Consonant Release Research Articles

Auto Annotation of Linguistic Features for Audio Deepfake Discernment

We present an innovative approach to auto-annotate Expert Defined Linguistic Features (EDLFs) as subsequences in audio time series to improve audio deepfake discernment. In our prior work, these linguistic features – namely pitch, pause, breath, consonant release bursts, and overall audio quality, labeled by experts on the entire audio signal – have been shown to improve detection of audio deepfakes with AI algorithms. We now expand our approach to pilot a way to auto annotate subsequences in the time series that correspond to each EDLF. We developed an ensemble of discords, i.e. anomalies in time series, detected using matrix profiles across multiple discord lengths to identify multiple types of EDLFs. Working closely with linguistic experts, we evaluated where discords overlapped with EDLFs in the audio signal data. Our ensemble method to detect discords across multiple discord lengths achieves much higher accuracy than using individual discord lengths to detect EDLFs. With this approach and domain validation we establish the feasibility of using time series subsequences to capture EDLFs to supplement annotation by domain experts, for improved audio deepfake detection.

Voiced aspirates with mixed voicing in Yemba, a Grassfields Bantu language of Cameroon

AbstractUsing electroglottography and acoustic measures, we characterize the strength and quality of voicing in voiced aspirated and unaspirated consonants (stops, fricatives, and approximants) in Yemba (Grassfields Bantu, Cameroon). We show that the Yemba voiced aspirates exhibit mixed voicing: modal voicing during the consonant constriction, but voiceless aspiration after release. Breathy or whispery phonation extends slightly into consonant constrictions preceding, and across the entire duration of vowels following, aspiration; this non-modal phonation extends further into prenasalized consonants. Mixed voicing has typically been excluded from the possible range of laryngeal–supralaryngeal coordinative patterns in consonants, and is thought to be unattested in the world’s languages; most previous work on this topic assumes that non-modal phonation after voiced consonant release is breathy-voiced. However, we argue that Yemba voiced aspirates differ from more commonly studied breathy-release aspirates only in the settings of some gestural parameters: the late glottal spread gesture is larger in magnitude and more resistant to coarticulation, yielding consistently devoiced aspiration which may even be more perceptually recoverable compared to breathiness.

المشاكل التي يواجهها الناطقون بالإنجليزية عند قراءة القرآن الكريم: عرضها ومعالجتها

In this article, I conducted research on the similarities and differences between consonants and vowels of both Arabic and English, so that it may help teachers of the Quran in English- speaking countries, for example, the United Kingdom. I explored the translation of very difficult terms used in tajweed such as قلقلة which is in the phonetics of English “consonant release” if this term is literally translated to the native speaker by saying that it is the giving of movement to a sound in the place of articulationمخرج they will not understand it. Similarly, I explored the terms the place of articulation and manner of articulation for مخرج and صفة consecutively, homorganic for مشترك المخرج from the books of phonetics. If the Arabic terms are translated in this way, for example, the exit of sound it will be difficult for the students to understand. We usually find in the books that some consonants like ج، ذال ، وزاء are equivalent to Z , th, and ز sound , but I added the condition that they must be between two voiced sounds like vowels. For example, I concluded that ذال is like “th” in “within”, because in other contexts they become devoiced and resemble ث، س and چ. Dark L of English was judged as equal to لام مفخمة and it was made clear that if the teachers of Quranic recitation translate it to dignified L or honored L the native students will not understand them. It was suggested to the teachers to obverse the كاف sound in the reading of the students when it is in the stressed syllables such as "كلّا، "كالوا"so that aspiration after كاف is checked, because such kind of aspiration does not exist in Arabic. Similarly, when نون ساكن occurs before كاف in words like “إن كنتم" English speakers are liable to add ng sound (like that in “bank” of English) before كاف. The article has covered much information, the most important of which has been put forth in the summary also.

Perception of consonantal release: English /l/ and /n/ by Korean listeners

Perception of consonantal release: English /l/ and /n/ by Korean listeners

Perception of consonantal release: English /l/ and /n/ by Korean listeners

Perception of consonantal release: English /l/ and /n/ by Korean listeners

Rasgos acústicos de las obstruyentes simples y eyectivas en el maya yucateco del Camino Real: bases para una caracterización dialectal.

El presente análisis de las obstruyentes eyectivas y simples del maya yucateco en contextos pre y postvocálicos contribuye a la necesidad de contar con trabajos que identifiquen acústicamente a los miembros no pulmonares de este contraste. Los resultados de nuestro estudio sugieren la existencia de pistas acústicas bien definidas para los sonidos en cuestión. Las oclusivas simples y eyectivas prevocálicas del maya yucateco pueden distinguirse mínimamente por la duración del voice onset time (VOT) y la intensidad de la explosión; el mismo contraste en posición postvocálica se identifica por diferencias en la duración del periodo de silencio entre el cierre de la vocal y la explosión de la oclusiva, así como por la intensidad misma de esta explosión. Por su parte, las africadas simples y eyectivas prevocálicas contrastan acústicamente tanto por la duración e intensidad del ruido, como por el VOT y la presencia/ausencia de silencio previo a la liberación de la oclusión glotal. Finalmente, el contraste simple/eyectiva en africadas postvocálicas está dado por diferencias en la duración del silencio tras el cierre vocálico, así como por la intensidad y la duración de la fricción. Original recibido: 2016/10/16Dictamen enviado al autor: 2017/04/15Aceptado: 2017/11/09

The Mechanism and Representation of Korean Three-Way Phonation Contrast: External Photoglottography, Intra-Oral Air Pressure, Airflow, and Acoustic Data

This paper aims to refine our understanding of the speech mechanism and laryngeal features involved in the Korean lenis (/p t k/), aspirated (/ph th kh/), and fortis (/p' t' k'/) plosives. For this purpose we made measurements using a new noninvasive technique called external lighting and sensing photoglottography (ePGG) as well as intra-oral air pressure (Pio) above the glottis, airflow, and acoustic data. From simultaneous recordings of the experimental data, we were ableto quantify the laryngeal-oral coordination of glottal opening and a consonant release, and the covariance of airflow peak and duration of aspiration with glottal opening. The phasing of glottal opening and the 3-way phonation contrast occurs in the order, from early to late, fortis, lenis < aspirated plosives, and the glottal opening peak ranges from low to high in the same order. We also found that a Pio peak, the durations of a high Pio plateau and an oral closure, and F0 are independent of the glottal opening mechanism, varying in the order lenis < aspirated and fortis plosives. From these findings, we propose that the 2 independent patterns are accounted for by the articulator-based features [±spread glottis] and [±tense], respectively.

Effects of Stress, Stop Release, and Familiarization on Speech Recognition Thresholds

Objective: This study examined outcomes of common procedural variations of speech recognition threshold (SRT) testing, specifically related to the effects of equal syllable stress, word-final stop consonant release, and prior-familiarization, with the participants’ language status taken into account. Methods: SRTs were obtained from 40 adults with normal hearing. Twenty participants received prior-familiarization with the spondee list and the other 20 received no prior-familiarization. Repeated SRT tests were administered using three different recordings which varied in syllable stress and word-final stop release patterns. Results: The group with prior-familiarization demonstrated a threshold that was significantly lower than the group without prior-familiarization, by approximately 5 dB HL. Despite the statistically significant effects of equal syllable stress and word-final stop release on SRTs, the magnitude of SRT changes elicited by these acoustic-phonetic variations was only slightly above 1 dB HL. The monolinguals generally outperformed the bilinguals in SRT outcomes with the threshold difference less than 3 dB HL. Conclusion: Findings from the present study suggest that familiarizing listeners with test vocabulary prior to SRT administration should continue to remain an important procedural requirement. Future research addressing the extent to which acoustic-phonetic variations of spondee production affect SRTs in individuals with hearing impairments is warranted.

Durational Interaction of Stops and Vowels in English and Korean Child-Directed Speech

The current study observes the durational interaction of tautosyllabic consonants and vowels in the word-initial position of English and Korean child-directed speech (CDS). The effect of phonological laryngeal contrasts in stops on the following vowel duration, and the effect of the intrinsic vowel duration on the release duration of preceding stops in addition to the acoustic realization of the contrastive segments are explored in different prosodic contexts - phrase-initial/medial, focal accented/non-focused - in a marked speech style of CDS. A trade-off relationship between Voice Onset Time (VOT), as consonant release duration, and voicing phonation time, as vowel duration, reported from adult-to-adult speech, and patterns of durational variability are investigated in CDS of two languages with different linguistic rhythms, under systematically controlled prosodic contexts. Speech data were collected from four native English mothers and four native Korean mothers who were talking to their one-word staged infants. In addition to the acoustic measurements, the transformed delta measure is employed as a variability index of individual tokens. Results confirm the durational correlation between prevocalic consonants and following vowels. The interaction is revealed in a compensatory pattern such as longer VOTs followed by shorter vowel durations in both languages. An asymmetry is found in CV interaction in that the effect of consonant on vowel duration is greater than the VOT differences induced by the vowel. Prosodic effects are found such that the acoustic difference is enhanced between the contrastive segments under focal accent, supporting the paradigmatic strengthening effect. Positional variation, however, does not show any systematic effects on the variations of the measured acoustic quantities. Overall vowel duration and syllable duration are longer in English tokens but involve less variability across the prosodic variations. The constancy of syllable duration, therefore, is not found to be more strongly sustained in Korean CDS. The stylistic variation is discussed in relation to the listener under linguistic development in CDS.

Patterns of development in stop-release cues to coda contrasts in American-English-speaking 2-year-olds.

A distinctive characteristic of early speech is “exaggerated” release of stop consonants including multiple bursts and heavy post-release noise [Imbrie (2005)]. This study provides an in-depth investigation of changes in the acoustic characteristics of stop coda release from 1;6 to 2;6 years, focusing on their interaction with emerging linguistic contrasts. We examined how various non-spectral coda-release-related cues, such as the presence and duration of post-release noise, varied with coda voicing and place (alveolar versus velar) in spontaneous speech from six American-English-speaking mother-child dyads. Starting as early as 1;6, children produced many of the same types of correlates of voicing and place as adults did, but they exhibited more frequent occurrences and longer durations of these cues compared to mothers; this exaggeration decreased by 2;6. In utterance-medial position, where the decrease was particularly striking, it occurred in a selective, stepwise fashion rather than simultaneously in all contexts. That is, the decrease was larger for voiceless codas than for voiced codas, and for place contrasts, the values decreased first for velars, and only later for alveolars. This observation raises the question of whether such selective fine-tuning is characteristic of phonological development across languages, and if so, what the underlying mechanisms might be. [Work supported by NIH R01HD057606.]

Prosodic conditioning of phonetic detail in German plosives

This study investigates the prosodic conditioning of phonetic details which are candidate cues to phonological contrasts. German /b, d, g, p, t, k/ were examined in three prosodic positions. Lenis plosives /b, d, g/ were produced with less glottal vibration at larger prosodic boundaries, whereas their VOT showed no effect of prosody. VOT of fortis plosives /p, t, k/ decreased at larger boundaries, as did their burst intensity maximum. Vowels (when measured from consonantal release) following fortis plosives and lenis velars were shorter after larger boundaries. Closure duration, which did not contribute to the fortis/lenis contrast, was heavily affected by prosody. These results support neither of the hitherto proposed accounts of prosodic strengthening (Uniform Strengthening and Feature Enhancement). We propose a different account, stating that the phonological identity of speech sounds remains stable not only within, but also across prosodic positions (contrast-over-prosody hypothesis). Domain-initial strengthening hardly diminishes the contrast between prosodically weak fortis and strong lenis plosives.

Detecting anticipatory effects in speech articulation by means of spectral coefficient analyses

Few acoustic studies have attempted to examine anticipatory effects in the earliest part of the release of stop consonants. We investigated the ability of spectral coefficients to reveal anticipatory coarticulation in the burst and early aspiration of stops in monosyllables. Twenty American English speakers produced stop (/k,t,p/) – vowel (/æ,i,o/) – stop (/k,t,p/) sequences in two phrase positions. The first four spectral coefficients (mean, standard deviation, skewness, kurtosis) were calculated for one window centered on the burst of the onset consonant and two subsequent, non-overlapping windows. All coefficients showed an influence of vowel-to-consonant anticipatory coarticulation. Which onset consonant showed the strongest vowel effect depended on the specific coefficient under consideration. A context-dependent consonant-to-consonant anticipatory effect was observed for onset /p/. Findings demonstrate that spectral coefficients can reveal subtle anticipatory adjustments as early as the burst of stop consonants. Different results for the four coefficients suggest that comprehensive spectral analyses offer advantages over other approaches. Studies using these techniques may expose previously unobserved articulatory adjustments among phonetic contexts or speaker populations.

Typological implications of Kalam predictable vowels

AbstractKalam is a Trans New Guinea language of Papua New Guinea. Kalam has two distinct vowel types: full vowels /a e o/, which are of relatively long duration and stressed, and reduced central vowels, which are shorter and often unstressed, and occur predictably within word-internal consonant clusters and in monoconsonantal utterances. The predictable nature of the reduced vowels has led earlier researchers, e.g. Biggs (1963) and Pawley (1966), to suggest that they are a non-phonemic ‘consonant release’ feature, leading to lexical representations with long consonant strings and vowelless words. Here we compare Kalam to other languages with similar sound patterns and assess the implications for phonological theory in the context of Hall's (2006) typology of inserted vowels. We suggest that future work on predictable vowels should explore the extent to which clusters of properties are explained by evolutionary pathways.

Phonetic Description of a Three-way Stop Contrast in Northern Paiute

This paper presents the phonetic description of a three-way phonemic contrast in the medial stops (lenis, fortis, and voiced fortis stops) of a southern dialect of Northern Paiute. Qualitative and quantitative analysis of VOT, closure duration, and voice quality was performed on field recordings of a female speaker from the 1950s. The findings include that: 1) voiced fortis stops are realized phonetically as voiceless unaspirated stops; 2) the difference between fortis and voiced fortis and between voiced fortis and lenis in terms of VOT is subtle; 3) consonantal duration is a robust acoustic characteristic differentiating the three classes of stops; 4) lenis stops are characterized by a smooth VC transition, while fortis stops often exhibit aspiration at the VC juncture, and voiced fortis stops exhibit occasional glottalization at the VC juncture. These findings suggest that the three-way contrast is realized by combination of multiple phonetic properties, particularly the properties that occur at the vowel-consonant boundary rather than the consonantal release.

Asymmetry of voice onset time-processing in adult developmental dyslexics

Objective The human auditory cortex codes speech temporally according to sequential acoustico-phonetic cues like the voice onset time (VOT). This coding is predominantly left-lateralized in normal readers. We examined VOT-processing asymmetries in adults with a history of developmental dyslexia (DD-history+). Methods Auditory-evoked potentials (AEPs) to voiced (/ba/) and voiceless (/pa/) speech stimuli were recorded from 10 DD-history+ adults and 8 controls. Source modelling of the “release component” (RC: ∼240 ms; time-locked to voiced consonantal release and considered reflective of VOT-processing) was conducted to explore VOT asymmetries. Results Controls demonstrated L > R RC source probe amplitude asymmetry in the auditory cortex. DD-history+ subjects with little persistent reading deficit ( n = 5) demonstrated normal temporal coding but rightward asymmetry. DD-history+ subjects with severe persistent deficits ( n = 5) exhibited numerous supplemental AEP components (notably left hemispheric) and inconsistent asymmetry (leftward or alternating). Conclusions These preliminary findings suggest that DD-history+ adults process auditory speech cues differently than adults without previous DD. The nature of this processing may relate to the severity of persistent reading deficits. Significance Previous dyslexics with little persistent deficit can exhibit atypical functional asymmetry with normal auditory temporal coding. Source modelling represents an effective, non-invasive means of exploring processing asymmetries in clinical populations.

Perceptual explanations of articulatory variability in the realisation of the nasal feature for the consonants

This paper illustrates how the perceptual constraints explain the observed articulatory and aerodynamic inter‐speakers and inter‐languages variability between initial nasal consonants (/n/), and unreleased final consonants (/N/). In the case of /n/, only a short, well‐defined region around the consonant release needs to be nasalized for the consonant to be perceived as nasal. In the case of /N/, because of the lack of a nasal release, a longer span is needed for deciphering the presence of nasalisation; contextual nasalization of the preceding vowel can be avoided, but in that case, the nasal murmur has to be long to be perceived (since the first part of the murmur is perceptually masked by the vowel), and/or /N/ has to be released. Such observations argue for the division of the nasal consonant into three parts, onset, murmur and release: the realisation of the nasal feature is aligned around the /N/ onset, and around the /n/ offset. It also argues for a better integration of the perceptual requirements and masking phenomena in modelling the observed variability. Finally, it shows that a feature can be realized very differently depending on the position of the phoneme in the syllable.

Nasal consonant production in Broca’s and Wernicke’s aphasics: Speech deficits and neuroanatomical correlates

The present study investigated the articulatory implementation deficits of Broca’s and Wernicke’s aphasics and their potential neuroanatomical correlates. Five Broca’s aphasics, two Wernicke’s aphasics, and four age-matched normal speakers produced consonant–vowel–(consonant) real word tokens consisting of [m, n] followed by [i, e, a, o, u]. Three acoustic measures were analyzed corresponding to different properties of articulatory implementation: murmur duration (a measure of timing), amplitude of the first harmonic at consonantal release (a measure of articulatory coordination), and murmur amplitude over time (a measure of laryngeal control). Results showed that Broca’s aphasics displayed impairments in all of these parameters, whereas Wernicke’s aphasics only exhibited greater variability in the production of two of the parameters. The lesion extent data showed that damage in either Broca’s area or the insula cortex was not predictive of the severity of the speech output impairment. Instead, lesions in the upper and lower motor face areas and the supplementary motor area resulted in the most severe implementation impairments. For the Wernicke’s aphasics, the posterior areas (superior marginal gyrus, parietal, and sensory) appear to be involved in the retrieval and encoding of lexical forms for speech production, resulting in increased variability in speech production.

On distinctions between stops and similar-place weak fricatives

There is an extensive body of literature on the acoustic properties of both stops and fricatives. However, little attention has been paid to the acoustic features that might distinguish these two sound categories. This is unsurprising in the case of stops versus strong fricatives since these sounds are rarely confused. Stops and weak fricatives, on the other hand, are frequently confused [G.A. Miller and P.E. Nicely, J. Acoust. Soc. Am. 27, 338–352 (1955)]. The present study was undertaken in a search for acoustic features that might distinguish the stop/weak fricative pairs /b/–/v/ and /d/–/dh/ (i.e., /d/ vs voiced th). Speech material consisted of CV and VCV syllables spoken by five men and five women, using the vowels /a/, /i/, and /u/. A combination of two features reliably separated the stops from weak fricatives: (1) intensity during the consonant occlusion interval (typically greater for the fricatives), and (2) the rate of increase in mid- and high-frequency energy (above 1 kHz) associated with consonant release (typically greater for the stops).

Interacting effects of phrasal and syllable position on consonant production

The complexities of how prosodic structure, both at the phrasal and syllable levels, shapes speech production have begun to be illuminated through studies of articulatory behavior. Here, we pursue the goal of understanding prosodic signatures on articulation by examining the effects of phrasal and syllable position on the constriction formation and release of consonants. Articulatory kinematic data were collected for five subjects using electromagnetic articulography (EMA) to record target consonants (labial, labiodental, and tongue tip), located in (1) either syllable final or initial position and (2) either at a phrase edge or phrase-medially: ♯C, ♯♯C, C♯, C♯♯. The duration, displacement, and time-to-peak velocity of constriction formation and release were determined for the target consonants based on kinematic landmarks in the articulator velocity profiles (zero crossings and extrema). ANOVA determines that syllable and phrasal position consistently affect the movement duration; however, effects on displacement were more variable. For the majority of subjects, the boundary-adjacent portion of the movement (release for a preboundary coda and constriction formation for a postboundary onset) is not differentially affected in terms of phrasal lengthening; both lengthen equivalently. However, two subjects show an interaction such that the codas are lengthened more than the onsets. [Work supported by NIH.]

Physiological evidence for a temporal processing mechanism underlying voice-onset time (VOT) encoding

Despite decades of psychoacoustical research, the detailed neural mechanisms underlying VOT encoding remain obscure. Evidence collected from direct recordings in auditory cortex of human subjects undergoing surgical evaluation for medically intractable epilepsy, and from primary auditory cortex in monkeys, supports a temporal processing mechanism as a principal means by which VOT is encoded by the brain. This mechanism, as proposed by Pisoni [J. Acoust. Soc. Am. 77, 1352–1361 (1977)], argues that the perceptual discrimination of voiced from unvoiced stop consonants is based, in part, on whether consonant release and voicing onset are perceived as occurring simultaneously or sequentially. Neural activity in auditory cortex offers physiologically plausible parallels to this perceptual scheme that can help account for the distribution of typical VOT values used by the majority of the world’s languages, categorical perception of VOT, and perceptual boundary shifts that occur with changes in stop-consonant place of articulation and when nonspeech analogs of VOT are used. These responses in primary auditory cortex are poised to provide powerful inputs to later processing areas, where they can be integrated with other acoustical, visual, and language-related inputs known to modulate VOT perception. [Work supported by DC00657 and DC00120.]