling with multi-path syllable models. Index Terms: speech recognition, hidden Markov models, multi-path syllable models, Kullback-Leibler distance,

Contains fulltext : 44262.pdf (Publisher’s version ) (Closed access)

Speech quality of patients with oral or oropharyngeal carcinoma was assessed by perceptual and acoustic-phonetic analyses. Speech recordings of running speech of patients before and 6 and 12 months after treatment for oral or oropharyngeal cancer and of 18 control speakers were evaluated regarding intelligibility, nasality and articulation, which revealed deteriorated speech in 20% of the patients before treatment, and in 75% 6-12 months after treatment. Acoustic analyses comprised formant, duration, perturbation and noise measures of the vowels /i/, /a/, and /u/ and were performed on the speech samples 6 months after treatment and the controls. Patients appeared to have a smaller vowel space compared to controls, which was clearly related to speech intelligibility. Furthermore, voice perturbation appeared to be higher in patients. Although oropharyngeal treatment does not effect the function of the larynx itself, the acoustic coupling between source and filter may effect the smoothness of the voicing characteristics. The presented speech analyses may serve as part of an outcome measurement protocol for assessing efficacy of speech rehabilitation.

No abstract available.

In this paper, we introduce a novel method for clustering speech gestures, represented as continuous trajectories in acoustic parameter space. Trajectory Clustering allows us to avoid the conditional independence assumption that makes it difficult to account for the fact that successive measurements of an articulatory gesture are correlated. We apply the trajectory clustering method for developing multiple parallel hidden Markov models (HMMs) for a continuous digits recognition task. We compare the performance obtained with data-driven clustering to the recognition performance obtained with conventional head-body-tail models, which use knowledge-based criteria for building multiple HMMs in order to obviate the trajectory folding problem. The results show that trajectory clustering is able to discover structure in the the training database that is different from the structure assumed by the knowledge-based approach. In addition, the data-derived structure gives rise to significantly better recognition performance, and results in a 10% word error rate reduction.

This paper builds on previous work aimed at unraveling the structure of the speech signal using probabilistic representations. The context of this work is a multi-pass speech recognition system in which a phone lattice is created and used as a basis for a lexical decoding pass (search) that allows symbolic mismatches at certain costs. The focus is on the optimization of the costs of the phone insertions, deletions and substitutions that are used in the lexical decoding pass. Two optimization approaches are presented, one related to a multi-pass computational model for human speech recognition, the other based on a decoding that minimizes Bayes' risks. In the final section, the advantages of the two optimization methods are discussed and compared

Recent research suggests that modeling coarticulation in speech is more appropriate at the syllable level. However, due to a number of additional factors that affect the way syllables are articulated, creating multiple paths through syllable models might be necessary. Our previous research on longer-length multi-path models in connected digit recognition has proved trajectory clustering to be an attractive approach to deriving multi-path models. In this paper, we extend our research to large vocabulary continuous speech recognition by deriving trajectory clusters for 94 very frequent syllables in a 20-hour data set of Dutch read speech. The resulting clusters are compared with a knowledge-based classification. The comparison results suggest that multi-path models for syllables are difficult to build based on phonetic and linguistic knowledge. When multi-path models based on trajectory clustering are used, speech recognition performance improves significantly. Thus, it is concluded that data-driven trajectory clustering is a very effective approach to developing multi-path models

Most large speech corpora are delivered with a lexicon that contains a canonical transcription of every word in the orthographic transcription.Such a lexicon can be used for generating a hypothetical 'canonical' phonetic transcription from the orthography.In addition, time and money permitting, some speech corpora are provided with a manually verified broad phonetic transcription of at least part of the material.Since the manual verification of phonetic transcriptions is time-consuming and expensive, we investigated whether existing automatic transcription procedures and combinations of such procedures can offer a quick and cheap alternative for the generation of phonetic transcriptions like the manually verified transcriptions delivered with large speech corpora.In our study, we used ten automatic transcription procedures to generate a broad phonetic transcription of well-prepared speech (readaloud texts) and spontaneous speech (telephone dialogues) from the Spoken Dutch Corpus.The performance was assessed in terms of the number and the nature of the discrepancies between the emerging phonetic transcriptions and the corresponding manually verified phonetic transcriptions delivered with the Spoken Dutch Corpus.The resulting automatic transcriptions appeared to be comparable to the manually verified transcriptions.

Recent research suggests that it is more appropriate to model pronunciation variation with syllable-length acoustic models than with context-dependent phones. Due to the large number of factors contributing to pronunciation variation at the syllable level, the creation of multi-path model topologies appears necessary. In this paper, we propose a novel approach for constructing multi-path models for frequent syllables. The suggested approach uses phonetic knowledge for the initialisation of the parallel paths, and a data-driven solution for their re-estimation. When applied to 94 frequent syllables in a 37-hour corpus of Dutch read speech, it leads to improved recognition performance when compared with a triphone recogniser of similar complexity.

Item does not contain fulltext

It is generally acknowledged that many experts and almost all lay persons have difficulty in formulating requests for information in such a manner that conventional off-line Information Extraction systems can find optimal answers. Therefore, it is increasingly evident that there is a need for an interactive dialog between information seekers and information extraction systems. In this paper we describe the demonstrator of an interactive and multimodal information extraction system that is under construction in the NWO funded research program IMIX.

This paper presents an exploratory study on the relations between gender and everyday parlance. A “data-mining” approach is used to explore gender-specific characteristics in a large number of spontaneous telephone and face-to-face conversations. Our study focuses on speech rate (speaking rate and articulation rate), disfluencies (filled pauses and repetitions), pronunciation variation (phoneme substitutions, deletions and insertions), and preferences for particular parts of speech. Our study reveals interesting similarities and differences in everyday male and female speech, and proves that data-mining on large spoken language corpora is a promising approach for obtaining information on spontaneous speech phenomena and for generating new hypotheses for research.

Contains fulltext : 42030.pdf (Publisher’s version ) (Closed access)

Audio recordings of speakers using speech-driven systems show phenomena that are characteristic for on-line speech responses, such as out-of-task utterances, self-talk and speech disfluencies. This paper focuses on a survey of these phenomena as they were recorded during interactions by subjects using a multimodal system, and reports on experiments concerning the treatment of these phenomena for automatic speech recognition. This study is a starting point for the study of a richer set of on-line phenomena in speech addressed to multimodal systems and the implications for automatic speech recognition.

\n Contains fulltext :\n 61762.pdf (author's version ) (Open Access)\n

Within the Aurora2 experimental framework, the aim of this study is to determine what the relative contributions of spectral shape and energy features are to the mismatch observed between clean training and noisy test data. In addition to measurements on the baseline Aurora2 system, recognition performance was also evaluated after the application of time domain noise reduction (TDNR) and histogram normalisation (HN) in the cepstral domain. The results indicate that, for the Aurora2 digit recognition task, TDNR, HN, as well as a combination of the two techniques achieve higher recognition rates by reducing the mismatch in the energy part of acoustic feature space. The corresponding mismatch reduction in the spectral shape features yields hardly any gain in recognition performance.

In this paper, we present a novel computational model of human speech recognition - called SpeM - based on the theory underlying Shortlist. We will show that SpeM, in combination with an automatic phone recogniser (APR), is able to simulate the human speech recognition process from the acoustic signal to the ultimate recognition of words. This joint model takes an acoustic speech file as input and calculates the activation flows of candidate words on the basis of the degree of fit of the candidate words with the input.\n\nExperiments showed that SpeM outperforms Shortlist on the recognition of 'real-life' input. Furthermore, SpeM performs only slightly worse than an off-the-shelf full-blown automatic speech recogniser in which all words are equally probable, while it provides a transparent computationally elegant paradigm for modelling word activations in human word recognition.

This paper presents a method designed to select a limited set of maximally information rich speech data from a database for optimal training and diagnostic testing of Automatic Speech Recognition (ASR) systems. The method uses Principal Component Analysis (PCA) to map the variance of the speech material in a database into a low-dimensional space, followed by clustering and a selection technique. It appears that a very straightforward implementation of this procedure automatically detects at least two criteria for a classification of speakers of standard Dutch, viz. gender and the way in which the /r/ is produced. To verify the power of the technique to improve ASR, data sets of equal size selected with this method and obtained randomly were used to train a recognition system on Dutch connected digits. The results show an improvement in the recognition performance when optimal data sets were used, especially for the conditions where the sub-corpora used for training were relatively small.

The current version of the psycholinguistic model of human word recognition Shortlist suffers from two unrealistic constraints. First, the input of Shortlist must consist of a single string of phoneme symbols. Second, the current version of the search in Shortlist makes it difficult to deal with insertions and deletions in the input phoneme string. This research attempts to fully automatically derive a phoneme string from the acoustic signal that is as close as possible to the number of phonemes in the lexical representation of the word. We optimised an Automatic Phone Recogniser (APR) using two approaches, viz. varying the value of the mismatch parameter and optimising the APR output strings on the output of Shortlist. The approaches show that it will be very difficult to satisfy the input requirements of the present version of Shortlist with a phoneme string generated by an APR.

This paper investigates possibilities to automatically find a low-dimensional, formant-related physical representation of the speech signal, which is suitable for automatic speech recognition (ASR). This aim is motivated by the fact that formants have been shown to be discriminant features for ASR. Combinations of automatically extracted formant-like features and `conventional', noise-robust, state-of-the-art features (such as MFCCs including spectral subtraction and cepstral mean subtraction) have previously been shown to be more robust in adverse conditions than state-of-the-art features alone. However, it is not clear how these automatically extracted formant-like features behave in comparison with true formants. The purpose of this paper is to investigate two methods to automatically extract formant-like features, and to compare these features to hand-labeled formant tracks as well as to standard MFCCs in terms of their performance on a vowel classification task.

In this paper, the accumulated Kullback divergence (AKD) is used to analyze ASR performance deterioration due to the presence of background noise. The AKD represents a distance between the feature value distribution observed during training and the distribution of the observations in the noisy test condition for each individual feature vector component. In our experiments the AKD summed over all feature vector components shows a high correlation with word error rate and AKD computed per component can be used to pinpoint those feature vector components that substantially contribute to recognition errors. It is argued that the distance measure could be a useful evaluation tool for analyzing the strengths and weaknesses of existing noise robustness approaches and might help to suggest research strategies that focus on those elements of the acoustic feature vector that are most severely affected by the noise.

Recognising directory listings for national telephone number inquiry is slowly getting within reach for modern ASR technology. Two key factors for a successful system design are (1) optimal extent of lexical modelling and (2) an effective utterance rejection method. In this paper we show how a choice for the first has consequences for the second.

This paper describes the noise robust feature extraction meth ods developed by France Telecom and Alcatel for the noise robust front-end standardisation of ETSI Aurora.It is shown that both noise reduction methods give a substantial im provement when compared to a standard MFCC feature ex traction algorithm for speech recognition in noisy environ ments.In addition, blind equalisation and feature vector se lection were used for further improvement of recognition performance.Results are discussed for the ETSI Aurora 2 task and the SDC-Italian task as well.It was found that the combi nation of noise reduction with the proposed methods is capa ble to achieve around 50% reduction of the error rate.In the context of the open ETSI Aurora standardisation, two propos als were submitted based on these methods, they achieved the best results among all the proposals.

So far most attempts to automate Directory Assistance services focused on private listings, because it is not known precisely how callers will refer to a business listings. The research described in this paper, carried out in the SMADA project, tries to fill this gap. The aim of the research is to model the expressions people use when referring to a business listing by means of rules, in order to automatically create a vocabulary, which can be part of an automated DA service. In this paper a rule-base procedure is proposed, which derives rules from the expressions people use. These rules are then used to automatically create expressions from directory listings. Two categories of businesses, viz. hospitals and the hotel and catering industry, are used to explain this procedure. Results for these two categories are used to discuss the problem of the over- and undergeneration of expressions.

\n Contains fulltext :\n 75051.pdf (author's version ) (Open Access)\n

Within the context of robust acoustic features for automatic speech recognition (ASR), we evaluated mel-frequency cepstral coefficients (MFCCs) derived from two spectral representation techniques, i.e. the fast Fourier transform (FFT) and linear predictive coding (LPC). ASR systems based on the two feature types were tested on a digit recognition task using continuous density hidden Markov phone models. System performance was determined in clean acoustic conditions as well as in different simulations of adverse acoustic conditions. The LPC-based MFCCs outperformed their FFT counterparts in most of the adverse acoustic conditions that were investigated in this study. A tentative explanation for this difference in recognition performance is given.

This paper addresses differences between operational speaker verification (SV) systems and laboratory experiments in terms of performance and methods for measuring performance. It is concluded that operational SV systems need an indication of the quality of newly enrolled speaker models, to decide whether to re-enrol or request more enrolment material. We have investigated the impact of ASR errors on model quality. While attempting to design measures for the quality of speaker models we have developed a novel method for assigning weights to the contribution of models in accordance with their discriminative ability.

Within the context of automatic speech recognition (ASR) applications for telephony, we investigate the acoustic preprocessing issues that are at stake in going from the fixed line to the cellular network. Because the spectral representation used in enhanced full rate GSM is linear prediction, we investigate the relative advantages and drawbacks of conventional mel-frequency cepstral coefficient (MFCC) parameters derived from a non-parametric fast Fourier transform (FFT) and MFCC parameters derived from a linear predictive coding (LPC) spectral estimate. Robust formant parameters, also derived from an LPC description of the spectrum, are studied as an alternative to MFCCs. Within the framework of connected digit recognition based on hidden Markov models, ASR performance was measured for clean conditions, as well as for three different additive noise conditions. In addition, the performance of a conventional recognition procedure was compared with the performance of an ASR system based on our acoustic backing-off implementation of missing feature theory (MFT).

\n Contains fulltext :\n 75039.pdf (author's version ) (Open Access)\n

There are very few, if any, published accounts of practical expe rience with Speaker Verification as a means to provide secure access to telematics services.Yet, there is no reason to expect that Speaker Verification is very different from speech recogni tion, for which many deployed services have shown the need for close and intensive on-line monitoring during the time when the service becomes operational.In this paper we present our expe rience with a monitoring scheme for Speaker Verification dur ing the field test of a financial investment game.Many of the issues that were monitored were suggested by our experience with a semi-operational service, viz.free access to Directory Assistance for visually impaired.A newly developed enrolment procedure, that can flag potentially weak speaker models, is an essential part of the monitoring procedure.

This study looks at two ways of extracting a glottal waveform from recorded speech.One way is to inverse filter the flow at the mouth.Another is to inverse filter the microphone signal.Theoretically, the microphone signal is considered to be the equivalent of a first order differentiation of the flow signal recorded at the lips.Recording the oral airflow is more complicated than the recording of a microphone signal, as it requires the use of a mask, with constant adjustments during the recording.Recording of the microphone signal is more straightforward for the experimenter and less intrusive for the subject.If the two inverse filtering procedures can be shown to produce similar glottal flow waveforms for both types of recorded speech, this would support the use of only the microphone signal for those types of glottal flow analysis where the DC component of the flow is not essetial, making voice source analysis applicable in less specialised situations.In this study, we used recordings of microphone signal and recordings of oral flow to compare the results of inverse filtering.A group of twenty subjects produced repetitions of the utterance /pae/.We recorded oral flow, EGG, and the microphone signal.The flow and microphone signals were analysed using an automatic inverse filtering program and values for parameters which are extracted from the source wave are compared.The results were not as similar as expected, although in some respects, they correlated well.This may be due to experimental design, the degree of insight of the subject into the voicing task, and the fact that the speech material used for the comparison was not identical.

The loss in performance caused by mismatch between train and test material suggests a need for task specific acoustic models, especially for highly demanding tasks.However, since the training of these models is extremely expensive, general pur pose models are more attractive.In this paper we address the impact of mismatch in speaking style and task.We trained three sets of acoustic models on data from different tasks, involving both read and extemporaneous speech.The average utterance length in the training corpora varied between 10.5 and 1.2 words.The models were tested on matched as well on very dif ferent tasks.The results suggest that general purpose models trained from short utterances are to be preferred in most spoken dialog systems.However, these models might not perform ade quately in dictation tasks.

The use of confidence measures (CMs) in spoken dialog system (SDS) applications to suppress the number of verification turns for 'reliably correctly recognised utterances' can greatly reduce average dialog length which enhances usability and increases user satisfaction. This paper gives a clear review of the method of CM assessment, which was presented by Boves, Herberts and Russel (see Proceedings of the IEEE Third Workshop Interactive Voice Technology for Telecommunications Applications, Granada, Spain, p.89-92, 1996). It proceeds by demonstrating how the Dutch ARISE (Automatic Railways Information Systems in Europe) SDS was equipped with this technology and shows in deep detail how the parameters involved are to be optimised. The evaluation reveals and explains a typical behaviour of this method with train timetable information-alike systems. This results in a set of conclusions that were not foreseen when the method was first developed for a directory information system. The paper ends with an outlook for solutions in new research directions.

The problem of labelling speaker turns by automatically segmenting and clustering a continuous audio stream is addressed. A new clustering scheme is presented and evaluated using a clustering efficiency score which treats both agglomerative and divisive clustering strategies equally. Results show an efficiency of 70% can be obtained on both manually and automatically derived segments on the 1996 Hub4 development data. For the task of identifying potentially unknown anchor speakers within broadcast news shows, the frame classification error rate is very important. To reflect this, a frame-based cluster efficiency is defined and the results show a 90% frame-based efficiency can be achieved. Finally a frame-based comparison between the manually and automatically derived segment/cluster sets shows that approximately one third of the errors are introduced during segmentation and two-thirds during clustering.

There have been many working spoken dialogue systems, but very few of them are able to handle concurrent topics consistently if the user freely surfs among different topics at any time, because it is really difficult to resolve semantic ambiguities and check the knowledge consistencies among correlated topics. The issues include not only how to handle the information across multiple topics and domains with shared slots, but how to infer a reasonable dialogue state considering the new information or inconsistencies among knowledge structures after some correlated topics are activated. In this paper, a plan-based dialogue control mechanism that is capable of handling the very complicated dialogue problem is proposed. This mechanism can keep the knowledge consistent among multiple topics and domains when the topic is switched. Also, by representing the problem-solving procedures of different goals as loadable trees, the dialogue strategies are easy to maintain and modify, and can be dynamically adapted to the users during the dialogue.

This paper explores the use of micro-prosody in improving the quality of synthesised speech in concatenated text to speech synthesis (TTS) systems. Micro-prosody are defined as prosodic signals within context-dependent triphone units and across neighbouring triphones. Micro-prosody parameters are modelled using a Markovian model whose state distributions depend on the current linguistic-prosodic state as well as the current and the neighbouring phones. The use of various speech unit selection criteria in the design of the TTS sound inventory and their effects in reducing the variance of micro-prosodic parameters in concatenated speech and on the TTS output speech are explored. The effect of the variability of the prosodic parameters of speech in the recorded samples from a given speaker, and the influence of accents, such as the US and the UK accented English, on speech prosody variability and on the design of TTS are considered.

In this paper we investigate acoustic backing-off as an operationalization of Missing Feature Theory with the aim to increase recognition robustness. Acoustic backing-off effectively diminishes the detrimental influence of outlier values by using a new model of the probability density function of the feature values. The technique avoids the need for explicit outlier detection. Situations that are handled best by Missing Feature Theory are those where only part of the coefficients are disturbed and the rest of the vector is unaffected. Consequently, one may predict that acoustic feature representations that smear local spectrotemporal distortions over all feature vector elements are inherently less suitable for automatic speech recognition. Our experiments seem to confirm this prediction. Using additive band limited noise as a distortion and comparing four different types of feature representations, we found that the best recognition performance is obtained with recognizers that use acoustic backingoff and that operate on feature types that minimally smear the distortion.

The aim of our current research is to investigate the possibility of using likelihood ratios to perform utterance verification within the context of automatic oral proficiency assessment. The likelihood ratios under investigation have the appealing feature that they may be computed simply by using an off-theshelf automatic speech recognition system in two different recognition modes (forced and free phone) instead of using a system with specifically trained anti-models. We achieved 93% correct classification for 10 phonetically rich sentences uttered by 60 non-native language students. 1. INTRODUCTION The long-term goal of our research is to employ ASR technology in an automatic pronunciation test for Dutch as a second language. As a consequence of this aim we are not concerned with learners of Dutch with a specific mother tongue, but rather with a group of speakers who are highly varied in this respect. In this sense our situation is different from that of many studies on the use of...

\n Contains fulltext :\n 74998.pdf (author's version ) (Open Access)\n

\n Contains fulltext :\n 75007.pdf (author's version ) (Open Access)\n

Phase-corrected RASTA is a new technique for channel normalization that consists of classical RASTA filtering followed by a phase correction operation. In this manner, the channel bias is as effectively removed as with classical RASTA, without introducing a left context dependency. The performance of the phase-corrected RASTA channel normalization technique was evaluated for a continuous speech recognition task. Using context-independent hidden Markov models we found that phase-corrected RASTA reduces the best-sentence word error rate (WER) by 23% compared to classical RASTA. For contextdependent models phase-corrected RASTA reduces WER by 15% compared to classical RASTA. 1. INTRODUCTION In order to reduce the linear filtering effect of communication channels, different channel normalization (CN) techniques have been proposed (e.g. [1, 2, 3]). Recently, a new, extended version of the classical RASTA filtering technique was proposed and tested in the context of connected digit recogn...

In this paper we propose to introduce backing-off in the acoustic contributions of the local distance functions used during Viterbi decoding as an operationalisation of missing feature theory for increased recognition robustness. Acoustic backing-off effectively removes the detrimental influence of outlier values from the local decisions in the Viterbi algorithm. It does so without the need for prior knowledge that specific features are missing. Acoustic backing-off avoids any kind of explicit outlier detection. This paper provides a proof of concept of acoustic backing-off in the context of connected digit recognition over the telephone, using artificial distortions of the acoustic observations. It is shown that the word error rate can be maintained at the level of 2:5% obtained for undisturbed features, even in the case where a conventional local distance computation without backing-off leads to a word error rate ? 80:0%. The approach appears to be able to handle up to four independe...

We investigate acoustic differences between vowels in syllables that do or don't carry lexical stress. The speech material on which the investigation is based differs from the type of material used in previous research: we used phonetically rich sentences from the Dutch POLYPHONE corpus. We discuss the definition of the linguistic feature 'lexical stress' and its possible impact on the phonetic realization. We then proceed to explain the experiments that were carried out and the presentation of the results. Although most of the duration, energy and spectral tilt features that we used in the investigation show statistically significant differences for the population means for stressed and unstressed vowels, it also appears that the distributions overlap to such an extent that automatic detection of stressed and unstressed syllables yields accuracy scores of not much more than 65%. It is argued that this is due to the large variety in the ways in which the abstract linguistic feature 'lexical stress' is realized in the acoustic speech signal.

In this paper we propose an extension to the classical RASTA technique. The new method consists of classical RASTA filtering followed by an all-pass phase correction filter. In this manner, the influence of the communication channel is as effectively removed as with classical RASTA. However, our proposal does not introduce a left-context dependency like classical RASTA. Therefore the new method is better suited for automatic speech recognition based on context-independent modeling with Gaussian mixture hidden Markov models. We tested this in the context of connected digit recognition over the phone. In case we used context-dependent hidden Markov models (i.e., word models), we found that classical RASTA and phase-corrected RASTA performed equally well. For context-independent phone-based models, we found that phase-corrected RASTA can outperform classical RASTA depending on the acoustic resolution of the models.

Item does not contain fulltext

One of the frequently used assumptions in Speaker Verification is that two speech segments (phonemes, subwords, words) are considered to be independent. And therefore, the log-likelihood of a test utterance is just the sum of the log-likelihoods of the speech segments in that utterance. This paper reports about cases in which this observation-independence assumption seems to be violated, namely for those test utterances which call a certain speech model more than once. For example, a pin code which contains a non-unique digit set performs worse in verification than a pin code which consists of four different digits. Results illustrate that violating the independence assumption too much might result in increasing EERs while more information (in form of digits) is added to the test utterance. 1. INTRODUCTION In Speaker Verification (SV) systems using passwords in the form of fixed or prompted digit strings it seems usual to compute the log-likelihood of a claimant being the true speaker...

In this paper we propose a new measure to classify speakers with respect to their behaviour in speaker recognition systems. Taking the proposal made by EAGLES as a point of departure we show that it fails to yield results that are consistent between closely related speaker recognition methods and between different amounts of speech available for the recognition task. We show that measures based on a straightforward confusion matrices, that take only the 1-best classification into account, cannot result in consistent classifications. As an alternative we propose a measure based on n-best scores in a speaker identification paradigm, and show that it yields more consistent performance. 1. CONTEXT In Speaker Verification (SV) research it has been customary to base comparisons and evaluations on Equal Error Rates (EER) [2]. Recently, Receiver Operator Curves (ROC) seem to gain popularity, as a richer representation of performance results. However, ROC representations are still based on re...

\n Contains fulltext :\n 74985.pdf (author's version ) (Open Access)\n

In this research we have studied several human factors problems that are connected to the deployment of speaker verification technology in telecommunication services. We investigate the perception of the safety of a calling card service when it is protected by speaker verification on the 14 digit card number, and compare it to the perceived safety of speaker verification and PIN. Moreover, we compare a voice based interface to the service with a DTMF based interface. The results are crucial for guiding the introduction and deployment of speaker verification technology in actual applications. 1. INTRODUCTION Of all Speech Technology fields Speaker Recognition has received least attention over the past decade. Up to now hardly any research results have been published in the open literature on the performance and acceptance of speaker verification technology in telephone services for the general public. The LE project CAVE (Caller Verification in Telecommunication and Banking, LE1-1930) ...

The acoustic realization of vowels with lexical stress generally differs substantially from their unstressed counterparts, which are more reduced in spectral quality, shorter in duration, weaker in intensity and tend to have a flatter spectral tilt.Therefore, in an automatic speech recognizer it would appear profitable to train separate models for the stressed and unstressed variants of each vowel.A problem is how to define the mapping from the theoretical stress of words to the actual realization of stress in fluent speech.We compared several hypotheses about this mapping applied in both training and testing of the recognizer.The recognition results on an independent test-set showed that recognition rates did not increase by this use of stress in our ASR.Possible explanations are discussed and future research plans are outlined.

This paper reports on the development o f a spoken dialogue system for providing information about public transport in the Netherlands.It is explained how a German prototype was adapted for Dutch.Emphasis is laid on the specific approach chosen to collect speech material that could be used to gradually improve the system.The pros and cons of this method are discussed.

We compared three different channel normalisation (CN) meth ods in the context of a connected digit recognition task over the phone: ceptrum mean substraction (CMS), RASTA filtering and the Gaussian dynamic cepstrum reprsentation (GDCR).Using a small set of context-independent (CI) continuous Gaussian mixture hid den Markov models (HMMs) we found that CMS and RASTA out performed the GDCR technique.We show that the main cause for the superiority of CMS compared to RASTA is the phase distortion introduced by the RASTA filter.Recognition results for a phasecorrected RASTA technique are identical to those of CMS.Our re sults indicate that an ideal cepstrum based CN method should (1) effectively remove the DC-component, (2) at least preserve modu lation frequencies in the range 2-16 Hz and (3) introduce no phase distortion in case CI HMMs are used for recognition.

Voice source parameters can be estimated by fitting a voice source model to the glottal flow signal which is obtained by means of inverse filtering. In this paper we investigate the behaviour of the LF-model in a number of non-linear parameter estimation procedures. It is concluded that (1) the parameter estimates are robust against additive (white and narrow band) noise in the flow waveforms, (2) simplex search algorithms perform better than steepest descent algorithms, provided that (3) the LF-pulse is generated with an algorithm that treats all parameters as real numbers. 1.

Item does not contain fulltext

Item does not contain fulltext

Item does not contain fulltext

Item does not contain fulltext

Item does not contain fulltext

The experiments described show that RARMA (robust autoregressive moving average) analysis can be applied to the extraction of speech synthesis rules for nasal consonants that bear relationship to articulatory gestures. This is an important advantage in comparison to the results of the usual all-pole techniques. The poles and zeros resolved by RARMA could be directly translated into target frequencies for the vocal tract (anti-)resonators in this synthesis model. Inspection of short-term FFTs of natural speech utterances showed that during the realization of the nasal consonant an additional zero-per-pole is present around 800 Hz that is not resolved by RARMA. This special event rides on the steep spectral slope of the first resonance peak in the nasal consonant.>

In text-to-speech systems, syntactic knowledge is necessary for correct grapheme-to-phoneme conversion, and calculation of prosody. Markov models can assign the correct word classes needed for grapheme-to-phoneme conversion in most of the cases, but only a small part of the information needed for correct prosody. Conventional syntactic parsers almost never produce unique output: they either over-generate or fail to yield a syntactic analysis. Moreover, they cannot handle input that is not syntactically correct, and they are slow and expensive. Wild card parsing (WCP) can overcome these problems and provide the syntactic information needed for prosody calculations in unrestricted tests. The WCP system is implemented in a language-independent way. All language-dependent knowledge is kept strictly separated from the program. The WCP system has been integrated into a multilingual text-to-speech system where it provides syntactic information for the generation of prosodic contours for any input sentence. As a by-product of the syntactic parsing it disambiguates the word classes of the words in the sentence, thus helping grapheme-to-phoneme conversion.>

Item does not contain fulltext

Item does not contain fulltext

Item does not contain fulltext

Item does not contain fulltext

Two methods for deriving synthesizer control rules are discussed. In one approach a trained phonetician interprets the results of a robust autoregressive moving average (RARMA) analysis and makes a separation between spectral events attributable to the source and the vocal tract, respectively, in order to derive rules for controlling the poles and zeros in the tract model. This procedure appears to be fruitful but rather time-consuming. In the other approach an attempt is made to support the procedure as much as possible by automatic enhancement of the analysis results. The experiments indicate that RARMA analysis delivers reliable pole and zero tracks that can be associated with dynamic characteristics of the vocal tract.>

In order to derive voice source control rules from natural speech, the parameters of a source model must be derived from the acoustic signal. This is done by parameterizing the results of glottal inverse filtering. A number of different inverse filtering procedures and the ease with which their results can be parameterized are compared. It is shown that closed glottic interval covariance linear predictive coding is as powerful as more sophisticated techniques, because it is the only known method that can strictly be limited to the closed glottis interval.>

Item does not contain fulltext

Item does not contain fulltext

DEMSI is a database environment for the management of speech-related information for acoustic phonetic research. Structured around a relational DBMS, DEMSI offers a flexible and comprehensive framework for the storage and retrieval of a wide range of linguistic, phonetic and acoustic parameters. The structure of this database is discussed with respect to the entities that are defined and their relationships. Special attention is given to speech specific characteristics inherent in this type of data.>

Theoretical studies of the voice source in speech production are hampered by the lack of empirical measurements of the aerodynamic events around the glottis. This paper describes an attempt to obtain direct measurements of pressure-flow relations just below and above the glottis during normal speech production. To that end high resolution pressure recordings were made at two points in the trachea and in the pharynx. From these recordings accurate estimates of the flow in the trachea and pharynx can be obtained. It is shown how these estimates can be converted into estimates of transglottal flow. It appeared that the estimates of transglottal flow from tracheal flow are not dependent on the vowel produced. The estimates based on pharyngeal flow, on the other hand, depend to a large extent on the vowel. These findings are interpreted in the framework of acoustic models of speech production.

Over the years a number of inverse filtering methods for recovering the volume flow at the glottis from the acoustic speech signal have been designed. Formal tests of these methods have, however, been the exception rather than the rule. We describe an attempt to use registrations of the electroglottogram, photo-electroglottogram and the subglottal pressure, recorded at the same time as the speech signals, to establish the instants of opening, closing and maximal width of the glottis. Although the determination of those epochs from the registrations proved to be less straighforward than anticipated we have succeeded in showing their usefulness for evaluating the accuracy of flow pulses reconstructed by means of inverse filtering techniques.