Short Utterances Research Articles

A Unified Joint Model to Deal With Nuisance Variabilities in the i-Vector Space

The past decade has witnessed a significant improvement in speaker recognition SR technology in terms of performance with the introduction of the i-vectors framework. Despite these advances, the performance of SR systems considerably suffers in the presence of acoustic nuisances and variabilities. In this paper, we develop a data-driven nuisance compensation technique in the i-vector space without referring to the effects of the targeted nuisances in the temporal domain. This approach is nonparametric as it does not suppose a specific relationship between a “good” version of an i-vector and its corrupted version. Instead, our algorithm models directly the joint distribution of both representations the good i-vector and its corrupted version and takes advantage of the reproducibility of acoustic corruptions to generate the corrupted i-vectors. We then build an MMSE estimator that computes an improved version of a corrupted test i-vector, given this joint distribution. Experiments are carried out on NIST SRE 2010 and speakers in the wild databases where the proposed algorithm is used to deal with additive noise and short utterances. Our technique is shown to be efficient, improving the baseline system performance in terms of equal-error rate by up to 70% when used on known test noises and up to 65% in the context of unseen noises using a generic model. It was also proven efficient in the context of duration mismatch reaching up to 40% of relative improvement when used on short utterances using multiple models corresponding to different durations and up to 36% when used on arbitrary duration test segments.

Nonparametrically trained PLDA for short duration i-vector speaker verification

The duration of speech segments can significantly impact the performance of text-independent speaker verification systems. In real world applications which require high accuracy on short utterances, the performance of i-vector speaker verification framework degrades significantly considering that i-vectors extracted from short utterances are less reliable (i.e., uncertainty is higher) than those extracted from long utterances. Therefore, to handle duration variability properly, a more realistic approach seems to be required. This study is an extension to our recently proposed nearest neighbor probabilistic linear discriminant analysis (NN-PLDA) which estimates the parameters of PLDA in i-vector speaker verification framework using a nonparametric form rather than maximum likelihood estimation (MLE) obtained by an EM algorithm, and has been shown to provide superior performance. In NN-PLDA, the between-speaker covariance matrix that represents global information about the speaker variability is replaced with a local estimation computed on a nearest neighbor basis for each target speaker. Compared to their parametric counterparts, the nonparametric between- and within-speaker scatter matrices can better exploit the discriminant information in training data and are more adapted to sample distributions. In this paper, we provide further analysis on the proposed nonparametrically trained PLDA as well as introduce a duration variability modeling technique in the estimation of the within-speaker scatter matrix as to compensate for the effect of limited speech data. We evaluate our approach using core–10sec and 10sec–10sec telephone trial conditions of NIST 2010 SRE as well as on the truncated test utterances in extended core condition with duration less than 10 s. We also present the results obtained by the successful incorporation of NN-PLDA on the recent NIST 2016 speaker recognition evaluation. In all experiments, considerable performance improvement is obtained with the proposed technique compared to a generatively trained PLDA model.

Phonetic Temporal Neural Model for Language Identification

Deep neural models, particularly the LSTM-RNN model, have shown great potential for language identification (LID). However, the use of phonetic information has been largely overlooked by most existing neural LID methods, although this information has been used very successfully in conventional phonetic LID systems. We present a phonetic temporal neural model for LID, which is an LSTM-RNN LID system that accepts phonetic features produced by a phone-discriminative DNN as the input, rather than raw acoustic features. This new model is similar to traditional phonetic LID methods, but the phonetic knowledge here is much richer: it is at the frame level and involves compacted information of all phones. Our experiments conducted on the Babel database and the AP16-OLR database demonstrate that the temporal phonetic neural approach is very effective, and significantly outperforms existing acoustic neural models. It also outperforms the conventional i-vector approach on short utterances and in noisy conditions.

Age Estimation in Short Speech Utterances Based on LSTM Recurrent Neural Networks

Age estimation from speech has recently received increased interest as it is useful for many applications such as user-profiling, targeted marketing, or personalized call-routing. This kind of applications need to quickly estimate the age of the speaker and might greatly benefit from real-time capabilities. Long short-term memory (LSTM) recurrent neural networks (RNN) have shown to outperform state-of-the-art approaches in related speech-based tasks, such as language identification or voice activity detection, especially when an accurate real-time response is required. In this paper, we propose a novel age estimation system based on LSTM-RNNs. This system is able to deal with short utterances (from 3 to 10 s) and it can be easily deployed in a real-time architecture. The proposed system has been tested and compared with a state-of-the-art i-vector approach using data from NIST speaker recognition evaluation 2008 and 2010 data sets. Experiments on short duration utterances show a relative improvement up to 28% in terms of mean absolute error of this new approach over the baseline system.

LID-Senones and Their Statistics for Language Identification

Recent research on end-to-end training structures for language identification has raised the possibility that intermediate language-sensitive feature units exist which are analogous to phonetically sensitive senones in automatic speech recognition systems. Termed language identification (LID)-senones, the statistics derived from these feature units have been shown to be beneficial in discriminating between languages, particularly for short utterances. This paper examines the evidence for the existence of LID-senones before designing and evaluating LID systems based on low- and high-level statistics of LID-senones with both generative and discriminative models. For the standard NIST LRE 2009 task on 23 languages, LID-senone-based systems are shown to outperform state-of-the-art deep neural network/i-vector methods both when LID-senones are used directly for classification and when LID-senone statistics are used for i-vector formation.

DEIXIS FOUND IN KAMIO YOUKO’S HANA YORI DANGO IN ENGLISH VERSION

Deixis cannot be separated to language itself because deixis is a part of language. One of language forms in our environment are comics. It is the popular literary works that teenager read besides novel and short story. The objectives of this research can be further detailed as the following; (1) To find kinds of deixis that are in Kamio Yuoko’s Hana Yori Dango. (2) To know the situation of deixis are in Kamio Youko’s Hana Yori Dango. The design of this research is descriptive qualitative research. It describes deixis found in Hana Yori Dango comic. This research uses the descriptive qualitative research. The result of this research are; (1) The kinds of English deixis found in Hana Yori Dango comic are; eighty five person deixis, twelve place deixis, thirteen time deixis, thirteen social deixis, and seventeen discourse deixis. So, it can be concluded the the dominan deixis which is used in Hana Yori Dango comic is person deixis and the minor deixis is place deixis and (2) The dominant situation in Hana Yori Dango comic is utterance situation related to person and social deixis in finding. So, the theory of deixis analysis from Jaszczolt (2002, p.191) is sticked with the result of deixis found in Hana Yori Dango comic. In this study, the kinds of deixis most commonly found is person deixis. It indicates that the author extremely want to show the figures in the story of comic. It can be seen from the portion of conversation made by the author is simple and short utterance and also related to some elements in comic itself.

Improved i-vector extraction technique for speaker verification with short utterances

A major challenge in ASV is to improve performance with short speech segments for end-user convenience in real-world applications. In this paper, we present a detailed analysis of ASV systems to observe the duration variability effects on state-of-the-art i-vector and classical Gaussian mixture model-universal background model (GMM-UBM) based ASV systems. We observe an increase in uncertainty of model parameter estimation for i-vector based ASV with speech of shorter duration. In order to compensate the effect of duration variability in short utterances, we have proposed adaptation technique for Baum-Welch statistics estimation used to i-vector extraction. Information from pre-estimated background model parameters are used for adaptation method. The ASV performance with the proposed approach is considerably superior to the conventional i-vector based system. Furthermore, the fusion of proposed i-vector based system and GMM-UBM further improves the ASV performance, especially for short speech segments. Experiments conducted on two speech corpora, NIST SRE 2008 and 2010, have shown relative improvement in equal error rate (EER) in the range of 12–20%.

Speaker verification with short utterances: a review of challenges, trends and opportunities

Automatic speaker verification (ASV) technology now reports a reasonable level of accuracy in its applications in voice-based biometric systems. However, it requires adequate amount of speech data for enrolment and verification; otherwise, the performance becomes considerably degraded. For this reason, the trade-off between the convenience and security is difficult to maintain in practical scenarios. The utterance duration remains a critical issue while deploying a voice biometric system in real-world applications. A large amount of research work has been carried out to address the limited data issue within the scope of SV. The advancements and research activities in mitigating the challenges due to short utterance have seen a significant rise in recent times. In this study, the authors present an extensive survey of SV with short utterances considering the studies from recent past and include latest research offering various solutions and analyses. The review also summarises the major findings of the studies of duration variability problem in ASV systems. Finally, they discuss a number of possible future directions promoting further research in this field.

Multi-style speaker recognition database in practical conditions

This work describes the process of collection and organization of a multi-style database for speaker recognition. The multi-style database organization is based on three different categories of speaker recognition: voice-password, text-dependent and text-independent framework. Three Indian institutes collaborated for the collection of the database at respective sites. The database is collected over an online telephone network that is deployed for speech based student attendance system. This enables the collection of data for a longer period from different speakers having session variabilities, which is useful for speaker verification (SV) studies in practical scenario. The database contains data of 923 speakers for the three different modes of SV and hence termed as multi-style speaker recognition database. This database is useful for session variability, multi-style speaker recognition and short utterance based SV studies. Initial results are reported over the database for the three different modes of SV. A copy of the database can be obtained by contacting the authors.

Alterations in Spanish Language Interpretation During Pediatric Critical Care Family Meetings.

To characterize alterations in Spanish language medical interpretation during pediatric critical care family meetings. Descriptive, observational study using verbatim transcripts of nine PICU family meetings conducted with in-person, hospital-employed interpreters. A single, university-based, tertiary children's hospital. Medical staff, family members, ancillary staff, and interpreters. None. Interpreted speech was compared with original clinician or family speech using the qualitative research methods of directed content analysis and thematic analysis. Alterations occurred in 56% of interpreted utterances and included additions, omissions, substitutions, editorializations, answering for the patient/clinician, confessions, and patient advocacy. Longer utterances were associated with more alterations. To minimize interpreter alterations during family meetings, physicians should speak in short utterances (fewer than 20 words) and ask interpreters to interrupt in order to facilitate accurate interpretation. Because alterations occur, physicians may also regularly attempt to assess the family's understanding.

EXPLICIT BACK-CHANNEL STRATEGY TRAINING AND IMPROVEMENT OF SPEAKING SKILL: CASE OF IRANIAN EFL LEARNERS

This study explores the effect of explicit back-channel strategy training on Iranian EFL language learners� success in speaking skill. Back-channels are turn taking responses uttered by the listeners. Forty homogenous learners (20 male, 20 female) were randomly assigned to four groups (2 experimental, 2 control). They were at intermediate level in terms of general English proficiency based on the results of Oxford Placement Test (OPT) and pre-test oral Proficiency Interview. After 18 sessions of treatment of back-channels for experimental groups, that were based on the methodological theory of English language teaching suggested by Doff (1990) and Harmer (1991), post- test oral proficiency test was administered for experimental and control groups. The findings of pretest and posttest oral exams revealed that EFL language learners� oral performance (speaking) in experimental groups improved. The result revealed that females employed back-channels more often than males when they were participating in a conversation. The type of back-channel both genders employed the most was short utterances, such as �yes�, and the most used function of back-channel was request for clarification. The results of this thesis will provide language teachers, EFL language learners, researchers, material developers, and readers, useful information about the types and functions of back channels that are needed to develop speaking ability appropriately through explicit teaching.Keywords: back channel strategy, communication strategy, explicit training

Incorporating pass-phrase dependent background models for text-dependent speaker verification

In this paper, we propose pass-phrase dependent background models (PBMs) for text-dependent (TD) speaker verification (SV) to integrate the pass-phrase identification process into the conventional TD-SV system, where a PBM is derived from a text-independent background model through adaptation using the utterances of a particular pass-phrase. During training, pass-phrase specific target speaker models are derived from the particular PBM using the training data for the respective target model. While testing, the best PBM is first selected for the test utterance in the maximum likelihood (ML) sense and the selected PBM is then used for the log likelihood ratio (LLR) calculation with respect to the claimant model. The proposed method incorporates the pass-phrase identification step in the LLR calculation, which is not considered in conventional standalone TD-SV systems. The performance of the proposed method is compared to conventional text-independent background model based TD-SV systems using either Gaussian mixture model (GMM)-universal background model (UBM) or hidden Markov model (HMM)-UBM or i-vector paradigms. In addition, we consider two approaches to build PBMs: speaker-independent and speaker-dependent. We show that the proposed method significantly reduces the error rates of text-dependent speaker verification for the non-target types: target-wrong and impostor-wrong while it maintains comparable TD-SV performance when impostors speak a correct utterance with respect to the conventional system. Experiments are conducted on the RedDots challenge and the RSR2015 databases that consist of short utterances.

Language Samples From Children Who Use Speech-Generating Devices: Making Sense of Small Samples and Utterance Length.

Mean length of utterance in morphemes (MLUm) is underreported in people who use augmentative and alternative communication (AAC). MLUm is difficult to measure in people who use AAC because of 2 challenges described in literature: the challenge of small language samples (difficulty collecting representative samples) and the challenge of transcribing short utterances (difficulty transcribing 1-morpheme utterances). We tested solutions to both challenges in a corpus of language samples from children who use speech-generating devices. The first challenge was addressed by adjusting the length of the sampling window to obtain representative language samples. The second challenge was addressed by using mean syntactic length (MSL) as an alternative to MLUm. A 24-hour sample window consistently failed to yield representative samples. An extended 1-month sample window consistently yielded representative samples. A significant positive prediction of MLUm by MSL was found in a normative sample. Observed measures of MSL were used to predict MLUm in representative language samples from children who use AAC. Valid measures of utterance length in people who use AAC can be obtained using extended sampling windows and MSL. Research is needed to characterize the strengths and limitations of both solutions.

Transfer learning for PLDA-based speaker verification

Currently, the majority of the state-of-the-art speaker verification systems are based on i-vector and PLDA; however, PLDA requires a huge volume of development data from multiple different speakers. This makes it difficult to learn PLDA parameters for a domain with scarce data. In this paper, we study and extend an effective transfer learning method based on Bayesian joint probability, in which the Kullback–Leibler (KL) divergence between the source domain and the target domain is added as a regularization factor. This method utilizes the development data from the source domain to help find the optimal PLDA parameters for the target domain. Specifically, speaker verification of short utterances can be viewed as a task in the domain with a limited amount of long utterances. Therefore, transfer learning for PLDA can also be adopted to learn discriminative information from other domains with a great deal of long utterances. Experimental results based on the NIST SRE and Switchboard corpus demonstrate that the proposed method offers a significant performance gain when compared with the traditional PLDA.

Text-independent speaker identification using robust statistics estimation

It is well-known that the performance of Gaussian mixture model-based text-independent speaker identification systems deteriorates significantly with the presence of noise and spectral distortion in the training and testing utterances. In this paper, we propose a novel GMM-based speaker identification system based on two robust-statistics estimation methods: the minimum volume ellipsoid method, and the minimum covariance determinant method. Compared to the traditional maximum likelihood estimation method, the proposed methods are less sensitive to outliers in the feature-vector space caused by additive noise and spectral distortion. Moreover, in the testing phase, we propose a simple distance metric to be used for comparing the unknown testing utterance against the speakers’ models. Furthermore, we derive a more robust version of the i-vector extractor, named robust i-vector, which utilizes our proposed robust estimation methods for estimating the parameters of the base universal background model. The proposed classification system has been applied to the NIST 2000 speaker recognition evaluation and the COSINE database. It has also been compared against state-of-the-art techniques such as the GMM/UBM method, the super-vectors method, and the i-vector methods. Experimental results show that the proposed classification system provides up to 16% relative improvement in the identification performance over the i-vector methods for short utterances in the NIST 2000 database and up to 8% when the utterances of the NIST 2000 database are contaminated by different types of artificial noise for signal-to-noise ratio ranging from 0 to 20 dB. For the COSINE database, the robust i-vector estimation provides an absolute improvement of up to 8%. Finally, the real time factor of the proposed distance metric for testing is 55% higher than the RT of the regular ML scoring.

CHILD COMPREHENSION OF ADULTS’ VERBAL INPUT: A CASE OF BILINGUAL ACQUISITION IN INFANCY

Research concerning comprehension in early simultaneous bilingualism is still very limited. Thus, this study focuses on describing a bilingual infant’s comprehension of adults’ verbal input addressed to the child in an Indonesian-German language environment, and the child’s understanding of translation equivalents (TEs). The child, who was exposed to Indonesian and German simultaneously from birth, was observed from age 0;9 to age 1;3 using a diary supplemented with weekly video recordings. A “one parent-one language” system was applied in which the child received Indonesian language from the mother and German language from the father from birth. Since the family live in Indonesia and have regular contact to the collective family members, the child received dominant exposure in Indonesian compared to German. The data was transcribed and analysed using ELAN. The results show that the adults’ verbal inputs in the form of speech addressed to the child were in the form of short utterances which very often had a high-pitched sound and were rich in repetition. The adults’ speech was able to be discriminated by the child. In the pre-production stage, the child could understand approximately 6 (six) proper nouns, 18 (eighteen) Indonesian words and 14 (fourteen) German words. The result reveals that the child could comprehend more words in Indonesian than in German. It was also found that the child could understand some bilingual synonyms, which implies that at the pre-production stage, the child already went through a process of bilingual development.

Pidgin in Creative Works in English in Cameroon

Summary This study, which deals with code-switching and language choice in multilingual contexts, describes the use of Pidgin in creative works in English in Cameroon, with the focus on the forms that this language takes in the works, the types of characters who are made to speak this language, and the functions that this language plays in these works. The data comprise three plays and two novels, all published between 2000 and 2006 by experienced writers who have a good command of English and yet make their characters speak in Pidgin. The analysis shows that Pidgin in the corpus takes the form of individual lexemes like salaka (libation, sacrifice) and relatively short utterances like This sun fit kill man (This sun is so hot that it can kill someone.). The characters who speak Pidgin in these literary works are generally low-ranking and rural people, illiterates and other people who are hardly looked up to in the Cameroonian society. Finally, Pidgin helps writers to realise some stylistic effects such as variations on the scale of formality, with English being used when addressing a superior person and Pidgin when addressing an inferior person. Most importantly, creative writers reproduce in their works what is observed in the Cameroonian society and this can be regarded as a formal way of enhancing their readers’ plurilingual competence.

Influence of noise on the speaker verification in the air traffic control voice communication

The voice communication between pilots and the air traffic controllers is vulnerable to various types of attacks. Speaker verification could be used as an add-on security feature; however, there are several factors that make the use of voice biometry in this scenario difficult to apply. These are among others: open set of speakers, very short utterances, speaker noises, signal clipping, foreign accent of non-native speakers, and high content of background and channel noises in the signal. This paper identifies sources of noise in the entire communication channel and analyzes the influence of these noise components of different types and levels on the reliability of the speaker verification. An i-vector based speaker recognizer with PLDA scoring is used for the experiments. Cockpit noises of several aircrafts and limited-band channel noises are simulated by a software noise-generator. The sensitivity of the speaker verification to the noises of different frequency bands is studied in comparison to the long-term speech spectrum and its variability. Possible measures for increasing the noise robustness of the system are discussed.

A study on the effects of using short utterance length development data in the design of GPLDA speaker verification systems

This paper studies the performance degradation of Gaussian probabilistic linear discriminant analysis (GPLDA) speaker verification system, when only short-utterance data is used for speaker verification system development. Subsequently, a number of techniques, including utterance partitioning and source-normalised weighted linear discriminant analysis (SN-WLDA) projections are introduced to improve the speaker verification performance in such conditions. Experimental studies have found that when short utterance data is available for speaker verification development, GPLDA system overall achieves best performance with a lower number of universal background model (UBM) components. As a lower number of UBM components significantly reduces the computational complexity of speaker verification system, that is a useful observation. In limited session data conditions, we propose a simple utterance-partitioning technique, which when applied to the LDA-projected GPLDA system shows over 8% relative improvement on EER values over baseline system on NIST 2008 truncated 10---10 s conditions. We conjecture that this improvement arises from the apparent increase in the number of sessions arising from our partitioning technique and this helps to better model the GPLDA parameters. Further, partitioning SN-WLDA-projected GPLDA shows over 16% and 6% relative improvement on EER values over LDA-projected GPLDA systems respectively on NIST 2008 truncated 10---10 s interview-interview, and NIST 2010 truncated 10---10 s interview-interview and telephone-telephone conditions.

The perception and interpretation of contrastive focus by Polish children and adults

AbstractThe study aimed at testing Polish children’s and adult’s ability to identify and interpret three different contrastive focus types; narrow final focus, narrow non-final focus, and broad focus. In the first experiment eight-year-old children (N = 15) and young adults (N = 15, age range 19–23) were asked to spot the focus in short utterances, and produce contrastive elements in response to three types of inputs with three distinct focus types. In the second experiment eight and four year old children (N = 14 and N = 15, respectively) and young adults (N = 16) listened to two types of utterances – with final contrastive and non-final contrastive focus and were asked to select a picture that best matched the implied meaning of the auditory input. Although in each group there were individuals with ceiling results and chance results, on average the adults outperformed the children in all the tasks. Implications of the findings for testing prosody development are discussed.