Voiceless Fricatives Research Articles

Speech cue enhancement for the hearing impaired: I. Altered vowel durations for perception of final fricative voicing.

Hearing-impaired persons with reduced voicing perception for final fricatives were studied for improvement in perception via training or enhancement of the vowel duration cue. Ten utterances each of the syllables /baef, baes, baev, baez/ composed three stimulus sets that were graded for the magnitude of the vowel duration cue. These sets were used differentially among 25 hearing-impaired listeners to train and test voicing perception. When enhanced, the vowels of the utterances were lengthened before voiced fricatives and shortened before voiceless fricatives. Training for the utterances with unmodified vowels yielded a small performance improvement for a subgroup of 14 listeners. In comparison, this subgroup showed considerably more improvement for the utterances with vowel duration enhancements. Another subgroup of 5 listeners displayed good fricative voicing perception independent of training or the vowel duration enhancements. A remaining subgroup of 6 listeners manifested poor use of the maximum unmodified vowel duration cues, even subsequent to training. This group showed a significant improvement in perception only for the voiced fricatives when extra duration enhancement and further training were used. This subgroup also demonstrated the poorest performance for a measure of vowel duration discrimination. Among the auditory variables examined, the listeners' tone thresholds at 250 Hz showed the highest relation to perception of fricative voicing.

The Segmental and Superasegmental Status of Preaspiration in Modern Icelandic

The paper discusses the historical origin, present day distribution and prosodic status of preaspiration in Icelandic. Having developed as a movement of the openness of the glottis, originally a feature of “hard” stops, towards the nucleus of the stressed syllable, its distribution must be accounted for by a “segmental” rule operating on historical (underlying) hard stops. Its status with respect to length and stress is, however, shown to be that of a prosody of the nucleus of the Modern Icelandic syllable, since it carries length when the syllable is stressed. The development of preaspiration is seen as a part of a more general rule also responsible for voiceless sonorants and fricatives in the nucleus.

Spectral envelopes and perceptual target zones for consonants and vowels: Preliminary estimates

Preliminary estimates of the sizes, shapes, and locations of perceptual target zones (PTZ) of ten vowels, six dipthongs, a twenty‐nine consonants have been made. The PTZ's are defined as enclosed volumes in the space x = log(PF3/PF2), y = log(PF1/PR), and z = log(PF2/PF1), where PR is the perceptual reference in Hertz and PF1, PF2, and PF3 are the center frequencies of the perceptual formants. The perceptual formants are estimated from spectral envelopes as guided by rules for distinguishing burst‐friction spectra and glottal‐source spectra and for naming and numbering formants based on their logarithmic relations to the value of the reference. The estimation of PTZ's will be illustrated for at least one allophone from each of the following classes: vowels, diphthongs, approximates, nasals, voiceless fricatives, and voiced fricatives. These preliminary estimates indicate that PTZ's: (1) are large, thus encompassing a considerable range of spectral variants as might be associated with coarticulatory effects and articulatory style; (2) have irregularly shaped, abutting boundaries; and (3) for the most part, exhibit little overlap. [NINCDS grant R0 I‐NS21994‐01.]

Fricative‐vowel coarticulatory effects in Arabic

Using LPC analysis, an investigation of anticipatory vowel coarticulation was made of 100‐ms segments of Arabic voiceless fricatives (/h, x, ∫, and s/) spoken in isolation and in three prevocalic contexts (/ɑ, i, and u/) by three native speakers. The degree of coarticulation for each prevocalic fricative segment was determined by assessing both the amount of spectral change from the isolated production, and the relationship of prevocalic fricative peaks to vowel formats. Strong coarticulation was observed for all speakers in all three vowel contexts for the pharyngeal fricative /h/, while minimal coarticulatory effects were present for the dental fricative /s/. Vowel specific trends were evident for the palatal /∫/ and velar /x/ fricatives: strong coarticulation was seen for /xi/, /xu/, /∫i/ and/ ∫ɑ/ and not the remaining utterances. Individual differences were evident with respect to degree of coarticulation. The three speakers behaved similarly for only three of the 12 fricative‐vowel utterances. Results are discussed in terms of the articulatory compatibility principle. Present address: Biosciences Lab., 3M Center, St. Paul, MN.

Speech coding in the auditory nerve: III. Voiceless fricative consonants.

Responses of auditory-nerve fibers in anesthetized cats were recorded for synthetic voiceless fricative consonants. The four stimuli (/x/, /s/, /s/, and /f/) were presented at two levels corresponding to speech in which the levels of the vowels would be approximately 60 and 75 dB SPL, respectively. Discharge patterns were characterized in terms of PST histograms and their power spectra. For both stimulus levels, frequency regions in which the stimuli had considerable energy corresponded well with characteristic-frequency (CF) regions in which average discharge rates were the highest. At the higher level, the profiles of discharge rate against CF were more distinctive for the stimulus onset than for the central portion. Power spectra of PST histograms had large response components near fiber characteristic frequencies for CFs up to 3-4 kHz, as well as low-frequency components for all fibers. The relative amplitudes of these components varied for the different stimuli. In general, the formant frequencies of the fricatives did not correspond with the largest response components, except for formants below about 3 kHz. Processing schemes based on fine time patterns of discharge that were effective for vowel stimuli generally failed to extract the formant frequencies of fricatives.

Phonetic influences on stuttering in monolingual and bilingual stutterers

The difficulty stutterers might have with individual sounds was investigated with respect to two modes of speaking (oral reading versus spontaneous speech) and two languages (English versus Kannada). Ten monolingual and ten bilingual stutterers read 16 lists of words (8 in each language). Analysis of stuttering was made with respect to a three-way classification of sounds (vowels, voiceless consonants, and voiced consonants) as well as an eight-way classification (short vowels, long vowels, voiceless stops, voiceless fricatives, voiced stops, voiced fricatives, nasals, and semivowels). Analysis was made with respect to both word-initial and total stuttering. The results indicated that both monolingual and bilingual stutterers were more dysfluent on voiceless consonants and especially on voiceless fricatives, when total stuttering was considered. Results of the analysis of word-initial stuttering on an eight-way classification of sounds showed that the bilingual stutterers stuttered more on the nasal sounds. The results of the bilingual comparison indicated the possibility that the phonetic influences on stuttering might be dependent on the number of languages spoken by the subjects as well as the specific language in which the effects were observed.

Airflow characteristics of fricative consonants produced by normally hearing and hearing-impaired speakers.

The purpose of this study was to investigate the rates of airflow for fricative consonants produced by normally hearing speakers and by hearing-impaired speakers who varied in degree of hearing loss and speech intelligibility. Subjects consisted of 20 normally hearing adults (10 men and 10 women), 20 moderately-to-severely hearing-impaired adults with intelligible speech (10 men and 10 women), and 20 profoundly hearing-impaired adults with semi-intelligible speech (10 men and 10 women). Average airflow rates were obtained for the fricatives /s/, /z/, /f/, and /v/ in CV and VCV syllables. In general, it was found that normally hearing, intelligible hearing-impaired, and some semi-intelligible hearing-impaired speakers produced voiceless fricatives with substantially greater airflow rates than they produced voiced cognates. The data also revealed that both male and female semi-intelligible hearing-impaired speakers produced fricatives in the CV context with greater amounts of airflow and fricatives in the VCV environment with less airflow than did the normally hearing and intelligible hearing-impaired speakers. Further, the male and female semi-intelligible hearing-impaired speakers failed to demonstrate the trend, as exhibited by the other speakers, for lingua-alveolar fricatives to be produced with greater air flow than the labiodental fricatives. The data from this investigation provide additional evidence to support the assumption that some hearing-impaired speakers fail to control and coordinate the biomechanical aspects of speech properly.

Acoustic and articulatory asymmetries in fricative-vowel and vowel-fricative productions

When synthetic noise replaces the natural friction in FV syllables and VF syllables, the influence of the vocalic segment on the perception of the noise as /s/ or /∫/ is greater for the FV syllables [S. D. Soli and V. A. Mann, J. Acoust. Soc. Am. Suppl. 1 71, S75 (1982)]. The acoustic and articulatory bases of this perceptual asymmetry were examined. Mean formant tracks displaying confidence intervals and amplitudes were computed for natural FV and VF syllables, where F = /∫/ and /s/, and V = /ɑ/ and /u/. FV and VF transitions were equally stable, although FV transitions exhibited larger coarticulatory differences due to fricative context. FV transitions also occurred during the highest amplitude portion of the vocalic segment, while VF transitions occurred as voicing amplitude was decreasing rapidly before the upcoming voiceless fricative. These acoustic differences would lead to larger vowel context effects for FV syllables. Thus the perceptual asymmetries appear to be a consequence of differences in the organization and timing of the supralaryngeal and laryngeal events which produce initial and final voiceless fricatives. [Supported by NICHD.]

Analysis and recognition of Japanese voiceless fricative consonants.

Analysis and recognition of Japanese voiceless fricative consonants.

母音および摩擦音発生時の発声器官の運動推定

It is of great interest to study control laws which produce speech sounds using a large number of muscles to move articulatory organs in precise order. There are many kinds of works on this problem in the linguistic, medical and engineering fields. However, our knowledge is still so small and restricted that we need many years to accumulate data and to organize them in a satisfactory manner. The estimation of the articulatory motion from speech waves is one of the important techniques to be developed in the research of speech production. In this paper, two problems are discussed in the analyses of articulatory movements which use speech production model. First, an adaptation scheme adjusting the articulatory model for each speaker is presented, which is necessary to improve the accuracy of the estimation. The estimated results are compared with the data observed by an X-ray microbeam system, and it is shown that a simple method which uses a linear transform of the base vector to modify the model using a set of speech sounds for 5 vowels, is very effective for the improvement of the estimation accuracy. Second, an extension of the estimation for consonants is presented. For voiceless fricative consonants, a noise source exists in the middle of the vocal tract and the acoustic characteristics of these consonants are mainly determined by the vocal tract configuration between the constriction made by the tongue and the lips. We tried to determine the vocal tract shape of these consonants by the spectral matching between real speech and the synthesized one from the model. Four parameters are used to determine the area function of the vocal tract. Estimation of the area function is tested through the simulation. Studying the distribution of the spectral error for various area functions, it is confirmed that the stable convergence is always assured. And further, the application of the method to real speech waves shows hopeful consequences.

Acoustic perceptual study of the Portuguese voiceless fricatives

A perceptual approach to the definition of the information bearing relevant acoustic features of the Portuguese voiceless fricatives /f/, /s/, /∫/ was tried. The experiment was conducted in order to (1) identify these acoustic features and (2) induce classification errors by means of a suppression of the obtained features using notch filtering. Although this method tended to attenuate major spectral peaks drastically few classification errors occured. Instead the experiment can be taken as a demonstration of an unexpected ability of the perceptual system to compensate for static filtering effects.

Laryngeal adjustments in the production of consonant clusters and geminates in American English

The glottal opening gesture and its timing control in various sequences of voiceless obstruents were investigated by the combined techniques of electromyography, photo-electric glottography, and fiberoptic endoscopy. The results obtained at both electromyographic and movement levels revealed that the glottal opening gesture is characterized by a one-, two-, or more-than-two-peaked pattern in a regular fashion according to the phonetic nature of the voiceless segments: each voiceless obstruent or geminate accompanied by aspiration or frication noise tends to require a single separate peak of the opening gesture, while an unaspirated stop embedded in a voiceless environment can be produced within the opening gesture attributed to an adjacent aspirated stop or fricative. Such an independent opening gesture of the glottis for the production of voiceless aspirated stops or voiceless fricatives even in sequentially unvoiced contexts can be interpreted as assuring the aerodynamic requirements for turbulent noise production during the aspirated stop or fricative segment.

Acoustic Factors that May Contribute to Categorical Perception

The perception of the voiced and voiceless velar and pharyngeal fricatives / , x, , h/ and of /∫, s/ in Colloquial Egyptian Arabic was examined to determine if the presence of the first two or three formants in / , x, , h/ results in continuous perception, in contrast to an expected categorical perception of /∫, s/, which lack these formants. Three twelve-step series of VFV nonsense words were synthesized. For the /∫/-/s/ series, the center of a band of high-frequency noise was varied in equal steps. For the /x/-/h/ and / /-/ / series, the first formant was varied. Eight native speakers were asked to identify the stimuli and discriminate two-step differences in a 4IAX discrimination task. While the voiced / /-/ / series showed continuous or less categorical perception than the /∫/-/s/ series, the voiceless /x/-/h/ series was perceived somewhat categorically. This suggests that voicing alone, or in combination with acoustic information about the lower formants, may be a necessary condition for continuous perception.

Auditory discrimination of voiceless fricatives in children.

Auditory discrimination of the voiceless fricatives magnitude of theta f integral of s was studied in 200 subjects between the ages of four and six and a half years. In the test task the subject was asked to remember one of the sounds and then to indicate each time this sound was presented. Two types of errors were analyzed. An error of omission was a failure to identify the remembered sound, and this type of error did not change with age. An error of commission, a failure to discriminate between the remembered sound and one of the other voiceless fricatives, decreased with age. The results are interpreted as support for a theory of the development of auditory perception of speech in which discrimination of some properties in speech, such as relatively weak spectral cues and second formant transitions, have to be learned by a child.

Spontaneous nasalization

Distinctively nasal vowels usually develop from earlier sequences of vowel + nasal consonant. There are exceptions, however, called instances of “spontaneous nasalization,” where an earlier nasal is lacking, e.g., Hindi [sɑ̃:p] “snake” from Sanskrit sarpa; English [hɑ̃vəd] “Harvard.” Curiously, most such spontaneously nasalized vowels have obstruents with high airflow next to them, e.g., voiceless fricatives. We tested a hypothesis that vowels flanking such consonants might sound nasalized without being nasilized if produced with a slightly open glottis via assimilation to the greater-than-normal glottal opening required by the high airflow segments. The spectral effects of open glottis duplicate those of nasilization: increased formant bandwidth, etc. Physiological recordings of syllables such as [bis], [bɑs], by two speakers each of American English and Mexican Spanish showed no true nasalization on the vowels but did show appreciable glottal opening just before the [s]. Using digital techniques, the last complete period of these vowels before the [s] was iterated into a steady-state vowel 400 ms long and presented to listeners who, as predicted, judged them to sound significantly more “nasalized” than control stimuli. [Supported by NSF.]

The role of formant transitions in the identification of Spanish fricatives

The role played by formant transitions in the identification of Argentine Spanish voiceless fricatives: [f, s, ∫] was evaluated by cross‐splicing technique of natural syllables. The test tape consisted of a set of CV syllables which were formed by the recombination of each friction portion with each vocalic portion. The vocalic portion was one of the Spanish vowels [i, a, u]. The average responses of ten listeners to these stimuli showed that the acoustic cues provided by the friction portion were sufficient for the identification of the fricative sounds. In the case of [f], the results are not in agreement with previous data obtained for English listeners [K. Harris, Lang. Speech 1(1), 1–7 (1958)], a discrepancy which may be explained by the different linguistic systems. Formant transitions seem to play some role in s/∫ distinction. But its relevant role seems to be preventing a vowel effect on the preceding fricative.

Laryngeal-oral coordination in American English obstruent production

Voiceless obstruent production involves simultaneous activity at both laryngeal and supralaryngeal levels. The laryngeal and oral articulations have to be temporally coordinated, and the present experiment was designed to examine the nature of this coordination. Laryngeal activity was monitored by transillumination, and an artificial palate was used to obtain information on tongue-palate contact patterns. Voiceless obstruents in various positions and under different stress conditions spoken at two different rates provided the linguistic material. Preliminary results indicate a fixed relation between onset of glottal adduction and oral release in aspirated stops. In voiceless fricatives, peak glottal opening occurs close to the middle of the oral constriction. The results also suggest that temporal ratios between oral and laryngeal gestures tend to remain constant across changes in stress and speaking rate. [Work supported by NINCDS and BRSG.]

Discriminant analysis of the acoustics of seven English voiceless fricatives

The analysis of fricatives has typically been either acoustic (by spectroscopy and analysis‐by‐synthesis) or perceptual (by multi‐dimensional scaling of responses to natural stimuli and bivariate analysis of responses to specially prepared stimuli). While these contribute to our knowledge, there are too many possible acoustic cues within the speech signal to be accounted for in an economical way. Further, of the English voiceless fricatives, only [f θ s ∫] have been studied extensively; the three fricative‐like sounds [h ʍç], occurring at a systematic phonetic level in many English dialects, have been largely neglected. The present study used multiple discriminant analysis to find the most effective acoustic cues for discriminating and characterizing phonologically all seven English voiceless fricatives. Acoustic information was extracted digitally using several techniques whose relative effectiveness was then compared. The most successful measures were the quartiles of spectral energy distribution, expressed in mels; VOT; and the peak amplitude of the fricative relative to the peak of the neighboring vowels. Three discriminant functions were found to account for 98.5% of the variance; posterior classification was successful in 90% of the samples.

Laryngeal adjustments in Japanese voiceless sound production

As part of a series of investigations on the production of sequences of unvoiced sounds in different languages, the current experiment was conducted using the combined techniques of photoelectric glottography, fiberoptics, and electromyography. Particular attention was paid to devoiced vowel production in various voiceless consonantal environments including geminates. The movement data show that the glottal opening gesture during a voiceless sequence containing a devoiced vowel is characterized by a monomodal pattern, unless the vowel occurs between voiceless fricatives. The results also suggest that the velocity and size of the glottal opening gesture vary according to the nature of the adjacent voiceless obstruents. The speed of the opening phase is slow when a stop precedes the vowel, and fast when a fricative precedes it. The peak glottal opening during the devoiced vowel is larger when a fricative either precedes or follows it, and smaller when the vowel is surrounded on both sides by single or geminated stops. [Work supported by NINCDS and BRSG.]

A dynamic processing approach to phoneme recognition (part I)--Feature extraction

This paper presents a new approach to phonemic feature extraction, which is one of the essential components of the study to develop an automatic speech recognition system. The feature extraction method is composed of three levels, i.e., frequency analysis, peak-band enhancement, and local peak extraction. The latter two levels consist of a multistage processing structure and are made by using analogical formulations to mechanical dynamics. Extracted features include local peak frequencies and their amplitudes, and local valley frequencies and their amplitudes, which are estimated by the computational process with no feedback loop and no backtracking. The feasibility is first confirmed by using synthesized speech, then by application to continuous vowels, multisyllable words, voiceless stop consonants, and voiceless fricatives. The advantages are discussed in comparisons with other methods.