Syllable Rate Research Articles

Performance constraints and the production of birdsong

The role of physical and physiological constraints in determining the performance limits on the tempo and frequency bandwidth of birdsong was investigated. One series of experiments examined the mechanism by which a vocal mimic, the northern mockingbird (Mimus polygottos), copied the songs of other species with which it was tutored as a juvenile. Other experiments analyzed the motor basis of special canary (Serinus canaria) syllables eliciting sexual responses from females. In each case, the mechanism of vocalization was determined by measuring the respiratory dynamics and sound produced on each side of the songbirds duplex vocal organ, the syrinx. When mockingbirds copied the songs of other species the accuracy of their copy depended on the accuracy with which they reproduced the motor pattern used by the tutor species. Motor difficulty of various acoustic features was assessed by the accuracy of its copy. The high repetition rate, broadband canary syllables preferred by females required especially demanding bilateral motor skills. The results indicate that constraints on the rate of respiratory ventilation and bilateral syringeal coordination can set an upper limit on syllable repetition rate and frequency bandwidth. [Work supported by NIH and NSF.]

Alternating motion rate as an index of speech motor disorder in traumatic brain injury

The task of syllable alternating motion rate (AMR) (also called diadochokinesis) is suitable for examining speech disorders of varying degrees of severity and in individuals with varying levels of linguistic and cognitive ability. However, very limited information on this task has been published for subjects with traumatic brain injury (TBI). This study is a quantitative and qualitative acoustic analysis of AMR in seven subjects with TBI. The primary goal was to use acoustic analyses to assess speech motor control disturbances for the group as a whole and for individual patients. Quantitative analyses included measures of syllable rate, syllable and intersyllable gap durations, energy maxima, and voice onset time (VOT). Qualitative analyses included classification of features evident in spectrograms and waveforms to provide a more detailed description. The TBI group had (1) a slowed syllable rate due mostly to lengthened syllables and, to a lesser degree, lengthened intersyllable gaps, (2) highly correlated syllable rates between AMR and conversation, (3) temporal and energy maxima irregularities within repetition sequences, (4) normal median VOT values but with large variation, and (5) a number of speech production abnormalities revealed by qualitative analysis, including explosive speech quality, breathy voice quality, phonatory instability, multiple or missing stop bursts, continuous voicing, and spirantization. The relationships between these findings and TBI speakers' neurological status and dysarthria types are also discussed. It was concluded that acoustic analyses of the AMR task provides specific information on motor speech limitations in individuals with TBI.

Towards robustness to speech rate in mandarin all-syllable recognition

In mandarin all-syllable recognition, many insert errors occur due to the influence of non-consonant syllables. Introducing the duration model into the recognition process is a direct way to lessen these errors. But that usually could not work well as expected, for the duration is sensitive to speech rate. Hence, aiming at this problem, a novel context dependent duration distribution normalized by speech rate is proposed in this paper and applied to a speech recognition system based on the frame of improved Hidden Markov Model (HMM). To realize this algorithm, the authors employ a new method to estimate the speech rate of a sentence; then compute the duration probability combined with speech rate; and finally implement this duration information in the post-processing stage. With little change in the recognition process and resource demand, the duration model is adopted efficiently in the system. The experimental results indicate that the syllable error rates decrease significantly in two different speech corpora. Especially for the insertions, the error rates reduce about sixty to eighty percent.

Temporal pattern recognition based on instantaneous spike rate coding in a simple auditory system.

Auditory pattern recognition by the CNS is a fundamental process in acoustic communication. Because crickets communicate with stereotyped patterns of constant frequency syllables, they are established models to investigate the neuronal mechanisms of auditory pattern recognition. Here we provide evidence that for the neural processing of amplitude-modulated sounds, the instantaneous spike rate rather than the time-averaged neural activity is the appropriate coding principle by comparing both coding parameters in a thoracic interneuron (Omega neuron ON1) of the cricket (Gryllus bimaculatus) auditory system. When stimulated with different temporal sound patterns, the analysis of the instantaneous spike rate demonstrates that the neuron acts as a low-pass filter for syllable patterns. The instantaneous spike rate is low at high syllable rates, but prominent peaks in the instantaneous spike rate are generated as the syllable rate resembles that of the species-specific pattern. The occurrence and repetition rate of these peaks in the neuronal discharge are sufficient to explain temporal filtering in the cricket auditory pathway as they closely match the tuning of phonotactic behavior to different sound patterns. Thus temporal filtering or "pattern recognition" occurs at an early stage in the auditory pathway.

Effects of speaking rate on EMA‐derived lingual kinematics: a preliminary investigation

Electromagnetic articulography (EMA) was used to investigate how tongue movement characteristics (i.e., velocity, acceleration, duration, distance) change with, or indeed affect, increased rates of speech. Eight young adult males repeated /ta/ and /ka/ syllables first at a moderate rate that had been modelled at three syllables per second, and then ‘as fast as possible’. Distance travelled by the tongue appeared to be the principal lingual kinematic feature manipulated by the group of speakers in producing increased syllable repetition rates, with velocity found to increase, decrease or remain unchanged. Acceleration remained unchanged, except in the case of increased velocity. One participant formed an exception in terms of manipulating distance by exhibiting marginally increased lingual velocities rather than distance changes. This preliminary study serves to direct future EMA‐based studies of speech rate control as to the speech tasks that should be employed and the possible underlying anatomical and acoustic bases or constraints that could possibly influence the kinematic strategies employed to increase speech rate.

Modeling partial pronunciation variations for spontaneous Mandarin speech recognition

The high error rate in spontaneous speech recognition is due in part to the poor modeling of pronunciation variations. An analysis of acoustic data reveals that pronunciation variations include both complete changes and partial changes. Complete changes are the replacement of a canonical phoneme by another alternative phone, such as ‘b’ being pronounced as ‘p’. Partial changes are the variations within the phoneme, such as nasalization, centralization, voiceless, voiced, etc. Most current work in pronunciation modeling attempts to represent pronunciation variations either by alternative phonetic representations or by the concatenation of subphone units at the hidden Markov state level. In this paper, we show that partial changes are a lot less clear-cut than previously assumed and cannot be modeled by mere representation by alternate phones or a concatenation of phone units. We propose modeling partial changes through acoustic model reconstruction. We first propose a partial change phone model (PCPM) to differentiate pronunciation variations. In order to improve the model resolution without increasing the parameter size too much, PCPM is used as a hidden model and merged into the pre-trained acoustic model through model reconstruction. To avoid model confusion, auxiliary decision trees are established for PCPM triphones, and one auxiliary decision tree can only be used by one standard decision tree. The acoustic model reconstruction on triphones is equivalent to decision tree merging. The effectiveness of this approach is evaluated on the 1997 Hub4NE Mandarin Broadcast News corpus (1997 MBN) with different styles of speech. It gives a significant 2.39% syllable error rate absolute reduction in spontaneous speech.

Directional female preference for an exaggerated male trait in canary (Serinus canaria) song.

Motor constraints on vocal production impose a trade-off between trill rate and frequency bandwidth within birdsong. We tested whether domesticated canary (Serinus canaria) females, reared either in acoustic isolation or in aviary conditions, had a preference for broad bandwidth songs with artificially increased syllable rates. The copulation solicitation display (CSD) was used as an index of female preference. As predicted, both naive and experienced females were especially responsive to syllables with a broad bandwidth emitted at an artificially increased rate. Female preference for supernormal stimuli provide support for the honest-signalling hypothesis and our results are consistent with recent findings indicating that production of song phrases maximizing both bandwidth and syllable rate may be a reliable indicator of male physical or behavioural qualities. We suggest that female preference for vocal emissions, which simultaneously maximize these two parameters, could be a widespread pattern within songbirds.

Primary closure versus radial forearm flap reconstruction after hemiglossectomy: functional assessment of swallowing and speech.

The authors compared the postoperative speech and swallowing function of six patients who underwent free radial forearm flap reconstruction after hemiglossectomy with that of six control patients who underwent primary closure of the defect. Clinical speech pathologic evaluations included the Fletcher time-to-time maximum repetition rate of syllables, multiple rhyme test, and overall quality and intelligibility of the patients' speech. Evaluation of swallowing included the duration of deglutition, bolus volume, and ingestion rate. Speech quality, including intelligibility and articulation, was better in patients with primary closure. However, the bolus volume and ingestion rate in deglutition were better in those with flap reconstruction. These results suggest that the flap adds bulk, thus improving pharyngeal clearance by maintaining the tongue-to-mouth roof contact that is necessary in the swallowing process. The nonfunctional flap, however, hinders articulation by restricting the mobility of the remaining portion of the normal tongue.

Temporal cues in the antiphonal long-calling behaviour of cottontop tamarins

In primates, long-range communication is often mediated by the use of ‘long’ (or ‘loud’) calls. Beyond the acoustic classification of these calls and descriptions of the behavioural context in which they are produced, few experimental studies have examined how species-typical information is encoded in the structure of these signals. We present the results of eight experimental conditions designed to isolate the mechanisms underlying the perception of long calls in cottontop tamarins, Saguinus oedipus. Our procedure involved presenting a combination of naturally produced and experimentally manipulated long calls to individuals isolated from their group, and then recording the relationship between signal design and the production of antiphonal long calls by the test subject. Tamarins did not distinguish between normal calls and time-reversed or pitch-shifted long calls, but normal response rates did require the species-typical amplitude envelope. Furthermore, there was suggestive evidence that the number of syllables and the syllable rate may also influence antiphonal calling responses. We discuss these results in terms of the mechanisms of vocal recognition in primates and other taxa. Copyright 2002 The Association for the Study of Animal Behaviour. Published by Elsevier Science Ltd. All rights reserved.

Kinematics of normal lingual diadokokinesis

Speech-language clinicians use diadochokinetic (DDK) tasks as a behavioral measure of the status of the speech production system. The articulator kinematics of these repetitive syllable productions has been relatively little studied (most studies of speech DDK have been acoustic). As a result, many clinicians misunderstand the relationship between syllable rate and movement parameters, such as rate of articulator movement and range of articulator movement. For example, clinicians assume that because ‘‘kuh’’ syllable repetitions are relatively slow, tongue dorsum movements must be slow. However, Westbury and Dembowski [Ann. Bull. RILP No. 27 (1993)] showed that the tongue dorsum may produce larger and faster movements than other tongue points (such as the tongue tip). Therefore, syllable repetition rates may not reflect movement speeds for individual articulator points. This study replicates and extends these findings, using a larger speaker sample encompassing a wider age range. The goal of this study is to further clarify the relationship between articulator speed, range of articulator motion, and syllable rate. That is, it seeks to examine whether syllable rate is related to how quickly the tongue moves or how far the tongue moves. Data are derived from the University of Wisconsin X-ray Microbeam Database.

Characteristics of diadochokinesis in Parkinson’s Disease and Multiple Sclerosis

The current study applies a quantitative, acoustic analysis procedure for the study of rapid syllable productions outlined by Kent and colleagues [R. D. Kent et al., J. Med. Sp. Lang. Path. 7, 83–90 (1999)] to syllables produced by speakers with Parkinson’s Disease and Multiple Sclerosis. Neurologically healthy talkers will be studied for comparison purposes. Acoustic measures will be reported for syllable repetitions of /pə/, /tə/, and /kə/. Temporal measures will include syllable duration, syllable rate, and stop gap duration. The energy envelope of syllable repetitions will be quantified using measures of rms amplitude minima and maxima. Acoustic measures will be contrasted to determine the extent to which acoustic profiles of diadochokinesis distinguish hypokinetic dysarthria associated with Parkinson’s Disease, ataxic dysarthria secondary to Multiple Sclerosis, and spastic dysarthria secondary to Multiple Sclerosis. It also is of interest to determine whether speakers with Parkinson’s Disease and Multiple Sclerosis judged to be nondysarthric via perceptual analyses also demonstrate objective, acoustic profiles of diadochokinesis that are within normal limits. [Work supported by NIH.]

Mandarin pronunciation modeling based on CASS corpus

The pronunciation variability is an important issue that must be faced with when developing practical automatic spontaneous speech recognition systems. In this paper, the factors that may affect the recognition performance are analyzed, including those specific to the Chinese language. By studying the INITIAL/FINAL (IF) characteristics of Chinese language and developing the Bayesian equation, the concepts of generalized INITIAL/FINAL (GIF) and generalized syllable (GS), the GIF modeling and the IF-GIF modeling, as well as the context-dependent pronunciation weighting, are proposed based on a well phonetically transcribed seed database. By using these methods, the Chinese syllable error rate (SER) is reduced by 6.3% and 4.2% compared with the GIF modeling and IF modeling respectively when the language model, such as syllable or word N-gram, is not used. The effectiveness of these methods is also proved when more data without the phonetic transcription are used to refine the acoustic model using the proposed iterative forced-alignment based transcribing (IFABT) method, achieving a 5.7% SER reduction.

Differences of speech rate, interphoneme distance and likelihood caused by speaking style, their relationship, and recognition performance

AbstractWe have been developing a speech recognition system as part of a spoken dialogue system. There is a clear limit of recognition performance for dialogue speech using acoustic models built with a read speech database, mainly because various acoustic and linguistic phenomena, which reflect the characteristics of spontaneous speech, occur in dialogue speech. We made acoustic models of syllable unit using simulated dialogue speech or read speech database. We then analyzed the acoustic properties of the dialogue speech in terms of the distance between models and the distribution of duration and compared them with read speech. We found that the speech rate of dialogue speech was faster on average and had a greater variance of speech rate than read speech. We also found that the distance between vowels of the dialogue speech tended to be smaller than that of read speech, that is, the vowels tend to be confused in dialogue speech because of closer boundaries to one another. We also compared recognition performance using acoustic models trained with each database and found that dialogue speech had a low correct syllable rate. Dialogue speech was more prone to insertion and deletion errors and the influence of distance between vowels and the variance of speech rate was shown clearly in the results. © 2002 Wiley Periodicals, Inc. Syst Comp Jpn, 33(7): 50–60, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/scj.1141

Anomalous anatomy of speech–language areas in adults with persistent developmental stuttering

To the Editor: I read with interest the article by Foundas et al.,1 which is the latest in a line of neurologic reports that confuse the cause and effect of stuttering (stammering). The observations associated with this clinical problem of chronic involuntary blockage are not disputed. That these observations can be interpreted as a cause of blockage is neurologically impossible. Because unintended blockage of speech is an aberration of normal speech processing, which is a manifestation of synchronization of phonation (for speech power) with articulation (that shapes phonatory power into the sounds of speech), the essence of fluent speech is the timing of speech sound phonatory/articulatory synchronization. Any cause of blockage has to account for why it can be involuntary (otherwise it would never become chronic), and, more to the point, how high-speed speech sounds can be produced with low-speed, cognitive/linguistic equipment. Synchronization speed is revealed by the syllable rate (no sound can be produced outside the context of a syllable) multiplied by the number of sounds per syllable. Syllable and cognitive thinking rates turn out to be roughly the same, which is fortunate, otherwise we would be unable to speak at the same rates we think. Decades ago, George Miller demonstrated the cognitive rate to be 7 ± 2 per second.2 In 1964, I inadvertently discovered that this is also the syllable rate, when …

Temperature dependence of cicada songs (Homoptera, Cicadoidea).

The songs of male Portuguese cicadas Tettigetta argentata, T. josei and Tympanistalna gastrica were recorded at five to seven temperatures within the range 24-38.5 degrees C. To investigate the temperature dependence of the neuromuscular apparatus involved in song production, different temporal elements of the calling songs were measured. We report a strong temperature dependence for the syllable and the echeme rates in T. josei and Ty. gastrica. This suggests that in these species the neuromuscular structures involved in the timbal cycle and in generating the echeme succession of the song are strongly temperature dependent. In T. argentata, the syllable rate was again significantly temperature dependent; the echeme rate, however, increased between 25.5 degrees C and 33.5 degrees C but decreased with the highest temperature. This indicates that at least in T. argentata two separate neuronal networks control both song parameters. Other temporal elements of the song with potential behavioural significance were also measured and found to be temperature dependent (e.g. echeme duration and interval). The possible implications for intraspecific communication are discussed. We also demonstrate that the temperature of these small cicadas is not significantly influenced by the muscle activity involved in song production. On the other hand, exposure to sunlight can be used by these cicadas to elevate their body temperature by more than 10 degrees C.

Comparison of different implementations of MFCC

The performance of the Mel-Frequency Cepstrum Coefficients (MFCC) may be affected by (1) the number of filters, (2) the shape of filters, (3) the way in which filters are spaced, and (4) the way in which the power spectrum is warped. In this paper, several comparison experiments are done to find a best implementation. The traditional MFCC calculation excludes the 0th coefficient for the reason that it is regarded as somewhat unreliable. According to the analysis and experiments, the authors find that it can be regarded as the generalized frequency band energy (FBE) and is hence useful, which results in the FBE-MFCC. The authors also propose a better analysis, namely the auto-regressive analysis, on the frame energy, which outperform its 1st and/or 2nd order differential derivatives. Experiments with the “863” Speech Database show that, compared with the traditional MFCC with its corresponding auto-regressive analysis coefficients, the FBE-MFCC and the frame energy with their corresponding auto-regressive analysis coefficients form the best combination, reducing the Chinese syllable error rate (CSER) by about 10%, while the FBE-MFCC with the corresponding auto-regressive analysis coefficients reduces CSER by 2.5%. Comparison experiments are also done with a quite casual Chinese speech database, named Chinese Annotated Spontaneous Speech (CASS) corpus. The FBE-MFCC can reduce the error rate by about 2.9% on an average.

Backward-masking and syllable–rate thresholds have only minor associations with reading achievement, compared to several cognitive and linguistic measures

Rate thresholds for syllable–sequences [the Indiana Test of Auditory Memory and Processing Rate—ITAMPR, Watson and Eddins, J. Acoust. Soc. Am. 105, 1236 (1999)] and backward-masking thresholds, were obtained from 220 children. These measures were only weakly associated with reading achievement and with speech recognition. Other tests, however, accounted for roughly half of the variance in teacher-assigned reading grades. Results of 36 sensory, perceptual processing, linguistic, and cognitive tests and subtests administered to the population from which these children were obtained (approximately 98% of the 472 entering first-graders in Benton County Indiana from 1995 to 1997), were well-described by four factors. These factors and the systematic variance in reading achievement associated with them were: Factor 1 (31%): reading-related skills (phonological awareness, letter- and word-identification); Factor 2 (12%): visual cognition (visual perceptual abilities, spatial perception, visual memory); Factor 3 (9%): verbal cognition (language development, vocabulary, verbal concepts); and Factor 4 (1%): speech processing (the ability to understand speech, measured under difficult listening conditions). These data are inconsistent with the hypothesis that temporal processing deficits are causally related to reading impairments. [Work supported by NIH/NIDCD, Indiana Univ., and the Benton County (Indiana) School Corporation.]

Differential Contributions of Motor Cortex, Basal Ganglia, and Cerebellum to Speech Motor Control: Effects of Syllable Repetition Rate Evaluated by fMRI

In order to delineate the neuroanatomical correlates of speech motor control, functional magnetic resonance imaging was performed during silent repetitions of the syllable “ta” at three different rates (2.5, 4.0, and 5.5 Hz). Spatial extent and magnitude of hemodynamic responses at the level of the motor cortex showed a positive correlation to production frequencies. As concerns the basal ganglia, the lower rates (2.5 and 4.0 Hz) gave rise to higher magnitudes of activation within the left putamen as compared to the 5.5 Hz condition. In contrast, cerebellar responses were rather restricted to fast performance (4.0 and 5.5 Hz) and exhibited a shift in caudal direction during 5.5 as compared to 4.0 Hz. These findings corroborate the suggestion of a differential impact of various cortical and subcortical areas on speech motor control.

Seasonal Changes in the Song Pattern of the Non-Domesticated Island Canary (Serinus Canaria) a Field Study

[The domesticated canary ( Serinus canaria ) is one of the most widely used songbird species to study (1) neural mechanisms of behavioural plasticity and (2) mechanisms of song based female mate choice. Despite numerous studies of the singing and seasonal changes in the song of these animals under laboratory conditions, the present paper describes the results of the first systematic study of the song of the non-domesticated, free-living relatives of the domesticated canary, the island canary ( Serinus canaria ). The songs of ringed males of the canary population Ilheu Chao (Madeira) were studied at different times of the year including the spring breeding and autumnal non-breeding season. In the breeding season songs are on average longer and the repetition rate of syllables is higher compared to the non-breeding season. The syllable repertoire size does not change seasonally. Longitudinal studies of individual males confirm these results of the population average. Further, this approach showed that the repertoire composition of individual males changes seasonally with a significant increase of fast-frequency modulated syllables and a decrease of whistle-type syllables during the breeding season. Playback experiments showed that the fastfrequency modulated syllables of the male island canaries are sexually attractive, if used in standard courtship solicitation tests with female canaries under laboratory conditions. This suggests that seasonal changes in the song temporal pattern are a general feature of canaries, domesticated or not while seasonal changes in repertoire composition is an adaptive feature of island canaries, most likely lost during domestication., The domesticated canary ( Serinus canaria ) is one of the most widely used songbird species to study (1) neural mechanisms of behavioural plasticity and (2) mechanisms of song based female mate choice. Despite numerous studies of the singing and seasonal changes in the song of these animals under laboratory conditions, the present paper describes the results of the first systematic study of the song of the non-domesticated, free-living relatives of the domesticated canary, the island canary ( Serinus canaria ). The songs of ringed males of the canary population Ilheu Chao (Madeira) were studied at different times of the year including the spring breeding and autumnal non-breeding season. In the breeding season songs are on average longer and the repetition rate of syllables is higher compared to the non-breeding season. The syllable repertoire size does not change seasonally. Longitudinal studies of individual males confirm these results of the population average. Further, this approach showed that the repertoire composition of individual males changes seasonally with a significant increase of fast-frequency modulated syllables and a decrease of whistle-type syllables during the breeding season. Playback experiments showed that the fastfrequency modulated syllables of the male island canaries are sexually attractive, if used in standard courtship solicitation tests with female canaries under laboratory conditions. This suggests that seasonal changes in the song temporal pattern are a general feature of canaries, domesticated or not while seasonal changes in repertoire composition is an adaptive feature of island canaries, most likely lost during domestication.]

Brain correlates of stuttering and syllable production. A PET performance-correlation analysis.

To distinguish the neural systems of normal speech from those of stuttering, PET images of brain blood flow were probed (correlated voxel-wise) with per-trial speech-behaviour scores obtained during PET imaging. Two cohorts were studied: 10 right-handed men who stuttered and 10 right-handed, age- and sex-matched non-stuttering controls. Ninety PET blood flow images were obtained in each cohort (nine per subject as three trials of each of three conditions) from which r-value statistical parametric images (SPI¿r¿) were computed. Brain correlates of stutter rate and syllable rate showed striking differences in both laterality and sign (i.e. positive or negative correlations). Stutter-rate correlates, both positive and negative, were strongly lateralized to the right cerebral and left cerebellar hemispheres. Syllable correlates in both cohorts were bilateral, with a bias towards the left cerebral and right cerebellar hemispheres, in keeping with the left-cerebral dominance for language and motor skills typical of right-handed subjects. For both stutters and syllables, the brain regions that were correlated positively were those of speech production: the mouth representation in the primary motor cortex; the supplementary motor area; the inferior lateral premotor cortex (Broca's area); the anterior insula; and the cerebellum. The principal difference between syllable-rate and stutter-rate positive correlates was hemispheric laterality. A notable exception to this rule was that cerebellar positive correlates for syllable rate were far more extensive in the stuttering cohort than in the control cohort, which suggests a specific role for the cerebellum in enabling fluent utterances in persons who stutter. Stutters were negatively correlated with right-cerebral regions (superior and middle temporal gyrus) associated with auditory perception and processing, regions which were positively correlated with syllables in both the stuttering and control cohorts. These findings support long-held theories that the brain correlates of stuttering are the speech-motor regions of the non-dominant (right) cerebral hemisphere, and extend this theory to include the non-dominant (left) cerebellar hemisphere. The present findings also indicate a specific role of the cerebellum in the fluent utterances of persons who stutter. Support is also offered for theories that implicate auditory processing problems in stuttering.