An open-source toolbox for measuring vocal tract shape from real-time magnetic resonance images.

Real-time magnetic resonance imaging (rtMRI) is a technique that provides high-contrast videographic data of human anatomy in motion. Applied to the vocal tract, it is a powerful method for capturing the dynamics of speech and other vocal behaviours by imaging structures internal to the mouth and throat. These images provide a means of studying the physiological basis for speech, singing, expressions of emotion, and swallowing that are otherwise not accessible for external observation. However, taking quantitative measurements from these images is notoriously difficult. We introduce a signal processing pipeline that produces outlines of the vocal tract from the lips to the larynx as a quantification of the dynamic morphology of the vocal tract. Our approach performs simple tissue classification, but constrained to a researcher-specified region of interest. This combination facilitates feature extraction while retaining the domain-specific expertise of a human analyst. We demonstrate that this pipeline generalises well across datasets covering behaviours such as speech, vocal size exaggeration, laughter, and whistling, as well as producing reliable outcomes across analysts, particularly among users with domain-specific expertise. With this article, we make this pipeline available for immediate use by the research community, and further suggest that it may contribute to the continued development of fully automated methods based on deep learning algorithms.

Pterygoid muscle activity in speech: A preliminary investigation.

BACKGROUND: Speaking depends on refined control of jaw opening and closing movements. The medial pterygoid muscle (MPT), involved in jaw closing, and the lateral pterygoid muscle (LPT), involved in jaw opening, are two key mandibular muscles in mastication and are likely to be recruited for controlled movements in speech. OBJECTIVES: Three hypotheses were investigated, that during speech the MPT and LPT: (1) were both active, (2) but exhibited different patterns of activity, (3) which fluctuated with the vowels and consonants in speech. METHODS: Intramuscular EMG recordings were made from the right inferior head of the LPT and/or the right MPT in five participants during production of 40 target nonsense words (NWs) consisting of three syllables in the form /V1 C1 V2 C2 ə/ (V = vowel; C = consonant; ə = unstressed, reduced vowel), spoken by each participant 10 times per NW; analysis focussed on the target syllable, C1 V2 . RESULTS: Both MPT and LPT exhibited robust increases in EMG activity during utterance of most NWs, relative to rest. Peak LPT activation was time-locked to the final part of the target consonant (C1 ) interval when the jaw begins opening for the target vowel (V2 ), whereas peak MPT activation occurred around the temporal midpoint of V2 , when the jaw begins closing for C2 . EMG amplitude peaks differed in magnitude between "high" vowels, i.e., for which the tongue/jaw are high (e.g., in SEEK), and "low" vowels, i.e., for which the tongue/jaw are low (e.g., in SOCK). CONCLUSIONS: These novel findings suggest a key role for the LPT and MPT in the fine control of speech production. They imply that speech may impose major synergistic demands on the activities of the MPT and the LPT, and thereby provide insights into the possible interactions between speech activities and orofacial activities (e.g. mastication) and conditions (e.g. Temporomandibular Disorders) that involve the masticatory muscles.

A practical method of estimating the time-varying degree of vowel nasalization from acoustic features.

This paper presents a simple and easy-to-use method of creating a time-varying signal of the degree of nasalization in vowels, generated from acoustic features measured in oral and nasalized vowel contexts. The method is presented for separate models constructed using two sets of acoustic features: (1) an uninformed set of 13 Mel-frequency cepstral coefficients (MFCCs) and (2) a combination of the 13 MFCCs and a phonetically informed set of 20 acoustic features of vowel nasality derived from previous research. Both models are compared against two traditional approaches to estimating vowel nasalization from acoustics: A1-P0 and A1-P1, as well as their formant-compensated counterparts. Data include productions from six speakers of different language backgrounds, producing 11 different qualities within the vowel quadrilateral. The results generated from each of the methods are compared against nasometric measurements, representing an objective "ground truth" of the degree of nasalization. The results suggest that the proposed method is more robust than conventional acoustic approaches, generating signals which correlate strongly with nasometric measures across all vowel qualities and all speakers and accurately approximate the time-varying change in the degree of nasalization. Finally, an experimental example is provided to help researchers implement the method in their own study designs.

An acoustic description of Mixean Basque.

This paper presents an acoustic analysis of Mixean Low Navarrese, an endangered variety of Basque. The manuscript includes an overview of previous acoustic studies performed on different Basque varieties in order to synthesize the sparse acoustic descriptions of the language that are available. This synthesis serves as a basis for the acoustic analysis performed in the current study, in which the various acoustic analyses given in previous studies are replicated in a single, cohesive general acoustic description of Mixean Basque. The analyses include formant and duration measurements for the six-vowel system, voice onset time measurements for the three-way stop system, spectral center of gravity for the sibilants, and number of lingual contacts in the alveolar rhotic tap and trill. Important findings include: a centralized realization ([ʉ]) of the high-front rounded vowel usually described as /y/; a data-driven confirmation of the three-way laryngeal opposition in the stop system; evidence in support of an alveolo-palatal to apical sibilant merger; and the discovery of a possible incipient merger of rhotics. These results show how using experimental acoustic methods to study under-represented linguistic varieties can result in revelations of sound patterns otherwise undescribed in more commonly studied varieties of the same language.

Three-dimensional printable ultrasound transducer stabilization system.

When using ultrasound imaging of the tongue for speech recording/research, submental transducer stabilization is required to prevent the ultrasound transducer from translating or rotating in relation to the tongue. An iterative prototype of a lightweight three-dimensional-printable wearable ultrasound transducer stabilization system that allows flexible jaw motion and free head movement is presented. The system is completely non-metallic, eliminating interference with co-recorded signals, thus permitting co-collection and co-registration with articulometry systems. A motion study of the final version demonstrates that transducer rotation is limited to 1.25° and translation to 2.5 mm-well within accepted tolerances.

Using ultrasound and nasalance to separate oral and nasal contributions to formant frequencies of nasalized vowels.

The experimental method described in this manuscript offers a possible means to address a well known issue in research on the independent effects of nasalization on vowel acoustics: given that the separate transfer functions associated with the oral and nasal cavities are merged in the acoustic signal, the task of teasing apart the respective effects of the two cavities seems to be an intractable problem. The proposed method uses ultrasound and nasalance to predict the effect of lingual configuration on formant frequencies of nasalized vowels, thus accounting for acoustic variation due to changing lingual posture and excluding its contribution to the acoustic signal. The results reveal that the independent effect of nasalization on the acoustic vowel quadrilateral resembles a counter-clockwise chain shift of nasal compared to non-nasal vowels. The results from the productions of 11 vowels by six speakers of different language backgrounds are compared to predictions presented in previous modeling studies, as well as discussed in the light of sound change of nasal vowel systems.

The origins of babytalk: smiling, teaching or social convergence?

When addressing their young infants, parents systematically modify their speech. Such infant-directed speech (IDS) contains exaggerated vowel formants, which have been proposed to foster language development via articulation of more distinct speech sounds. Here, this assumption is rigorously tested using both acoustic and, for the first time, fine-grained articulatory measures. Mothers were recorded speaking to their infant and to another adult, and measures were taken of their acoustic vowel space, their tongue and lip movements and the length of their vocal tract. Results showed that infant- but not adult-directed speech contains acoustically exaggerated vowels, and these are not the product of adjustments to tongue or to lip movements. Rather, they are the product of a shortened vocal tract due to a raised larynx, which can be ascribed to speakers' unconscious effort to appear smaller and more non-threatening to the young infant. This adjustment in IDS may be a vestige of early mother-infant interactions, which had as its primary purpose the transmission of non-aggressiveness and/or a primitive manifestation of pre-linguistic vocal social convergence of the mother to her infant. With the advent of human language, this vestige then acquired a secondary purpose-facilitating language acquisition via the serendipitously exaggerated vowels.

A comparison of acoustic and articulatory methods for analyzing vowel differences across dialects: Data from American and Australian English.

In studies of dialect variation, the articulatory nature of vowels is sometimes inferred from formant values using the following heuristic: F1 is inversely correlated with tongue height and F2 is inversely correlated with tongue backness. This study compared vowel formants and corresponding lingual articulation in two dialects of English, standard North American English, and Australian English. Five speakers of North American English and four speakers of Australian English were recorded producing multiple repetitions of ten monophthongs embedded in the /sVd/ context. Simultaneous articulatory data were collected using electromagnetic articulography. Results show that there are significant correlations between tongue position and formants in the direction predicted by the heuristic but also that the relations implied by the heuristic break down under specific conditions. Articulatory vowel spaces, based on tongue dorsum position, and acoustic vowel spaces, based on formants, show systematic misalignment due in part to the influence of other articulatory factors, including lip rounding and tongue curvature on formant values. Incorporating these dimensions into dialect comparison yields a richer description and a more robust understanding of how vowel formant patterns are reproduced within and across dialects.

The articulatory dynamics of pre-velar and pre-nasal /æ/-raising in English: An ultrasound study.

Most dialects of North American English exhibit /æ/-raising in some phonological contexts. Both the conditioning environments and the temporal dynamics of the raising vary from region to region. To explore the articulatory basis of /æ/-raising across North American English dialects, acoustic and articulatory data were collected from a regionally diverse group of 24 English speakers from the United States, Canada, and the United Kingdom. A method for examining the temporal dynamics of speech directly from ultrasound video using EigenTongues decomposition [Hueber, Aversano, Chollet, Denby, Dreyfus, Oussar, Roussel, and Stone (2007). in IEEE International Conference on Acoustics, Speech and Signal Processing (Cascadilla, Honolulu, HI)] was applied to extract principal components of filtered images and linear regression to relate articulatory variation to its acoustic consequences. This technique was used to investigate the tongue movements involved in /æ/ production, in order to compare the tongue gestures involved in the various /æ/-raising patterns, and to relate them to their apparent phonetic motivations (nasalization, voicing, and tongue position).

High-resolution dynamic speech imaging with joint low-rank and sparsity constraints.

PURPOSE: To enable dynamic speech imaging with high spatiotemporal resolution and full-vocal-tract spatial coverage, leveraging recent advances in sparse sampling. METHODS: An imaging method is developed to enable high-speed dynamic speech imaging exploiting low-rank and sparsity of the dynamic images of articulatory motion during speech. The proposed method includes: (a) a novel data acquisition strategy that collects spiral navigators with high temporal frame rate and (b) an image reconstruction method that derives temporal subspaces from navigators and reconstructs high-resolution images from sparsely sampled data with joint low-rank and sparsity constraints. RESULTS: The proposed method has been systematically evaluated and validated through several dynamic speech experiments. A nominal imaging speed of 102 frames per second (fps) was achieved for a single-slice imaging protocol with a spatial resolution of 2.2 × 2.2 × 6.5 mm(3) . An eight-slice imaging protocol covering the entire vocal tract achieved a nominal imaging speed of 12.8 fps with the identical spatial resolution. The effectiveness of the proposed method and its practical utility was also demonstrated in a phonetic investigation. CONCLUSION: High spatiotemporal resolution with full-vocal-tract spatial coverage can be achieved for dynamic speech imaging experiments with low-rank and sparsity constraints.

Managing the distinctiveness of phonemic nasal vowels: articulatory evidence from Hindi.

There is increasing evidence that fine articulatory adjustments are made by speakers to reinforce and sometimes counteract the acoustic consequences of nasality. However, it is difficult to attribute the acoustic changes in nasal vowel spectra to either oral cavity configuration or to velopharyngeal opening (VPO). This paper takes the position that it is possible to disambiguate the effects of VPO and oropharyngeal configuration on the acoustic output of the vocal tract by studying the position and movement of the tongue and lips during the production of oral and nasal vowels. This paper uses simultaneously collected articulatory, acoustic, and nasal airflow data during the production of all oral and phonemically nasal vowels in Hindi (four speakers) to understand the consequences of the movements of oral articulators on the spectra of nasal vowels. For Hindi nasal vowels, the tongue body is generally lowered for back vowels, fronted for low vowels, and raised for front vowels (with respect to their oral congeners). These movements are generally supported by accompanying changes in the vowel spectra. In Hindi, the lowering of back nasal vowels may have originally served to enhance the acoustic salience of nasality, but has since engendered a nasal vowel chain shift.