Perception and speech analysisGlottal wave analysis with Pitch Synchronous Iterative Adaptive Inverse Filtering
Abstract
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.
Zusammenfassung
Im vorliegenden Artikel wird ein neues Verfahren zur Analyse des Glottissignals vorgestellt, das Pitch Synchronous Iterative Adaptive Inverse Filtering (PSIAIF). Der Algorithmus basiert auf einer vorher entwickelten Methode (IAIF). Bei der IAIF-Methode wird zunächst der glottale Beitrag zum Sprachspektrum mit einem iterativen Verfahren geschätzt. Die Übertragungsfunktion des Vokaltraktes wird anschlieβend nach Eliminierung des mittleren Glottalbeitrags modelliert. Die Anregung der Glottis wird ermittelt, indem die Einflüsse von Vokaltrakt und Lippenabstrahlung mittels inverser Filterung beseitigt werden. In der neuen PSIAIF-Methode wird die glottale Pulsform dadurch bestimmt, daβ der IAIF-Algorithmus zweimal auf dasselbe Signal angewendet wird. Die erste IAIF-Analyse liefert als Ergebnis die Glottisanregung über mehrere Grundperioden. Diese Pulsform wird dann dazu benutzt, Rahmenpositionen und -längen für die grundperioden-synchrone Analyse zu ermitteln. Das endgültige Ergebnis erhält man, indem das ursprüngliche Sprachsignal in jeweils einer Periode mit dem IAIF-Algorithmus analysiert wird. Der PSIAIF-Algorithmus wurde sowohl unter Verwendung von synthetischen als auch natürlichen Vokalen erprobt. Die Ergebnisse zeigen, daβ das Verfahren dazu in der Lage ist, eine hinreichend genaue Schätzung des Glottissignals anzugeben, mit Ausnahme von Vokalen mit einer niedrigen Frequenz des ersten Formanten und welche durch eine gepresste Aussprache erzeugt werden.
Résumé
On présente une nouvelle méthode d'analyse du flux glottique: le PSIAIF (Pitch Synchronous Iterative Adaptive Inverse Filtering). Cet algorithme se base sur une méthode (IAIF) développée précédemment. La contribution glottique totale au spectre de la parole y était tout d'abord évaluée itérativement. La fonction de transfert du conduit vocal était obtenue après élimination de la contribution glottique moyenne tandis que l'excitation glottique l'était en annulant les effets du conduit vocal et de la radiation labiale par filtrage inverse. Dans la nouvelle méthode, l'onde glottique est calculée en appliquant deux fois l'algorithme IAIF au même signal. La première analyse donne une estimation de l'excitation glottique qui s'étend sur plusieurs périodes. L'onde ainsi obtenue est utilisée ensuite pour déterminer les positions et les longueurs des fenêtres d'analyse synchronisées. Pour obtenir le résultat final, il ne rest plus qu'à analyser le signal original de la parole, période fondamentale par période fondamentale, avec l'algorithme IAIF. L'algorithme PSIAIF a été appliqué à l'analyse du signal glottique, dans le cas de voyelles naturelles et synthétiques. Les résultats montrent que la méthode est capable de fournir une estimation relativement précise de flux glottique, si l'on exclut l'analyse des voyelles à premier formant bas produites par un type de phonation pressée.
References (10)
- P. Alku et al.
A comparison of EGG and a new inverse filtering method in phonation change from breathy to normal
- P. Alku et al.
Analysis of glottal waveform in different phonation types using the new IAIF-method
- T.V. Ananthapadmanabha
Acoustic analysis of voice source dynamics
- B. Gold et al.
Analysis of digital and analog formant synthesizers
IEEE Trans. Audio Electroacoust.
(1968) - M.J. Hunt et al.
Interactive digital inverse filtering and its relation to linear prediction methods
Cited by (435)
What Might the Trombone Teach Us About the Singing Voice?—A Tutorial Review
2024, Journal of VoiceThe trombone and the male voice cover similar frequency ranges and, at a physical level, the basic anatomies of the voice and the trombone show some qualitative similarity: both have two vibrating flaps of muscular tissue (the vocal folds and the trombonist’s lips, respectively), and in each case, these are loaded acoustically by resonant ducts both upstream and downstream. There are also large differences. For example, the downstream ducts differ in length. The trombone usually operates with an oscillation frequency close to that of one of the downstream resonances; this is only occasionally true of the voice. Because the lips of a trombone player are much more readily accessible for experiments, they have yielded more detailed measurements of longitudinal and transverse motion, AC and DC pressures, and flow under varying acoustic loads. In normal operation, the downstream motion of the lips or vocal folds leads the lateral opening motion, resulting in a sweeping flow that leads the flow through the aperture. The relative timing of these flow components is related to the phases of the pressure across the tissues and the downstream acoustic load. Further, the work done on trombonists’ lips due to their sweeping motion makes an important contribution to maintaining oscillation with both inertive and compliant acoustic loads. This probably explains why trombonists can “lip” the pitch smoothly from above to below a downstream resonance. Similar calculations on measurements of vocal fold motion show a similar work contribution and suggest that this sweeping motion is significant in powering this component of laryngeal motion. Comparing and contrasting the operation of the two “instruments” gives new perspectives on the basic science of the voice, with practical applications including the use of resonances. This could be helpful to voice scientists but also useful background knowledge for singers and singing teachers.
Bibliometric Analysis and Review of Literature on the Relationship Between Voice Production and Bilingualism
2024, Journal of VoiceThe spoken language may influence the way we produce our voice. Therefore, speak more than one language could affect voice production. However, there is limited evidence on the relationship between voice production and bilingualism.
To characterize the relationship between bilingualism and voice production determined by means of voice acoustic parameters and voice perception.
A review of literature was performed using four computerized databases. The keywords used were voice, bilingualism, and multilingualism. Relevant data were extracted from the publications on country and year of study, participants, sample size, characterization of voice production (acoustics, perceptual), and languages. A bibliometric analysis was performed for co-occurrence analysis. Quality assessment was conducted with the EPHPP tool.
A total of 10 publications met our inclusion criteria. Nine of the articles compared English productions with another language. Included studies reported changes in fundamental frequency between English and Spanish, Cantonese Chinese, Mandarin, Russian, Finnish, Arabic, Korean, Welsh, and Hebrew. The co-occurrence analysis shows that fundamental frequency is the most reported parameter, and English is the most compared language among studies including bilingual speakers.
Speaking a second language influences voice production determined by voice acoustic parameters, such as fundamental frequency. Therefore, language may be an important factor to take into consideration when planning assessment and intervention programs on the field of voice.
Voice Feature Selection to Improve Performance of Machine Learning Models for Voice Production Inversion
2023, Journal of VoiceEstimation of physiological control parameters of the vocal system from the produced voice outcome has important applications in clinical management of voice disorders . Previously we developed a simulation-based neural network for estimation of vocal fold geometry, mechanical properties, and subglottal pressure from voice outcome features that characterize the acoustics of the produced voice. The goals of this study are to (1) explore the possibility of improving the estimation accuracy of physiological control parameters by including voice outcome features characterizing vocal fold vibration; and (2) identify voice feature sets that optimize both estimation accuracy and robustness to measurement noise.
Feedforward neural networks are trained to solve the inversion problem of estimating the physiological control parameters of a three-dimensional body-cover vocal fold model from different sets of voice outcome features that characterize the simulated voice acoustics, glottal flow, and vocal fold vibration. A sensitivity analysis is then performed to evaluate the contribution of individual voice features to the overall performance of the neural networks in estimating the physiologic control parameters.
While including voice outcome features characterizing vocal fold vibration increases estimation accuracy, it also reduces the network's robustness to measurement noise, due to high sensitivity of network performance to voice outcome features measuring the absolute amplitudes of the glottal flow and area waveforms, which are also difficult to measure accurately in practical applications. By excluding such glottal flow-based features and replacing glottal area-based features by their normalized counterparts, we are able to significantly improve both estimation accuracy and robustness to noise. We further show that similar estimation accuracy and robustness can be achieved with an even smaller set of voice outcome features by excluding features of small sensitivity.
Review of analysis methods for speech applications
2023, Speech CommunicationThis paper reviews methods used to analyze speech signals for various applications such as automatic recognition of associated text and speaker identity, and coding. The survey focuses on the requirements of different applications, as diverse speech tasks have often used the same methods, despite having very different objectives. As relevant information in a speech signal is distributed highly non-uniformly, a variety of time and frequency analysis techniques is examined. The utility of methods is noted in terms of performance, using accuracy, complexity, cost, and latency as criteria.
Pathological voice classification based on the features of an asymmetric fluid–structure interaction vocal cord model
2023, Applied AcousticsCurrently, the acoustic features used for classifying pathological voices do not take the physiological structure and vibration characteristics of the vocal cords into account. We proposed a feature extraction method based on the asymmetric fluid–structure interaction vocal cord model. The airflow viscous force was introduced into this model. The left and right asymmetric coefficients of the upper mass and the lower mass were designed to characterize the vibration mechanism of the diseased vocal cord. Then a hybrid simulated annealing genetic algorithm (HSAGA) optimized the vocal cord model inversion parameters to match the target voice source. The vocal cord model parameters corresponding to the actual voice signal were extracted and combined with the phonation pressure threshold and the phonation flow threshold to classify pathological voices. The experimental results (deposited in the Massachusetts Eye and Ear Infirmary (MEEI) database) showed that the weighted average relative error of the objective optimization function classifying the vibration characteristics of the vocal cords with lesions calculated by the HSAGA was 4.0%. When classifying pathological voices, the accuracy of the vocal cord model parameters was 77.18%, which is 33.18% higher than classifications using acoustic features, and effectively distinguishes different types of pathological voice.
Automated Assessment of Glottal Dysfunction Through Unified Acoustic Voice Analysis
2022, Journal of VoiceThis paper uses the recent glottal flow model for iterative adaptive inverse filtering to analyze recordings from dysfunctional speakers, namely those with larynx-related impairment such as laryngectomy. The analytical model allows extraction of the voice source spectrum, described by a compact set of parameters. This single model is used to visualize and better understand speech production characteristics across impaired and nonimpaired voices. The analysis reveals some discriminative aspects of the source model which map to a physiological class description of those impairments. Furthermore, being based on analysis of source parameters only, it is complementary to any existing techniques of vocal-tract or phonetic analysis. The results indicate the potential for future automated speech reconstruction systems that adapt to the method of reconstruction required, as well as being useful for mainstream speech systems, such as ASR, in which front-end analysis can direct back-end models to suit characteristics of impaired speech.