Search

Simultaneous speech of multiple speakers is known as overlapped speech, which causes problems for speech recognition and speaker diarization systems. The present work uses previously less utilized signal phase information in the task of overlapped speech detection. In this context, Instantaneous Frequency Cosine Coefficient (IFCC) and Modified Group Delay Cepstral Coefficient (MGDCC) features are explored. IFCC captures the time-varying phase characteristics, while MGDCC represents the frequency-varying information of the phase spectrum. A Convolutional Neural Network and Long Short-Term Memory (CNN-LSTM)-based classifier is used for the classification. The present work uses synthetically generated overlapped speech from the GRID corpus. The proposed method is benchmarked against three baseline approaches that use magnitude spectrum features. It is observed that the combination of IFCC and MGDCC features with CNN-LSTM classifier provides better performance than the baselines. The combination of phase features with magnitude-based MFCC feature provides the best performance, indicating the importance of complementary information. The present study also investigates the effect of segment duration, genders, and number of simultaneous speakers on the overlapped speech detection system. Finally, the proposed method is also evaluated on real overlapped data from the AMI corpus.

This paper reports the findings of an automatic dialect identification (DID) task conducted on Ao speech data using source features. Considering that Ao is a tone language, in this study for DID, the gammatonegram of the linear prediction residual is proposed as a feature. As Ao is an under-resourced language, data augmentation was carried out to increase the size of the speech corpus. The results showed that data augmentation improved DID by 14%. A perception test conducted on Ao speakers showed better DID by the subjects when utterance duration was 3 s. Accordingly, automatic DID was conducted on utterances of various duration. A baseline DID system with the Slmsfeature attained an average F1-score of 53.84% in a 3 s long utterance. Inclusion of source features, Silprand [Formula: see text], improved the F1-score to 60.69%. In a final system, with a combination of Silpr, [Formula: see text], Slms, and Mel frequency cepstral coefficient features, the F1-score increased to 61.46%.

The cleft of the lip and palate (CLP) is a congenital disability affecting the craniofacial region and it impacts the speech production system. The current work focuses on the modification of misarticulations produced for unvoiced stop consonants in CLP speech. Three types of misarticulations are studied: glottal, palatal, and velar stop substitutions. The stop consonants are misarticulated due to inadequate buildup of intra-oral pressure caused by velopharyngeal dysfunction and oro-nasal fistula. The misarticulated stops affect the speech intelligibility and quality, and this further affects the use of speech-based applications. The misarticulated stops are analyzed and modified using the speech data collected from 60 Kannada speaking children (normal and CLP). An event-based modification approach is used to correct the misarticulated stops. At first, automatic detection of burst onset and vowel onset events is carried out. Then, the region from vowel onset to 20 ms duration of the vowel is extracted. Further, the region from burst onset point to 20 ms duration of the vowel is defined as the region for modification. It is transformed using the nonnegative matrix factorization (NMF) method. The objective and subjective evaluation results show that the proposed event-based transformation approach provides a relative improvement compared to the entire-word modification (signal processed without using the knowledge of burst and vowel onset events). The event-based transformed misarticulated stops showed close similarity with the normal stops in perceptual quality. The improved performance accuracy of modified stops suggests that the speech distortion is minimized.

Smartphones have been employed with biometric-based verification systems to provide security in highly sensitive applications. Audio-visual biometrics are getting popular due to their usability, and also it will be challenging to spoof because of their multimodal nature. In this work, we present an audio-visual smartphone dataset captured in five different recent smartphones. This new dataset contains 103 subjects captured in three different sessions considering the different real-world scenarios. Three different languages are acquired in this dataset to include the problem of language dependency of the speaker recognition systems. These unique characteristics of this dataset will pave the way to implement novel state-of-the-art unimodal or audio-visual speaker recognition systems. We also report the performance of the bench-marked biometric verification systems on our dataset. The robustness of biometric algorithms is evaluated towards multiple dependencies like signal noise, device, language and presentation attacks like replay and synthesized signals with extensive experiments. The obtained results raised many concerns about the generalization properties of state-of-the-art biometrics methods in smartphones.

Discrimination between shouted and normal speech is an essential prerequisite for many speech processing applications. Existing works have established that excitation source information plays a significant role in shouted speech production. In speech processing literature, various features have been proposed to model different aspects of the excitation source. The principal contribution of this work is to explore three such features, Discrete Cosine Transform of Integrated Linear Prediction Residual (DCT-ILPR), Mel-Power Difference of Spectrum in Sub-bands (MPDSS), and Residual Mel-Frequency Cepstral Coefficient (RMFCC), for shouted and normal speech classification. The DCT-ILPR feature represents the shape of the glottal cycle, MPDSS estimates the periodicity of the excitation source spectrum, and RMFCC characterizes smoothed spectral information of the excitation source. The authors have also contributed a dataset containing shouted and normal speech. This work is evaluated on three datasets and benchmarked against three baseline methods. Deep neural networks are used to study the classification performance of individual features and their combinations. The generalization performance of features (and combinations) is also investigated. Fusion of excitation source features with Mel-Frequency Cepstral Coefficients (MFCC) provides the best performance compared to other combinations. Noise analysis shows that adding excitation features with MFCC+ Δ Δ provides a more robust classification system.

The presence of hypernasality in repaired cleft palate (CP) speech is a consequence of velopharyngeal insufficiency. The coupling of the nasal tract with the oral tract adds nasal formant and antiformant pairs in the hypernasal speech spectrum. This addition deviates the spectral and linear prediction (LP) residual characteristics of hypernasal speech compared to normal speech. In this work, the vocal tract constriction feature, peak to side-lobe ratio feature, and spectral moment features augmented by low-order cepstral coefficients are used to capture the spectral and residual deviations for hypernasality detection. The first feature captures the lower-frequencies prominence in speech due to the presence of nasal formants, the second feature captures the undesirable signal components in the residual signal due to the nasal antiformants, and the third feature captures the information about formants and antiformants in the spectrum along with the spectral envelope. The combination of three features gives normal versus hypernasal speech detection accuracies of 87.76%, 91.13%, and 93.70% for /a/, /i/, and /u/ vowels, respectively, and hypernasality severity detection accuracies of 80.13% and 81.25% for /i/ and /u/ vowels, respectively. The speech data are collected from 30 control normal and 30 repaired CP children between the ages of 7 and 12.

Ao is a Tibeto-Burman language spoken in Nagaland, India. It is a low resource, tonal language with three lexical tones, namely, high, mid, and low. However, tone assignment on lexical words may differ among the three dialects of Ao, namely, Chungli, Mongsen, and Changki. In this work, an acoustic study is conducted on the three tones in the three dialects of Ao. It was found that the acoustic characteristics of the tones in the Changki dialect are markedly different from that of the Chungli and the Mongsen dialects. Hence, in the latter part of the work, automatic dialect identification (DID) in the Ao dialects is attempted with Mel Frequency Cepstral Coefficients, Shifted Delta Cepstral coefficients, and F0 features using the Gaussian Mixture models. It is confirmed that in both text-dependent and text-independent DID, the F0 features improve the accuracy of classification.

Sonorant sounds are characterized by regions with prominent formant structure, high energy and high degree of periodicity. In this work, the vocal-tract system, excitation source and suprasegmental features derived from the speech signal are analyzed to measure the sonority information present in each of them. Vocal-tract system information is extracted from the Hilbert envelope of numerator of group delay function. It is derived from zero time windowed speech signal that provides better resolution of the formants. A five-dimensional feature set is computed from the estimated formants to measure the prominence of the spectral peaks. A feature representing strength of excitation is derived from the Hilbert envelope of linear prediction residual, which represents the source information. Correlation of speech over ten consecutive pitch periods is used as the suprasegmental feature representing periodicity information. The combination of evidences from the three different aspects of speech provides better discrimination among different sonorant classes, compared to the baseline MFCC features. The usefulness of the proposed sonority feature is demonstrated in the tasks of phoneme recognition and sonorant classification.

Assessment of intelligibility is required to characterize the overall speech production capability and to measure the speech outcome of different interventions for individuals with cleft lip and palate (CLP). Researchers have found that articulation error and hypernasality have a significant effect on the degradation of CLP speech intelligibility. Motivated by this finding, the present work proposes an objective measure of sentence-level intelligibility by combining the information of articulation deficits and hypernasality. These two speech disorders represent different aspects of CLP speech. Hence, it is expected that the composite measure based on them may utilize complementary clinical information. The objective scores of articulation and hypernasality are used as features to train a regression model, and the output of the model is considered as the predicted intelligibility score. The Spearman's correlation coefficient based analysis shows a significant correlation between the predicted and perceptual intelligibility scores (ρ = 0.77, p

Unlike aspiration in stops, occurrence of aspiration in non-stop consonants is quite rare. Most of the languages that have aspirated non-stop consonants are low-resource languages. Hence, data driven, quantitative, and statistical analysis of their aspiration phenomena is fairly limited. Rabha and Angami are considered in this study, as previous studies have confirmed the existence of aspiration contrast in fricatives and nasals. This study reports the acoustic characteristics of aspiration in stops, fricatives, and nasals. Among them, distinguishing the aspirated fricatives and aspirated nasals from their unaspirated counterparts is a challenging task. A set of acoustic features is proposed to automatically detect the presence of aspiration in fricatives and nasals. Acoustic features, such as vocal tract constriction (VTC), normalized autocorrelation peak strength (NAPS), strength of excitation (SoE), and variance of successive epoch intervals (VSEI) are used to detect aspiration in fricatives and nasals. These features are extracted from zero-frequency filtered signal of the speech sounds, as it preserves the aspiration information. Results show that VTC, NAPS, and SoE can detect aspiration in nasals, whereas SoE and VSEI can detect aspiration in fricatives. The proposed method improves the performance of an automatic phoneme recognizer by reducing the confusion between aspirated and unaspirated counterparts.

The segmentation of unconstrained handwriting is an important issue for both recognition and synthesis systems. In this direction, hidden Markov model (HMM) has been the most popular method for segmentation of continuous handwriting. It has been employed in both implicit and explicit segmentation-based recognition systems. The main advantage of HMM-based force alignment method is that it can segment the handwriting to basic units (e.g. characters) without any knowledge of ground truth boundaries. However, the alignment acquired from an HMM framework often fails to segment the basic units to their actual boundaries. This paper analyzes the segmentation errors made by the HMM in details and identifies an early character boundary problem. Accordingly, an automatic boundary correction method is proposed which utilizes both the forward and the reverse direction alignments of handwritten data with its transcription. This improves the alignment of data and thus the character boundaries in cursive handwriting. The proposed method enhances the word recognition performance for both the Gaussian mixture model (GMM)-HMM and deep neural network (DNN)-HMM frameworks compared to the conventional approach, when evaluated on UNIPEN ICROW-03 and IAM-onDB databases.

This work makes an investigation into speaker verification (SV) from the view of practical systems. Limited data SV is preferred in order to have user comfort and effective decision delivery for regular usage. However, reduction in speech data affects the SV performance that becomes a concern for field deployment. In this work, varied data conditions for SV are explored, and sufficient train with limited test data is presented as a preferable anatomy for practical systems. Different explorations are made from the perspective of improving performance in varied data conditions. These explorations include vocal tract constriction feature to include speaker-specific acoustic–phonetic information, different attributes of voice source features that carry alternative/complementary information from that carried by conventional mel-frequency cepstral coefficient features. Further, kernel discriminant analysis is performed at the back end of i-vector-based speaker modeling for channel/session compensation that is found to work well for varied data conditions. Finally, a framework is proposed in combination with the stated explorations to have a better speaker characterization, which is more effective in case of sufficient train and limited test data scenario. The proposed framework achieves significant improvement in performance [equal error rate (EER): 11.20%, detection cost function (DCF): 0.1990], compared to the baseline (EER: 22.31%, DCF: 0.4128) for sufficient train with 2-s test segment case, showing scope toward application-oriented systems.

Biometric recognition is a trending technology that uses unique characteristics data to identify or verify/authenticate security applications. Amidst the classically used biometrics, voice and face attributes are the most propitious for prevalent applications in day-to-day life because they are easy to obtain through restrained and user-friendly procedures. The pervasiveness of low-cost audio and face capture sensors in smartphones, laptops, and tablets has made the advantage of voice and face biometrics more exceptional when compared to other biometrics. For many years, acoustic information alone has been a great success in automatic speaker verification applications. Meantime, the last decade or two has also witnessed a remarkable ascent in face recognition technologies. Nonetheless, in adverse unconstrained environments, neither of these techniques achieves optimal performance. Since audio-visual information carries correlated and complementary information, integrating them into one recognition system can increase the system’s performance. The vulnerability of biometrics towards presentation attacks and audio-visual data usage for the detection of such attacks is also a hot topic of research. This paper made a comprehensive survey on existing state-of-the-art audio-visual recognition techniques, publicly available databases for benchmarking, and Presentation Attack Detection (PAD) algorithms. Further, a detailed discussion on challenges and open problems is presented in this field of biometrics. This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/ 3

Cleft lip and palate (CLP) refer to a congenital craniofacial condition that causes various speech-related disorders. As a result of structural and functional deformities, the affected subjects' speech intelligibility is significantly degraded, limiting the accessibility and usability of speech-controlled devices. Towards addressing this problem, it is desirable to improve the CLP speech intelligibility. Moreover, it would be useful during speech therapy. In this study, the cycle-consistent adversarial network (CycleGAN) method is exploited for improving CLP speech intelligibility. The model is trained on native Kannada-speaking childrens' speech data. The effectiveness of the proposed approach is also measured using automatic speech recognition performance. Further, subjective evaluation is performed, and those results also confirm the intelligibility improvement in the enhanced speech over the original., Comment: 8 pages, 4 figures, IEEE spoken language and technology workshop

Intelligibility is considered as one of the primary measures for speech rehabilitation of individuals with a cleft lip and palate (CLP). Currently, speech processing and machine-learning-based objective methods are gaining more research interest as a way to quantify speech intelligibility. In this work, joint spectro-temporal features computed from a time-frequency representation of speech are explored to derive speech representations based on Gaussian posteriograms. A comparative framework using dynamic time warping (DTW) is used to quantify the intelligibility of child CLP speech. The DTW distance is used to score sentence-level intelligibility and tested for correlation with perceptual intelligibility ratings obtained from expert speech-language pathologists. A baseline DTW system using the conventional Mel-frequency cepstral coefficients (MFCCs) is also developed to compare the performance of the proposed system. Spearman's rank correlation coefficient between the objective intelligibility scores and the perceptual intelligibility rating is studied. A Williams significance test is conducted to assess the statistical significance of the correlation difference between the methods. The results show that the system based on joint spectro-temporal features significantly outperforms the MFCC-based system.

This work focuses on long-enrollment with short-test speaker verification (SV) from the perspective of application-oriented systems. The importance of phonetic match between train and test models is explored in terms of having a text-constraint model-based framework on Part IV of RedDots database. This database has a text-dependent and a text-prompted-based enrollment conditions for speaker modeling. Two different text-constraint setups are formalized for evaluating the effect of text match on train and test sessions. Further, the excitation source features mel power difference of spectrum in subbands, residual mel frequency cepstral coefficient and discrete cosine transform of integrated linear prediction residual are investigated to determine their significance for text-constraint-based framework. Although the source features individually perform poorer compared to the conventional mel frequency cepstral coefficient (MFCC) features, their significance is reflected in fusion due to the complementary nature of information carried by them. Additionally, the source features become imperative for text-constraint-based models for long-enrollment with short-test SV in fusion to MFCC features and achieves commendable improvement from baseline framework of text-prompted-based enrollment condition. This thus minimizes the performance difference between text-dependent and text-prompted-based enrollment condition showing importance of text-constraint models and source information in long-enrollment with short-test-based framework favorable from the perspective of field deployable systems.

This work explores the significance of source information for speech enhancement resulting in better phoneme recognition of speech with background music segments. Standard procedure for speech enhancement in noisy conditions involves sequential processing in terms of the temporal, spectral and perceptual methods. This work follows the same sequential processing but with the additional modification of studying the effect of source, particularly in the temporal and perceptual-based enhancement techniques for enhancing speech with background music segments. The source information is studied in terms of the epoch locations and epoch strength, obtained after passing the sum of the mean and standard deviation of the component envelopes computed across frequencies obtained using the single frequency filter (SFF), through a zero frequency filter (ZFF). This method of obtaining epoch locations and epoch strength will be termed as SFF-ZFF in this work. The enhanced segments are passed through a phoneme recognizer built using Gaussian mixture model-hidden Markov model (GMM-HMM), subspace Gaussian mixture model-hidden Markov model (SGMM-HMM) and deep neural network-hidden Markov model (DNN-HMM) system, where the models are trained on clean speech. The enhanced audio files show a better phone error rate than the degraded audio files, which means that performing enhancement in terms of the source information is significant for the speech with background music regions.

In this paper, acoustic analysis of misarticulated trills in cleft lip and palate speakers is carried out using excitation source based features: strength of excitation and fundamental frequency, derived from zero-frequency filtered signal, and vocal tract system features: first formant frequency (F1) and trill frequency, derived from the linear prediction analysis and autocorrelation approach, respectively. These features are found to be statistically significant while discriminating normal from misarticulated trills. Using acoustic features, dynamic time warping based trill misarticulation detection system is demonstrated. The performance of the proposed system in terms of the F1-score is 73.44%, whereas that for conventional Mel-frequency cepstral coefficients is 66.11%.

This paper proposes a method using speech-specific knowledge to detect the begin and end points of speech under degraded condition. The method is based on vowel-like region (VLR) detection and uses both excitation source and vocal tract system information. Existing method for VLR detection uses excitation source information. Vocal tract system information from dominant resonant frequency is used to eliminate spurious VLRs in background noise. Foreground speech segmentation using excitation and vocal tract system information is carried out to remove spurious VLRs in the background speech region. Better localization of the end points is done using more detailed information about excitation source in terms of glottal activity to detect the sonorant consonants and missed VLRs. To include an unvoiced consonant, obstruent region detection is done at the beginning of the first VLR and at the end of last VLR. Detected begin and end points are evaluated by comparing with manually marked end points as well as by conducting the text-dependent speaker verification experiments. The proposed method performs better than some of the existing techniques.

An online handwriting recognition (HR) system is usually developed considering point-based features that describe different geometric attributes of handwriting. Often, due to the wide variations in writing styles, the use of point-based features results in high intra-class variability in feature space. To address this problem, we propose a set of features based on character class-conditional probabilities (posterior features) derived from Gaussian Mixture Model (GMM) for online HR task. The proposed features capture class dependent characteristics with a probabilistic framework, which in turn aid in minimizing the intra-class variability of the feature space. Also, the features well represent the inter-class variability of the feature space for a given classification task. The efficacy of the proposed GMM posterior features is shown for character and word recognition tasks, employing support vector machine (SVM) classifier. The experiments are conducted on three databases: the locally collected Assamese digit database, the UNIPEN English character database, and the UNIPEN ICROW-03 English word database. Recognition results are promising over the reported works employing point-based features.

A method to improve voicing decision using glottal activity features proposed for statistical parametric speech synthesis. In existing methods, voicing decision relies mostly on fundamental frequency F 0 , which may result in errors when the prediction is inaccurate. Even though F 0 is a glottal activity feature, other features that characterize this activity may help in improving the voicing decision. The glottal activity features used in this work are the strength of excitation (SoE), normalized autocorrelation peak strength (NAPS), and higher-order statistics (HOS). These features obtained from approximated source signals like zero-frequency filtered signal and integrated linear prediction residual. To improve voicing decision and to avoid heuristic threshold for classification, glottal activity features are trained using different statistical learning methods such as the k-nearest neighbor, support vector machine (SVM), and deep belief network. The voicing decision works best with SVM classifier, and its effectiveness is tested using the statistical parametric speech synthesis. The glottal activity features SoE, NAPS, and HOS modeled along with F 0 and Mel-cepstral coefficients in Hidden Markov model and deep neural network to get the voicing decision. The objective and subjective evaluations demonstrate that the proposed method improves the naturalness of synthetic speech.

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.

This work explores the use of speech enhancement for enhancing degraded speech which may be useful for text dependent speaker verification system. The degradation may be due to noise or background speech. The text dependent speaker verification is based on the dynamic time warping (DTW) method. Hence there is a necessity of the end point detection. The end point detection can be performed easily if the speech is clean. However the presence of degradation tends to give errors in the estimation of the end points and this error propagates into the overall accuracy of the speaker verification system. Temporal and spectral enhancement is performed on the degraded speech so that ideally the nature of the enhanced speech will be similar to the clean speech. Results show that the temporal and spectral processing methods do contribute to the task by eliminating the degradation and improved accuracy is obtained for the text dependent speaker verification system using DTW.

Cleft lip and palate is a craniofacial condition that affects the intelligibility of speech. The current work describes a modification of three types of misarticulated fricative /s/ in cleft lip and palate speech, which include palatalized /s/, phoneme-specific nasal air emission distorted /s/, and glottal stop substituted /s/. By using the knowledge of the glottal activity, frication, and silence, an approach is proposed for misarticulated fricative detection and categorization of the type of error. Based on the category of error, an appropriate modification technique is applied. The deviations of palatalized /s/ and phoneme-specific nasal air emission distorted /s/ are corrected by modifying the spectral energy. The high-frequency energy levels are emphasized to improve the perception of fricative /s/ using spectral tilt modification. Glottal stop substitution is modified by inserting artificially synthesized /s/, where the fricative /s/ signal is synthesized using white noise signal and linear prediction filter obtained from normal children fricative /s/. Further, the modified speech samples are subjected to objective and subjective evaluation. The evaluation scores obtained from objective and subjective assessment indicate speech intelligibility improvement of the modified signals compared to the misarticulated fricative /s/.

The objective of this work is to obtain meaningful time domain components, or Intrinsic Mode Functions (IMFs), of the speech signal, using Empirical Mode Decomposition (EMD), with reduced mode mixing, and in a time-efficient manner. This work focuses on two aspects – firstly, extracting IMFs of the speech signal which can better reflect its higher frequency spectrum; and secondly, to get a better representation and distribution of the vocal tract resonances of the speech signal in its IMFs, compared to that obtained from standard EMD. To this effect, modifications are proposed to the EMD algorithm for processing speech signals, based on the critical nature of the interpolation points (IPs) used for cubic spline interpolation in EMD. The effect of using different sets of IPs, other than the extrema of the residue – as used in standard EMD – is analyzed. It is found that having more IPs is beneficial only upto a certain limit, after which the characteristic dyadic filterbank nature of EMD breaks down. For certain sets of IPs, these modified EMD processes perform better than EMD, giving better frequency separability between the IMFs, and an enhanced representation of the higher frequency content of the signal. A detailed study of the distribution of the formants, in the IMFs of the speech signal, is done using Linear Prediction (LP) analysis of the IMFs. It is found that the IMFs of the EMD variants have a far better distribution of the formants structure within them, with reduced overlapping amongst their filter spectrums, compared to that of standard EMD. Henceforth, when subjected to the task of formants estimation of voiced speech, using LP analysis, the IMFs of the modified EMD processes cumulatively exhibit a superior performance than that of standard EMD, or the speech signal itself, under both clean and noisy conditions.

This paper proposes the use of speech-specific features for speech / music classification. Features representing the excitation source, vocal tract system and syllabic rate of speech are explored. The normalized autocorrelation peak strength of zero frequency filtered signal, and peak-to-sidelobe ratio of the Hilbert envelope of linear prediction residual are the two source features. The log mel energy feature represents the vocal tract information. The modulation spectrum represents the slowly-varying temporal envelope corresponding to the speech syllabic rate. The novelty of the present work is in analyzing the behavior of these features for the discrimination of speech and music regions. These features are non-linearly mapped and combined to perform the classification task using a threshold based approach. Further, the performance of speech-specific features is evaluated using classifiers such as Gaussian mixture models, and support vector machines. It is observed that the performance of the speech-specific features is better compared to existing features. Additional improvement for speech / music classification is achieved when speech-specific features are combined with the existing ones, indicating different aspects of information exploited by the former.

In this paper, the explicit and implicit modelling of the subsegmental excitation information are experimentally compared. For explicit modelling, the static and dynamic values of the standard Liljencrants–Fant (LF) parameters that model the glottal flow derivative (GFD) are used. A simplified approximation method is proposed to compute these LF parameters by locating the glottal closing and opening instants. The proposed approach significantly reduces the computation needed to implement the LF model. For implicit modelling, linear prediction (LP) residual samples considered in blocks of 5 ms with shift of 2.5 ms are used. Different speaker recognition studies are performed using NIST-99 and NIST-03 databases. In case of speaker identification, the implicit modelling provides significantly better performance compared to explicit modelling. Alternatively, the explicit modelling seem to be providing better performance in case of speaker verification. This indicates that explicit modelling seem to have relatively less intra and inter-speaker variability. The implicit modelling on the other hand, has more intra and inter-speaker variability. What is desirable is less intra and more inter-speaker variability. Therefore, for speaker verification task explicit modelling may be used and for speaker identification task implicit modelling may be used. Further, for both speaker identification and verification tasks the explicit modelling provides relatively more complimentary information to the state-of-the-art vocal tract features. The contribution of the explicit features is relatively more robust against noise. We suggest that the explicit approach can be used to model the subsegmental excitation information for speaker recognition.

This work demonstrates the significance of information about vowel onset points (VOPs) for speaker verification. VOP is defined as the instant at which the onset of vowel takes place. Vowel-like regions can be identified using VOPs. By production, vowel-like regions have impulse-like excitation and therefore impulse-response of vocal tract system is better manifested in them, and are relatively high signal to noise ratio (SNR) regions. Speaker information extracted from such regions may therefore be more discriminative. Due to this better speaker modeling and reliable testing may be possible using the features extracted from vowel-like regions. It is demonstrated in this work that for clean and matched conditions, relatively less number of frames from vowel-like regions are sufficient for speaker modeling and testing. Alternatively, for degraded and mismatched conditions, vowel-like regions provide better performance.

Speaker specific information present in the excitation signal is mostly viewed from sub-segmental, segmental and supra-segmental levels. In this work, the supra-segmental level information is explored for recognizing speakers. Earlier study has shown that, combined use of pitch and epoch strength vectors provides useful supra-segmental information. However, the speaker recognition accuracy achieved by supra-segmental level feature is relatively poor than other levels source information. May be the modulation information present at the supra-segmental level of the excitation signal is not manifested properly in pith and epoch strength vectors. We propose a method to model the supra-segmental level modulation information from residual mel frequency cepstral coefficient (R-MFCC) trajectories. The evidences from R-MFCC trajectories combined with pitch and epoch strength vectors are proposed to represent supra-segmental information. Experimental results show that compared to pitch and epoch strength vectors, the proposed approach provides relatively improved performance. Further, the proposed supra-segmental level information is relatively more complimentary to other levels information.

This work explores mel power difference of spectrum in subband, residual mel frequency cepstral coefficient, and discrete cosine transform of the integrated linear prediction residual for speaker verification under limited test data conditions. These three source features are found to capture different attributes of source information, namely, periodicity, smoothed spectrum information, and shape of the glottal signal, respectively. On the NIST SRE 2003 database, the proposed combination of the three source features performs better [equal error rate (EER): 20.19%, decision cost function (DCF): 0.3759] than the mel frequency cepstral coefficient feature (EER: 22.31%, DCF: 0.4128) for 2 s duration of test segments.

In this work we develop a speaker recognition system based on the excitation source information and demonstrate its significance by comparing with the vocal tract information based system. The speaker-specific excitation information is extracted by the subsegmental, segmental and suprasegmental processing of the LP residual. The speaker-specific information from each level is modeled independently using Gaussian mixture modeling--universal background model (GMM-UBM) modeling and then combined at the score level. The significance of the proposed speaker recognition system is demonstrated by conducting speaker verification experiments on the NIST-03 database. Two different tests, namely, Clean test and Noisy test are conducted. In case of Clean test, the test speech signal is used as it is for verification. In case of Noisy test, the test speech is corrupted by factory noise (9 dB) and then used for verification. Even though for Clean test case, the proposed source based speaker recognition system still provides relatively poor performance than the vocal tract information, its performance is better for Noisy test case. Finally, for both clean and noisy cases, by providing different and robust speaker-specific evidences, the proposed system helps the vocal tract system to further improve the overall performance.

In this paper, the task of identifying the speaker using limited training and testing data is addressed. Speaker identification system is viewed as four stages namely, analysis, feature extraction, modelling and testing. The speaker identification performance depends on the techniques employed in these stages. As demonstrated by different experiments, in case of limited training and testing data condition, owing to less data, existing techniques in each stage will not provide good performance. This work demonstrates the following: multiple frame size and rate (MFSR) analysis provides improvement in the analysis stage, combination of mel frequency cepstral coefficients (MFCC), its temporal derivatives (Δ, ΔΔ), linear prediction residual (LPR) and linear prediction residual phase (LPRP) features provides improvement in the feature extraction stage and combination of learning vector quantization (LVQ) and gaussian mixture model — universal background model (GMM-UBM) provides improvement in the modelling stage. The performance is further improved by integrating the proposed techniques at the respective stages and combining the evidences from them at the testing stage. To achieve this, we propose strength voting (SV), weighted borda count (WBC) and supporting systems (SS) as combining methods at the abstract, rank and measurement levels, respectively. Finally, the proposed hierarchical combination (HC) method integrating these three methods provides significant improvement in the performance. Based on these explorations, thiswork proposes a scheme for speaker identification under limited training and testing data.

This paper presents an experimental evaluation of the combined temporal and spectral processing methods for speaker recognition task under noise, reverberation or multi-speaker environments. Automatic speaker recognition system gives good performance in controlled environments. Speech recorded in real environments by distant microphones is degraded by factors like background noise, reverberation and interfering speakers. This degradation strongly affects the performance of the speaker recognition system. Combined temporal and spectral processing (TSP) methods proposed in our earlier study are used for pre-processing to improve the speaker-specific features and hence the speaker recognition performance. Different types of degradation like background noise, reverberation and interfering speaker are considered for evaluation. The evaluation is carried out for the individual temporal processing, spectral processing and the combined TSP method. The experimental results show that the combined TSP methods give relatively higher recognition performance compared to either temporal or spectral processing alone.

Most of the existing modelling techniques for the speaker recognition task make an implicit assumption of sufficient data for speaker modelling and hence may lead to poor modelling under limited data condition. The present work gives an experimental evaluation of the modelling techniques like Crisp Vector Quantization (CVQ), Fuzzy Vector Quantization (FVQ), Self-Organizing Map (SOM), Learning Vector Quantization (LVQ), and Gaussian Mixture Model (GMM) classifiers. An experimental evaluation of the most widely used Gaussian Mixture Model-Universal Background Model (GMM-UBM) is also made. The experimental knowledge is then used to select a subset of classifiers for obtaining the combined classifiers. It is proposed that the combined LVQ and GMM-UBM classifier provides relatively better performance compared to all the individual as well as combined classifiers.

A characterization of the voice source (VS) signal by the pitch synchronous (PS) discrete cosine transform (DCT) is proposed. With the integrated linear prediction residual (ILPR) as the VS estimate, the PS DCT of the ILPR is evaluated as a feature vector for speaker identification (SID). On TIMIT and YOHO databases, using a Gaussian mixture model (GMM)-based classifier, it performs on par with existing VS-based features. On the NIST 2003 database, fusion with a GMM-based classifier using MFCC features improves the identification accuracy by 12% in absolute terms, proving that the proposed characterization has good promise as a feature for SID studies. (C) 2015 Acoustical Society of America

In this paper, through different experimental studies we demonstrate that the excitation component of speech can be exploited for speaker recognition studies. Linear prediction (LP) residual is used as a representation of excitation information in speech. The speaker-specific information in the excitation of voiced speech is captured using the AutoAssociative Neural Network (AANN) models. The decrease in the error during training and recognizing correct speakers during testing demonstrates that the excitation component of speech contains speaker-specific information and is indeed being captured by the AANN models. The study on the effect of different LP orders demonstrates that for a speech signal sampled at 8 kHz, the LP residual extracted using LP order in the range 8-20 best represents the speaker-specific excitation information. It is also demonstrated that the proposed speaker recognition system using excitation information and AANN models requires significantly less amount of data both during training as well as testing, compared to the speaker recognition system using vocal tract information. Finally the speaker recognition studies on NIST 2002 database demonstrates that even though, the recognition performance from the excitation information alone is poor, when combined with evidence from vocal tract information, there is significant improvement in the performance. This result demonstrates the complementary nature of the excitation component of speech. � 2006 Elsevier B.V. All rights reserved.

We analyze the effectivity of different pseudo noise (PN) and orthogonal sequences for encrypting speech signals in terms of perceptual intelligence. Speech signal can be viewed as sequence of correlated samples and each sample as sequence of bits. The residual intelligibility of the speech signal can be reduced by removing the correlation among the speech samples. PN sequences have random like properties that help in reducing the correlation among speech samples. The mean square aperiodic auto-correlation (MSAAC) and the mean square aperiodic cross-correlation (MSACC) measures are used to test the randomness of the PN sequences. Results of the investigation show the effectivity of large Kasami sequences for this purpose among many PN sequences., {"references":["H. J. Beker and F. C. Piper, Secure Speech Communications, London:\nAcademic Press, 1985.","W. Stallings, Cryptography and Network Security, Englewoods Cliffs,\nNJ: Prentice Hall, 2003.","W. Diffe and M. E. Hellman, \"New directions in cryptography,\" IEEE\nTrans. Inform. Theory, vol. 22, pp. 644-654, Nov. 1976.","N. S. Jayant, B. J. McDermott, S. W. Christensen and A. M. Quinn, \"A\ncomparison of four methods for analog speech privacy,\" IEEE Trans.\nCommun., vol. COM-29, pp. 18-23, Jan. 1981.","B. Goldburg, S. Sridharan and E. Dawson, \"Design and cryptanalysis of\ntransform based speech scramblers,\" IEEE J. Selected Areas Commun.,\nvol. 11, no. 5, pp. 735-744, June 1993.","R. L. Pickholtz, D. L. Schilling and L. B. Milstein, \"Theory of spread\nspectrum communications ÔÇö A tutorial,\" IEEE Trans. Commun., vol.\nCOM-30, no. 5, May 1982.","E. H. Dinan and B. Jabbari, \"Spreading codes for direct sequence CDMA\nand wideband CDMA cellular networks,\" IEEE Commun. Magazine, vol.\n36, no. 4, pp. 48-54, Sep. 1998.","B. Sklar, Digital Communications: Fundamentals and Applications, 2nd\nEd., NJ: Prentice Hall, 2001.","J. H. Lindholm, \"An analysis of the pseudo randomness properties of\nthe subsequences of long m-sequences,\" IEEE Trans. Inform. Theory,\nvol. IT-14, pp. 569-576, July 1968.\n[10] I. Oppermann and B. S. Vucetic, \"Complex spreading sequences with a\nwide range of correlation properties,\" IEEE Trans. Commun., vol. COM-\n45, pp. 365-375, March 1997.\n[11] L. T. Wang and E. J. McCluskey, \"Linear feedback shift register design\nusing cyclic codes,\" IEEE Trans. Comput., vol. 37, pp. 1302-1306, Oct.\n1988.\n[12] A. Fuster and L. J. Garcia, \"An efficient algorithm to generate binary\nsequences for cryptographic purposes,\" Theoretical Computer Science,\nvol. 259, pp. 679-688, May 2001.\n[13] D. V. Sarwate and M. B. Pursley, \"Correlation properties of pseudo\nrandom and related sequences,\" Proc. IEEE, vol. 68, no. 5, pp. 593-\n619, May 1980.\n[14] S. W. Golomb and R. A. Scholtz, \"Generalized Barker sequences,\" IEEE\nTrans. Inform. Theory, vol. IT-11, no. 4, pp. 533-537, Oct. 1965.\n[15] D. G. Luenberger, \"On Barker codes of even length,\" Proc. IEEE, vol.\n51, pp. 230-231, Jan. 1963.\n[16] C. K. Chan and W. H. Lam, \"Generalised Barker-like PN sequences\nfor quasisynchronous spread spectrum multiple access communication\nsystems,\" IEE Proc. Commun., vol. 142, no. 2, pp. 91-98, April 1995.\n[17] X. Wang, Y. Wu and B. Caron, \"Transmitter identification using embedded\npseudo random sequences,\" IEEE Trans. Broadcasting, vol. 50, no.\n3, pp. 244-252, Sep. 2004.\n[18] V. Milosevic, V. Delic and V. Senk, \"Hadamard transform application\nin speech scrambling,\" Proc. IEEE, vol. 1, pp. 361-364, July 1997.\n[19] Tai-Kuo Woo, \"Orthogonal variable spreading codes for wideband\nCDMA,\" IEEE Trans. Vehicular Tech., vol. 51, no. 4, pp. 700-709, July\n2002.\n[20] B. Wysocki and T. A. Wysocki, \"Modified Walsh Hadamard\nsequences for DS-CDMA wireless systems,\" School of\nElectrical, Computer and Telecommunications Engineering,\nUniversity of Wollongong, Australia. (Online) Available:\nwww.elec.uow.edu.au/staff/wysocki/publications/J1.pdf.\n[21] A. Mitra, \"On Pseudo-Random and Orthogonal Binary Spreading Sequences\",\nto appear in Int. J. Info. Tech., 2007."]}

This paper proposes a method for the enhancement of reverberant speech using the knowledge of the excitation source of speech production. The degradation level in the reverberant speech is measured in terms of Speechto-Reverberation component Ratio (SRR). From perception and processing point of view high SRR regions are important. Hence the proposed method identifies and enhances the speech in high SRR regions. The high SRR regions are identified using the Hilbert envelope of the Linear Prediction (LP) residual, which contains information about the excitation source of speech production. The Hilbert envelope of the LP residual derived from the reverberant speech is processed by the covariance analysis to derive the weight function. The LP residual of the reverberant speech is multiplied with the weight function to enhance the excitations of speech in the high SRR regions. The speech signal synthesized from the modified LP residual is found to be less reverberant.

This work demonstrates the usefulness of multiple frame size and rate (MFSR) analysis for speaker recognition under limited data condition. Present day speaker recognition systems assume the availability of sufficient data for modelling and testing. Owing to this, speech signals are analysed with fixed frame size and rate, which may be termed as single frame size and rate (SFSR) analysis. In the limited data condition available training and testing data is small. If we use SFSR analysis, then it may not provide sufficient feature vectors to train and test the speaker. Further, insufficient feature vectors lead to poor speaker modelling during training and may not yield reliable decision during testing. As part of analysis, we demonstrate the use of multiple frame size (MFS), multiple frame rate (MFR) and MFSR analysis techniques for speaker recognition under limited data condition. These techniques are specifically useful to mitigate the sparseness of limited feature vectors during training and testing. These techniques produce relatively more number of feature vectors. This helps in better modelling and testing under limited data conditions. The experimental results show that use of MFS, MFR and MFSR analysis improves the performance significantly compared to SFSR analysis. The MFSR analysis even outperforms the Gaussian mixture model-universal background model (GMM-UBM) performance, the most widely used modelling technique.