Iterative Adaptive Inverse Filtering Research Articles

Evaluation of Glottal Inverse Filtering Techniques on OPENGLOT Synthetic Male and Female Vowels

Current articulatory-based three-dimensional source–filter models, which allow the production of vowels and diphtongs, still present very limited expressiveness. Glottal inverse filtering (GIF) techniques can become instrumental to identify specific characteristics of both the glottal source signal and the vocal tract transfer function to resemble expressive speech. Several GIF methods have been proposed in the literature; however, their comparison becomes difficult due to the lack of common and exhaustive experimental settings. In this work, first, a two-phase analysis methodology for the comparison of GIF techniques based on a reference dataset is introduced. Next, state-of-the-art GIF techniques based on iterative adaptive inverse filtering (IAIF) and quasi closed phase (QCP) approaches are thoroughly evaluated on OPENGLOT, an open database specifically designed to evaluate GIF, computing well-established GIF error measures after extending male vowels with their female counterparts. The results show that GIF methods obtain better results on male vowels. The QCP-based techniques significantly outperform IAIF-based methods for almost all error metrics and scenarios and are, at the same time, more stable across sex, phonation type, F0, and vowels. The IAIF variants improve the original technique for most error metrics on male vowels, while QCP with spectral tilt compensation achieves a lower spectral tilt error for male vowels than the original QCP.

Whispered Speech Detection Using Glottal Flow-Based Features

Recent studies have reported that the performance of Automatic Speech Recognition (ASR) technologies designed for normal speech notably deteriorates when it is evaluated by whispered speech. Therefore, the detection of whispered speech is useful in order to attenuate the mismatch between training and testing situations. This paper proposes two new Glottal Flow (GF)-based features, namely, GF-based Mel-Frequency Cepstral Coefficient (GF-MFCC) as a magnitude-based feature and GF-based relative phase (GF-RP) as a phase-based feature for whispered speech detection. The main contribution of the proposed features is to extract magnitude and phase information obtained by the GF signal. In the GF-MFCC, Mel-frequency cepstral coefficient (MFCC) feature extraction is modified using the estimated GF signal derived from the iterative adaptive inverse filtering as the input to replace the raw speech signal. In a similar way, the GF-RP feature is the modification of the relative phase (RP) feature extraction by using the GF signal instead of the raw speech signal. The whispered speech production provides lower amplitude from the glottal source than normal speech production, thus, the whispered speech via Discrete Fourier Transformation (DFT) provides the lower magnitude and phase information, which make it different from a normal speech. Therefore, it is hypothesized that two types of our proposed features are useful for whispered speech detection. In addition, using the individual GF-MFCC/GF-RP feature, the feature-level and score-level combination are also proposed to further improve the detection performance. The performance of the proposed features and combinations in this study is investigated using the CHAIN corpus. The proposed GF-MFCC outperforms MFCC, while GF-RP has a higher performance than the RP. Further improved results are obtained via the feature-level combination of MFCC and GF-MFCC (MFCC&GF-MFCC)/RP and GF-RP(RP&GF-RP) compared with using either one alone. In addition, the combined score of MFCC&GF-MFCC and RP&GF-RP gives the best frame-level accuracy of 95.01% and the utterance-level accuracy of 100%.

Determination of the vocal tract model order in iterative adaptive inverse filtering

This paper is part of a special issue on Music Technology. Accurate estimation of glottal wave is an important method for singing voice analysis via voicing patterns. In this paper, an optimising method is proposed for automatic determination of vocal tract linear prediction analysis order that follows the specific situation of different voicing scenes based on Iterative Adaptive Inverse Filtering (IAIF) aiming at getting a more accurate glottal wave from speech or singing voice signal in a non-invasive way. In comparison with existing methods that use fixed experience order, the proposed method can achieve up to 8.41% improvement in correlation coefficient with the real glottal wave.

New replay attack detection using iterative adaptive inverse filtering and high frequency band

Replay attack is most vulnerable to automatic speaker verification system, where the frauds get the access by replaying pre-recorded voice samples of the genuine speakers. In this proposed work, we mainly concentrated on glottal excitation and high frequency band. First, we demonstrate the importance of the glottal information of the speech signals to detect the replay attack for speaker verification system along with magnitude based discrimination power features set. Iterative Adaptive Inverse Filtering (IA-IF) technique is used to extract the glottal excitation information of the given speech spectrum which shows the difference in the characteristics of genuine and replayed speech sample. Using this technique the glottal information is gained by eliminating the lip radiation and vocal tract effect by applying the integration and inverse filtering respectively. Secondly, we have shown the prominence and discriminative information to detect the replayed attack which is present in high frequency band of the speech spectrums of genuine and replayed samples. Finally, Constant-Q Cepstral Coefficients (CQCC) is used to extract desired features set from the high-frequency bands of glottal excitation spectrum. From the experimental studies done on ASV spoof-2017 version 2.0 database it shows that the proposed method feature set significantly decreases the Equal Error Rate (EER) to 3.68% and 8.32% for development and evaluation set when compare to other state-of-art method feature set.

The Detection of Parkinson's Disease From Speech Using Voice Source Information

Developing automatic methods to detect Parkinson's disease (PD) from speech has attracted increasing interest as these techniques can potentially be used in telemonitoring health applications. This article studies the utilization of voice source information in the detection of PD using two classifier architectures: traditional pipeline approach and end-to-end approach. The former consists of feature extraction and classifier stages. In feature extraction, the baseline acoustic features-consisting of articulation, phonation, and prosody features-were computed and voice source information was extracted using glottal features that were estimated by iterative adaptive inverse filtering (IAIF) and quasi-closed phase (QCP) glottal inverse filtering methods. Support vector machine classifiers were developed utilizing the baseline and glottal features extracted from every speech utterance and the corresponding healthy/PD labels. The end-to-end approach uses deep learning models which were trained using both raw speech waveforms and raw voice source waveforms. In the latter, two glottal inverse filtering methods (IAIF and QCP) and zero frequency filtering method were utilized. The deep learning architecture consists of a combination of convolutional layers followed by a multilayer perceptron. Experiments were performed using PC-GITA speech database. From the traditional pipeline systems, the highest classification accuracy (67.93%) was given by combination of baseline and QCP-based glottal features. From the end-to-end-systems, the highest accuracy (68.56%) was given by the system trained using QCP-based glottal flow signals. Even though classification accuracies were modest for all systems, the study is encouraging as the extraction of voice source information was found to be most effective in both approaches.

Estimation of the glottal flow from speech pressure signals: Evaluation of three variants of iterative adaptive inverse filtering using computational physical modelling of voice production

The aim of this study is to comparatively review and evaluate three variants of the glottal inverse filtering algorithm based on iterative adaptive inverse filtering (IAIF): the Standard algorithm, and two recently proposed variants that use iterative optimal preemphasis (IOP) and a glottal flow model (GFM), respectively. To enable an objective evaluation, a computational physical model of voice production is used to generate time-domain signals pertaining to both the input glottal flow and the output speech pressure, for a wide range of vowels, fundamental frequencies, and voice qualities (involving co-variation of phonation type and loudness). Furthermore, for a fair comparison, the three key parameters of IAIF are selected by an exhaustive search to minimize the root-mean-square error between the estimated and reference glottal flow derivative in each analyzed frame and performance is assessed with two time-domain and two frequency-domain error measures. A conventional evaluation is also carried out with fixed parameter values determined by cross-validation. Results indicate that IOP tends to yield the lowest errors for nonback vowels (reducing errors by 31% on average compared with Standard), especially for not too high fundamental frequencies and not too pressed voice qualities; GFM becomes competitive for normal phonations when fixed parameter values are used; and in other cases, Standard IAIF is still recommended. In addition, the results suggest that not only the overall spectral tilt (as controlled by IOP and GFM) but also the balance between the levels of different spectral regions, can be important for accurate estimation of the glottal flow.

Noninvasive Detection of Potentially Precancerous Lesions of Vocal Fold Based on Glottal Wave Signal and SVM Approaches

Purpose: Preventive detection of premalignant lesions of vocal fold can play an important role for earlier detection of larynx cancer. In this paper, an automatic detection of premalignant lesions based on human voice production theory is designed. More specifically, premalignant lesions such as leukoplakia, Erythroplakia, Keratosis, etc are directly related to the vocal fold, features extracted from glottal fold waveform can be pertinent and critical. The proposed method performs as non-intrusive technique based on the extraction of vocal fold waveform from the recorded utterances. Method: The basic idea is to extract pertinent features from the source signal (glottal waveform). However, such signal is not directly accessible. Hence, we perform a first treatment of recorded speech using Iterative Adaptive Inverse Filtering (IAIF) to extract the glottal waveform. IAIF is applied on two databases: Massachusetts Eye and Ear Inrmary (MEEI) database and Saarbrucken Voice Database (SVD) with sustained vowel samples. From the glottal flow impulses, pertinent instances are used to build pertinent features. A deep exploration of extracted features, using statistical tools such as boxplot and Principal Component Analysis (PCA), leads to select the most critical and pertinent features. By using Support Vector Machine Technique as classification technique, a discrimination between normal and premalignant tumor is achieved. Results: Preliminary results show that some features are not discriminant. Those features are discarded and only pertinent ones are used in learning and testing processing using SVM. The performances are assessed using four criteria, sensitivity, specificity, precision and accuracy. When selected features are combined according to PCA analysis, an accuracy of about 92% is accomplished. Conclusion: It is shown in this study that is possible to detect the premalignant lesions with acceptable and fairly sensitivity, specificity, precision and accuracy. Such algorithm can help detect at earlier stage laryngeal cancer.

Alternating minimisation for glottal inverse filtering

A new method is proposed for solving the glottal inverse filtering (GIF) problem. The goal of GIF is to separate an acoustical speech signal into two parts: the glottal airflow excitation and the vocal tract filter. To recover such information one has to deal with a blind deconvolution problem. This ill-posed inverse problem is solved under a deterministic setting, considering unknowns on both sides of the underlying operator equation. A stable reconstruction is obtained using a double regularization strategy, alternating between fixing either the glottal source signal or the vocal tract filter. This enables not only splitting the nonlinear and nonconvex problem into two linear and convex problems, but also allows the use of the best parameters and constraints to recover each variable at a time. This new technique, called alternating minimization glottal inverse filtering (AM-GIF), is compared with two other approaches: Markov chain Monte Carlo glottal inverse filtering (MCMC-GIF), and iterative adaptive inverse filtering (IAIF), using synthetic speech signals. The recent MCMC-GIF has good reconstruction quality but high computational cost. The state-of-the-art IAIF method is computationally fast but its accuracy deteriorates, particularly for speech signals of high fundamental frequency (F0). The results show the competitive performance of the new method: With high F0, the reconstruction quality is better than that of IAIF and close to MCMC-GIF while reducing the computational complexity by two orders of magnitude.

Automatic glottal inverse filtering with the Markov chain Monte Carlo method

This paper presents a new glottal inverse filtering (GIF) method that utilizes a Markov chain Monte Carlo (MCMC) algorithm. First, initial estimates of the vocal tract and glottal flow are evaluated by an existing GIF method, iterative adaptive inverse filtering (IAIF). Simultaneously, the initially estimated glottal flow is synthesized using the Rosenberg–Klatt (RK) model and filtered with the estimated vocal tract filter to create a synthetic speech frame. In the MCMC estimation process, the first few poles of the initial vocal tract model and the RK excitation parameter are refined in order to minimize the error between the synthetic and original speech signals in the time and frequency domain. MCMC approximates the posterior distribution of the parameters, and the final estimate of the vocal tract is found by averaging the parameter values of the Markov chain. Experiments with synthetic vowels produced by a physical modeling approach show that the MCMC-based GIF method gives more accurate results compared to two known reference methods.

ANALYSIS OF VOICED SPEECH EXCITATION DUE TO ALCOHOL INTOXICATION

A significant part of information carried in speech signal refers to the speaker. This paper deals with investigating alcohol intoxication based on analyzing recorded speech signal. Speech changes resulting from alcohol intoxication were investigated in the waveform of glottal pulses estimated from speech by applying the Iterative Adaptive Inverse Filtering (IAIF). Experimental results show that analysis of glottal excitation appears to be a useful approach to provide evidence of alcohol intoxication of over 196o. At this alcohol level, the associated negative events influence professional performance and may involve fatal accidents in some cases. Via analyzing the speech signal, the speaker could be automatically monitored without their active co-operation. For use in our experiments, a new collection of Czech alcoholized speech consisting of phonetically identical speech data spoken in both sober and intoxicated state was created. http://dx.doi.org/10.5755/j01.itc.40.2.429

Analysis and Modeling of Snore Source Flow With Its Preliminary Application to Synthetic Snore Generation

With the emerging use of snore properties for clinical purposes, there is a need to understand the characteristics of snore source flow (SF)-the acoustic source in snore production. This paper attempts to analyze and model both SF and its derivative (SFD), along with its preliminary application to the generation of synthetic snores. SFs and SFDs were extracted from natural snores via an iterative adaptive inverse filtering approach, and subsequently parameterized into various time- and amplitude-based parameters to quantify the oscillatory maneuvers of snore excitation source (ES). The SF and SFD waveforms were also, respectively, modeled using the first and second derivatives of the Gaussian probability density function. Subjective and objective measures, including paired comparison score and sum-of-squared error, were assessed to appraise the performance of SFD model in producing natural-sounding snores. Results consistently show that: 1) the shapes of SF pulse are different among snores and can be associated with the dynamic biomechanical properties (e.g., compliance and elasticity) of ES; 2) changes to the SF or SFD pulse shape can affect the snore properties, both acoustically and perceptually; and 3) the proposed SFD model can generate close-to-natural sounding snores. Further research in this area can potentially yield valuable benefits to snore-oriented applications.

Perception of Emotional Valences and Activity Levels from Vowel Segments of Continuous Speech

This study aimed to investigate the role of voice source and formant frequencies in the perception of emotional valence and psychophysiological activity level from short vowel samples (approximately 150 milliseconds). Nine professional actors (five males and four females) read a prose passage simulating joy, tenderness, sadness, anger, and a neutral emotional state. The stress carrying vowel [a:] was extracted from continuous speech during the Finnish word [ta:k:ahan] and analyzed for duration, fundamental frequency (F0), equivalent sound level (L(eq)), alpha ratio, and formant frequencies F1-F4. Alpha ratio was calculated by subtracting the L(eq) (dB) in the range 50 Hz-1 kHz from the L(eq) in the range 1-5 kHz. The samples were inverse filtered by Iterative Adaptive Inverse Filtering and the estimates of the glottal flow obtained were parameterized with the normalized amplitude quotient (NAQ = f(AC)/(d(peak)T)). Fifty listeners (mean age 28.5 years) identified the emotional valences from the randomized samples. Multinomial Logistic Regression Analysis was used to study the interrelations of the parameters for perception. It appeared to be possible to identify valences from vowel samples of short duration ( approximately 150 milliseconds). NAQ tended to differentiate between the valences and activity levels perceived in both genders. Voice source may not only reflect variations of F0 and L(eq), but may also have an independent role in expression, reflecting phonation types. To some extent, formant frequencies appeared to be related to valence perception but no clear patterns could be identified. Coding of valence tends to be a complicated multiparameter phenomenon with wide individual variation.

Vocal-fold collision mass as a differentiator between registers in the low-pitch range

Register shift between the chest and falsetto register is generally studied in the higher-than-speaking pitch range. However, a similar difference can also be produced at speaking pitch level. The shift from breathy "falsetto" phonation to normal chest voice phonation was studied in normal female (pitch range 170-180 Hz) and male (pitch range 94-110 Hz) subjects. The phonations gliding from falsetto to normal chest voice were analyzed using iterative adaptive inverse filtering and electroglottography. Both trained and untrained, as well as female and male subjects, were able to produce an abrupt register shift from soft falsetto to soft chest register phonation. The differences between male and female speakers in the glottal flow waveforms were smaller than expected. The register shift is interpreted in terms of a "critical mass" concept of chest register phonation.

Glottal wave analysis with Pitch Synchronous Iterative Adaptive Inverse Filtering

A new glottal wave analysis method, Pitch Synchronous Iterative Adaptive Inverse Filtering (PSIAIF) is presented. The algorithm is based on a previously developed method, Iterative Adaptive Inverse Filtering (IAIF). In the IAIF-method the glottal contribution to the speech spectrum is first estimated with an iterative structure. The vocal tract transfer function is modeled after eliminating the average glottal contribution. The glottal excitation is obtained by cancelling the effects of the vocal tract and lip radiation by inverse filtering. In the new PSIAIF-method the glottal pulseform is computed by applying the IAIF-algorithm twice to the same signal. The first IAIF-analysis gives as a result a glottal excitation that spans over several pitch periods. This pulseform is used in order to determine positions and lengths of frames for the pitch synchronous analysis. The final result is obtained by analysing the original speech signal with the IAIF-algorithm one fundamental period at a time. The PSIAIF-algorithm was applied in glottal wave analysis using both synthetic and natural vowels. The results show that the method is able to give a fairly accurate estimate for the glottal flow excluding the analysis of vowels with a low first formant that are produced with a pressed phonation type.

Preliminary experiences in using automatic inverse filtering of acoustical signals for the voice source analysis

A new straightforward glottal wave analysis method, called Iterative Adaptive Inverse Filtering (IAIF), is presented. The method is based on the separated speech production model that consists of three processes: The glottal excitation, the vocal tract and the lip radiation effect. In order to estimate the vocal tract transfer function the effect of the source to the speech spectrum is first estimated with an iterative procedure. After eliminating the estimated glottal contribution, the vocal tract transfer function is estimated using linear predictive analysis (LPC). The estimate for the glottal excitation is finally obtained by cancelling the effects of the vocal tract and lip radiation by inverse filtering. The performance of the new algorithm is verified and compared to two other LPC-based glottal wave analysis methods using synthetic vowels. It is shown that the IAIF-method is able to give a fairly accurate estimate for the glottal excitation. The results of the analysis of natural speech using the I...