Acoustic Evidence Research Articles

Ensemble majority voting classifier for speech emotion recognition and prediction

Purpose– The purpose of this paper is to understand the emotional state of a human being by capturing the speech utterances that are used during common conversation. Human beings except of thinking creatures are also sentimental and emotional organisms. There are six universal basic emotions plus a neutral emotion: happiness, surprise, fear, sadness, anger, disgust and neutral.Design/methodology/approach– It is proved that, given enough acoustic evidence, the emotional state of a person can be classified by an ensemble majority voting classifier. The proposed ensemble classifier is constructed over three base classifiers: k nearest neighbors, C4.5 and support vector machine (SVM) polynomial kernel.Findings– The proposed ensemble classifier achieves better performance than each base classifier. It is compared with two other ensemble classifiers: one-against-all (OAA) multiclass SVM with radial basis function kernels and OAA multiclass SVM with hybrid kernels. The proposed ensemble classifier achieves better performance than the other two ensemble classifiers.Originality/value– The current paper performs emotion classification with an ensemble majority voting classifier that combines three certain types of base classifiers which are of low computational complexity. The base classifiers stem from different theoretical background to avoid bias and redundancy. It gives to the proposed ensemble classifier the ability to be generalized in the emotion domain space.

Nivkh palatalisation: articulatory causes and perceptual effects

Nivkh, a linguistic isolate of Eastern Siberia, displays a typologically uncommon process which affects /e/ in the context of a following uvular consonant, and which reportedly results in palatalisation of the consonant preceding /e/, e.g. /pʰeq/→[pʰʲq]. Phonologically, there are indeed reasonable grounds for analysing the consonants as palatalised. Phonetically, matters are less clear. Acoustic evidence shows that the clearest correlate of palatalisation is a steep and consistent fall of F2 during the realisation of the vowel. The fact that F2 lowering is also typical before uvulars suggests that palatalisation is the result of a perceptually motivated reanalysis, whereby the palatality of /e/ has come to be associated with the preceding consonant, and the vowel has a lowered and retracted reflex.

Czech trills revisited: An ultrasound EGG and acoustic study

The Czech language has two trills produced with the anterior tongue, represented in the Czech orthography as /r/ and /ř/. The phonetic characterization of the latter trill, currently transcribed in the IPA as [], has been especially controversial. The present study uses ultrasound, electroglottography (EGG) and acoustic evidence to examine these trills, with the particular aim of achieving a more phonetically grounded representation of /ř/. In the first part of the paper, ultrasound imaging is used to compare tongue positions between /r/ and /ř/, ruling out tongue height as the key factor distinguishing the two sounds. The second part uses EGG and acoustic evidence to examine phonatory aspects of production. Results implicate vocal fold abduction as the more robust factor distinguishing /r/ from /ř/. Consequently, we posit [] as a more appropriate phonetic characterization of /ř/.

Fluid emissions at the Aquitaine Shelf (Bay of Biscay, France): A biogenic origin or the expression of hydrocarbon leakage?

Fishery acoustic surveys conducted in the Bay of Biscay and dedicated to monitoring and predicting pelagic ecosystem evolution reveal numerous active seeps on the Aquitaine Shelf, east of the shelf break, at water depths ranging from 140 to 185m. Some acoustic anomalies recorded in the water column with hull-mounted single and multibeam echosounders are clearly caused by fluid escape at the seabed, most likely gases. These fluid emissions are associated at the seafloor with high backscatter subcircular small-scale mounds, on average less than 2m high and a few metres in diameter. Based on near-bottom video and acoustic surveys, these mounds are interpreted to be by-products of gas seepage, possibly methane-derived authigenic carbonates. The spatial distribution of the seeps and related structures, based on water column acoustic gas flares and high backscatter seabed patches, appears to be relatively broad, with a North–South extension of ~65km across the Parentis Basin and the Landes High, and a West–East extension along a few kilometres wide on the shelf. The seepage activity seems persistent through time at the annual scale, with acoustic evidence dating back to 1998. The spatial distribution of the fluid emissions at the Aquitaine Shelf may suggest possible sedimentary and tectonic controls in relation with the Pyrenean compression phase. The nature and the origin of the emitted fluids and seafloor mounds are unknown. The gases may correspond to biogenic methane from Late Pleistocene deposits or to thermogenic gases originating from deeper, Jurassic–Cretaceous levels. The oil province of the Parentis Basin raises questions regarding possible genetic links to the petroleum system.

Acoustic evidence of a submarine slide in the deepest part of the Arctic, the Molloy Hole

The western Svalbard continental margin contains thick sediment sequences with areas known to contain gas hydrates. Together with a dynamic tectonic environment, this makes the region prone to subm ...

Butqui c'est la différence? Discourse markers in Louisiana French: The case ofbutvs.mais

AbstractThis article examines the use of English discourse markers in Louisiana French, focusing in particular on Englishbutand its French counterpartmais. Based on data collected in Terrebonne and Lafourche Parishes, we examine the speech of bilinguals to determine the status of these markers, which provide a window onto the role of discourse markers in situations of language contact. Though the markers show an overlapping semantic and functional distribution,butmore often appears in the context of at least one pause. We also provide acoustic evidence and an analysis of the markers in different functions to conclude that the need for iconic contrast via language mixing (Maschler 1994, 1997; de Rooij 2000) is only one possible motivation for the use of foreign markers. We conclude that discourse markers may carry social meaning and be the site of identity construction as much as they are the site of text organization. (Discourse markers, bilingual discourse, codeswitching, language shift)*

Incipient tonogenesis in Phnom Penh Khmer: Acoustic and perceptual studies

Unlike many languages of Southeast Asia, Khmer (Cambodian) is not a tone language. However, in the colloquial speech of the capital Phnom Penh, /r/ is lost in onsets, reportedly supplanted by a range of other acoustic cues such as aspiration, a falling- or low-rising f0 contour, breathy voice quality, and in some cases diphthongization, e.g. /krɑː/‘poor’>[kɑ], [khɔˇɑ], [kɔ¨ɑ¨], /kruː/ ‘teacher’>[ku`ː], [khuˇː], [ku¨ː]. This paper presents the results of production and perception studies designed to shed light on this unusual sound change. Acoustic evidence shows that colloquial /CrV/ forms differ from reading pronunciation forms in terms of VOT, f0, and spectral balance measures, while a pair of perceptual studies demonstrate that f0 is a sufficient cue for listeners to distinguish underlying /CrV/-initial from /CV/-initial forms, but that F1 is not. I suggest that this sound change may have arisen via the perceptual reanalysis of changes in spectral balance, coupled with the coarticulatory influence of the dorsal gesture for /r/.

Acoustic evidence of distinctive temperatures in relaxor-multiferroics

We present an acoustic study of relaxor ferroelectrics with magnetic ordering PbFexB1−xO3 (B = Nb, Ta, W) in the wide temperature range from 100 to 750 K as well as relaxor PbMg1/3Nb2/3O3 at temperatures from 300 to 750 K. The longitudinal and transverse acoustic wave velocity and attenuation have been measured by the ultrasonic pulse-echo technique at frequencies of 5 and 10 MHz. The strong anomalies observed in the longitudinal sound velocity and attenuation versus temperature are correlated with a temperature range of the coexistence of relaxor ferroelectric and antiferromagnetic states. Attenuation peaks have been observed at distinctive temperatures for relaxors, i.e., at TB (Burns temperature) and T* (an additional distinctive temperature). Fluctuations of the polar nanoregion boundaries are considered as the source of the acoustic anomalies.

Acoustic Evidence for Phonologically Mismatched Speech Errors

Speech errors are generally said to accommodate to their new phonological context. This accommodation has been validated by several transcription studies. The transcription methodology is not the best choice for detecting errors at this level, however, as this type of error can be difficult to perceive. This paper presents an acoustic analysis of speech errors that uncovers non-accommodated or mismatch errors. A mismatch error is a sub-phonemic error that results in an incorrect surface phonology. This type of error could arise during the processing of phonological rules or they could be made at the motor level of implementation. The results of this work have important implications for both experimental and theoretical research. For experimentalists, it validates the tools used for error induction and the acoustic determination of errors free of the perceptual bias. For theorists, this methodology can be used to test the nature of the processes proposed in language production.

Acoustic Evidence for Hoary Bat Migration in the Coast Mountains of British Columbia

Acoustic Evidence for Hoary Bat Migration in the Coast Mountains of British Columbia

Killers in the dark—Acoustic evidence for night-time predation by mammal-eating killer whales (Orcinus orca) in Alaska

North Pacific killer whales that specialize on hunting marine mammals do not typically echolocate while searching for prey. This suggests that they detect prey by either relying on visual cues or listening for sounds generated by prey animals. If prey detection requires vision, hunting should be limited to the daylight hours. Documenting predation at night would therefore provide evidence supporting a passive listening hypothesis of prey detection. We used digital recording tags (DTAGs) to study the behavior of mammal-eating killer whales in Southeast Alaska. These tags recorded the underwater movements of the tagged individual and any sound emitted or received. Predation events were identified using distinctive sounds generated during prey capture and handling. We deployed 13 tags, of which 7 remained attached for at least part of the night. The majority of tags recorded night-time predation, even though nights were short (average of 4:18 h) during the study period. These findings show that mammal-eating killer whales can detect prey at night and thus suggest that passive listening is an important part of their hunting strategy. Acoustic data from digital recording tags can therefore provide valuable insights into the night-time activities and foraging behavior of killer whales and other marine mammals.

Geochemical and acoustic evidence for the occurrence of methane in sediments of the Polish sector of the southern Baltic Sea

This paper presents the results of geochemical and acoustic investigations of sediments in the Polish sector of the southern Baltic Sea. Its objective was to indicate areas of gas bubble formation and the occurrence of methane. Over 3000 nautical miles of transects were recorded using a variety of hydroacoustic instruments, and five coring points were selected for further analyses of pore waters (CH4, SO42−, H2S, NH4+, total alkalinity) and sediments (grain size distribution, Corg, Ntot, LOI and WC). Gas turned out to be present at shallow depths in different forms such as recent and buried pockmarks, and gas-saturated sediments (including gas pockets and seepages). It was found that methane was widespread in the sediments of the study area, both in the surface sediments, e.g. in the vicinity of the Hel Peninsula or in the central Gulf of Gdańsk, and in deeper sediment layers, e.g. in the Gdańsk Deep and the Słupsk Furrow. Chemical analysis showed that as a result of the rapid decomposition of organic matter, sulphates were depleted in the top 20cm layer of sediments and that methane was produced at relatively shallow depths (in some areas even at depths of 20–30cm bsf) compared to other regions of the Baltic, reaching concentrations of>6mmol l−1 in the 30–40cm layer below the sediment surface. The sulphate-methane transition zone (SMTZ) was 4–37cm thick and was situated in the uppermost 50cm of the sediments.

Level of interest sensing in spoken dialog using decision-level fusion of acoustic and lexical evidence

Automatic detection of a user's interest in spoken dialog plays an important role in many applications, such as tutoring systems and customer service systems. In this study, we propose a decision-level fusion approach using acoustic and lexical information to accurately sense a user's interest at the utterance level. Our system consists of three parts: acoustic/prosodic model, lexical model, and a model that combines their decisions for the final output. We use two different regression algorithms to complement each other for the acoustic model. For lexical information, in addition to the bag-of-words model, we propose new features including a level-of-interest value for each word, length information using the number of words, estimated speaking rate, silence in the utterance, and similarity with other utterances. We also investigate the effectiveness of using more automatic speech recognition (ASR) hypotheses (n-best lists) to extract lexical features. The outputs from the acoustic and lexical models are combined at the decision level. Our experiments show that combining acoustic evidence with lexical information improves level-of-interest detection performance, even when lexical features are extracted from ASR output with high word error rate.

An Information-Extraction Approach to Speech Processing: Analysis, Detection, Verification, and Recognition

The field of automatic speech recognition (ASR) has enjoyed more than 30 years of technology advances due to the extensive utilization of the hidden Markov model (HMM) framework and a concentrated effort by the speech community to make available a vast amount of speech and language resources, known today as the Big Data Paradigm. State-of-the-art ASR systems achieve a high recognition accuracy for well-formed utterances of a variety of languages by decoding speech into the most likely sequence of words among all possible sentences represented by a finite-state network (FSN) approximation of all the knowledge sources required by the ASR task. However, the ASR problem is still far from being solved because not all information available in the speech knowledge hierarchy can be directly integrated into the FSN to improve the ASR performance and enhance system robustness. It is believed that some of the current issues of integrating various knowledge sources in top-down integrated search can be partially addressed by processing techniques that take advantage of the full set of acoustic and language information in speech. It has long been postulated that human speech recognition (HSR) determines the linguistic identity of a sound based on detected evidence that exists at various levels of the speech knowledge hierarchy, ranging from acoustic phonetics to syntax and semantics. This calls for a bottom-up attribute detection and knowledge integration framework that links speech processing with information extraction, by spotting speech cues with a bank of attribute detectors, weighting and combining acoustic evidence to form cognitive hypotheses, and verifying these theories until a consistent recognition decision can be reached. The recently proposed automatic speech attribute transcription (ASAT) framework is an attempt to mimic some HSR capabilities with asynchronous speech event detection followed by bottom-up knowledge integration and verification. In the last few years, ASAT has demonstrated good potential and has been applied to a variety of existing applications in speech processing and information extraction.

Use of syntax in perceptual compensation for phonological reduction.

Listeners resolve ambiguity in speech by consulting context. Extensive research on this issue has largely relied on continua of sounds constructed to vary incrementally between two phonemic endpoints. In this study we presented listeners instead with phonetic ambiguity of a kind with which they have natural experience: varying degrees of word-final /t/-reduction. In two experiments, Dutch listeners decided whether or not the verb in a sentence such as Maar zij ren(t) soms 'But she sometimes run(s)' ended in /t/. In Dutch, presence versus absence of final /t/ distinguishes third- from first-person singular present-tense verbs. Acoustic evidence for /t/ varied from clear to absent, and immediately preceding phonetic context was consistent with more versus less likely deletion of /t/. In both experiments, listeners reported more /t/s in sentences in which /t/ would be syntactically correct. In Experiment 1, the disambiguating syntactic information preceded the target verb, as above, while in Experiment 2, it followed the verb. The syntactic bias was greater for fast than for slow responses in Experiment 1, but no such difference appeared in Experiment 2. We conclude that syntactic information does not directly influence pre-lexical processing, but is called upon in making phoneme decisions.

Feeling backwards? How temporal order in speech affects the time course of vocal emotion recognition

Recent studies suggest that the time course for recognizing vocal expressions of basic emotion in speech varies significantly by emotion type, implying that listeners uncover acoustic evidence about emotions at different rates in speech (e.g., fear is recognized most quickly whereas happiness and disgust are recognized relatively slowly; Pell and Kotz, 2011). To investigate whether vocal emotion recognition is largely dictated by the amount of time listeners are exposed to speech or the position of critical emotional cues in the utterance, 40 English participants judged the meaning of emotionally-inflected pseudo-utterances presented in a gating paradigm, where utterances were gated as a function of their syllable structure in segments of increasing duration from the end of the utterance (i.e., gated syllable-by-syllable from the offset rather than the onset of the stimulus). Accuracy for detecting six target emotions in each gate condition and the mean identification point for each emotion in milliseconds were analyzed and compared to results from Pell and Kotz (2011). We again found significant emotion-specific differences in the time needed to accurately recognize emotions from speech prosody, and new evidence that utterance-final syllables tended to facilitate listeners' accuracy in many conditions when compared to utterance-initial syllables. The time needed to recognize fear, anger, sadness, and neutral from speech cues was not influenced by how utterances were gated, although happiness and disgust were recognized significantly faster when listeners heard the end of utterances first. Our data provide new clues about the relative time course for recognizing vocally-expressed emotions within the 400–1200 ms time window, while highlighting that emotion recognition from prosody can be shaped by the temporal properties of speech.

Automatic Prosodic Break Detection and Feature Analysis

Automatic prosodic break detection and annotation are important for both speech understanding and natural speech synthesis. In this paper, we discuss automatic prosodic break detection and feature analysis. The contributions of the paper are two aspects. One is that we use classifier combination method to detect Mandarin and English prosodic break using acoustic, lexical and syntactic evidence. Our proposed method achieves better performance on both the Mandarin prosodic annotation corpus — Annotated Speech Corpus of Chinese Discourse and the English prosodic annotation corpus — Boston University Radio News Corpus when compared with the baseline system and other researches' experimental results. The other is the feature analysis for prosodic break detection. The functions of different features, such as duration, pitch, energy, and intensity, are analyzed and compared in Mandarin and English prosodic break detection. Based on the feature analysis, we also verify some linguistic conclusions.

Acoustic evidence for a gas migration and release system in Arctic glaciated continental margins offshore NW-Svalbard

High-resolution 3D and 2D seismic data offshore NW-Svalbard, west of Prins Karls Forland, provide geophysical evidence for geologically controlled fluid migration pathways, gas hydrate occurrence, and an active seabed gas expulsion system. The investigated seabed area covers ∼1600 km2 and lies between Kongsfjorden cross-shelf trough in the north and Isfjorden cross-shelf trough in the south ranging in water depths from ∼200 m on the shelf to 800 m on the upper continental slope. Acoustic evidence for present day methane release from the seabed to the water column comes from more than 220 gas flares at the outer shelf while past methane release activity at the mid-shelf area is evident from pockmarks without flares. The fluid migration pathways towards the seabed can be drawn from sub-seabed acoustic anomalies. Fluid migration towards the upper slope occurs mostly along strata in upslope direction and largely prevails over vertical focused migration. Fluids accumulate in the uppermost part of the slope just westward of the shelf break, where they are trapped beneath the prograding glacigenic sequence, which is not permeable enough for fluids to migrate through. Fluids are expelled on the shelf where the base of the glacigenic sequence outcrops. Some gas-charged fluids may originate from deep-seated hydrocarbon reservoirs and can be temporally stored in gas hydrates in the shallow subsurface. Though evidence for hydrates on the uppermost slope is missing the seismic data from the lower slope clearly suggest the occurrence of gas hydrate.

Lenition and Phonemic Overlap in Rome Italian

The purpose of this paper is to offer acoustic evidence for an unusual phonemic contrast in Rome Italian. In our corpus, about half of all tokens of intervocalic /p t k/ are realized with uninterrupted voicing (both word-internally and across word boundaries). Furthermore, the voiced realizations of /t/ and /k/ do not significantly differ from /d/ and /g/ in duration and/or degree of constriction (as acoustically determined). Phonemic contrast is maintained under substantial phonetic overlap. Regarding the labials, duration keeps /p/ and /b/ apart. Contrary to the universal tendency, it is /b/ that is considerably longer, due to complex diachronic facts.

Morphological and acoustic evidence for hybridisation between two broadly sympatric south-eastern Australian tree frogs Litoria ewingii and L. verreauxii (Anura: Hylidae)

The south-eastern Australian tree frogs Litoria ewingii and L. verreauxii are related species that share a broad zone of sympatry. Despite evidence of reproductive character displacement and positive assortative mating in sympatry, a small number of studies have provided inconclusive evidence that occasional hybridisation may occur. We assessed morphological and acoustic divergence between males from two localities where L. ewingii and L. verreauxii occur in sympatry. Several acoustic and morphological traits were found to be divergent between individuals assigned a priori to parental species based on a single acoustic character of the male advertisement call that was previously identified to be diagnostic (pulse number). In contrast, multivariate analysis indicated that at least five adult males identified a priori as putative hybrids (on the basis of intermediate pulse number) were also intermediate in morphology and other aspects of acoustic structure. The concordance of intermediate morphological and acoustic phenotypes supports the hypothesis that despite strong premating isolation, mismating results in at least a small number of hybrid adult males where these species occur in sympatry. This new evidence of hybridisation between L. ewingii and L. verreauxii underlines that there remains much to learn about this model system for understanding hybrid zone dynamics.