Filtering by
- All Subjects: Speech
- Creators: Daliri, Ayoub
- Creators: Liss, Julie M
- Status: Published
The purpose of this longitudinal study was to predict /r/ acquisition using acoustic signal processing. 19 children, aged 5-7 with inaccurate /r/, were followed until they turned 8 or acquired /r/, whichever came first. Acoustic and descriptive data from 14 participants were analyzed. The remaining 5 children continued to be followed. The study analyzed differences in spectral energy at the baseline acoustic signals of participants who eventually acquired /r/ compared to that of those who did not acquire /r/. Results indicated significant differences between groups in the baseline signals for vocalic and postvocalic /r/, suggesting that the acquisition of certain allophones may be predictable. Participants’ articulatory changes made during the progression of acquisition were also analyzed spectrally. A retrospective analysis described the pattern in which /r/ allophones were acquired, proposing that vocalic /r/ and the postvocalic variant of consonantal /r/ may be acquired prior to prevocalic /r/, and /r/ followed by low vowels may be acquired before /r/ followed by high vowels, although individual variations exist.
The distinctions between the neural resources supporting speech and music comprehension have long been studied using contexts like aphasia and amusia, and neuroimaging in control subjects. While many models have emerged to describe the different networks uniquely recruited in response to speech and music stimuli, there are still many questions, especially regarding left-hemispheric strokes that disrupt typical speech-processing brain networks, and how musical training might affect the brain networks recruited for speech after a stroke. Thus, our study aims to explore some questions related to the above topics. We collected task-based functional MRI data from 12 subjects who previously experienced a left-hemispheric stroke. Subjects listened to blocks of spoken sentences and novel piano melodies during scanning to examine the differences in brain activations in response to speech and music. We hypothesized that speech stimuli would activate right frontal regions, and music stimuli would activate the right superior temporal regions more than speech (both findings not seen in previous studies of control subjects), as a result of functional changes in the brain, following the left-hemispheric stroke and particularly the loss of functionality in the left temporal lobe. We also hypothesized that the music stimuli would cause a stronger activation in right temporal cortex for participants who have had musical training than those who have not. Our results indicate that speech stimuli compared to rest activated the anterior superior temporal gyrus bilaterally and activated the right inferior frontal lobe. Music stimuli compared to rest did not activate the brain bilaterally, but rather only activated the right middle temporal gyrus. When the group analysis was performed with music experience as a covariate, we found that musical training did not affect activations to music stimuli specifically, but there was greater right hemisphere activation in several regions in response to speech stimuli as a function of more years of musical training. The results of the study agree with our hypotheses regarding the functional changes in the brain, but they conflict with our hypothesis about musical expertise. Overall, the study has generated interesting starting points for further explorations of how musical neural resources may be recruited for speech processing after damage to typical language networks.
stimuli played prior to the onset of speech production. In this experiment, we are examining the
specificity of the auditory stimulus by implementing congruent and incongruent speech sounds in
addition to non-speech sound. Electroencephalography (EEG) data was recorded for eleven adult
subjects in both speaking (speech planning) and silent reading (no speech planning) conditions.
Data analysis was accomplished manually as well as via generation of a MATLAB code to
combine data sets and calculate auditory modulation (suppression). Results of the P200
modulation showed that modulation was larger for incongruent stimuli than congruent stimuli.
However, this was not the case for the N100 modulation. The data for pure tone could not be
analyzed because the intensity of this stimulus was substantially lower than that of the speech
stimuli. Overall, the results indicated that the P200 component plays a significant role in
processing stimuli and determining the relevance of stimuli; this result is consistent with role of
P200 component in high-level analysis of speech and perceptual processing. This experiment is
ongoing, and we hope to obtain data from more subjects to support the current findings.
When we produce speech movements, we expect a specific auditory consequence, but an error occurs when the predicted outcomes do not match the actual speech outcome. The brain notes these discrepancies, learns from the errors, and works to lower these errors. Previous studies have shown a relationship between speech motor learning and auditory targets. Subjects with smaller auditory targets were more sensitive to errors. These subjects estimated larger perturbations and generated larger responses. However, these responses were often ineffective, and the changes were usually minimal. The current study examined whether subjects’ auditory targets can be manipulated in an experimental setting. We recruited 10 healthy young adults to complete a perceptual vowel categorization task. We developed a novel procedure where subjects heard different auditory stimuli and reported the stimuli by locating the stimuli relative to adjacent vowels. We found that when stimuli are closer to vowel boundary, subjects are less accurate. Importantly, by providing visual feedback to subjects, subjects were able to improve their accuracy of locating the stimuli. These results indicated that we might be able to improve subjects’ auditory targets and thus may improve their speech motor learning ability.
A fully automated Vowel Space Area (VSA) computation method is proposed that can be applied to any type of speech. It is shown that the VSA provides an efficient and reliable measure and is correlated to speech intelligibility. A clinical tool that incorporates the automated VSA was proposed for evaluation and treatment to be used by speech language pathologists. Two exploratory studies are performed using two databases by analyzing mean formant trajectories in healthy speech for a wide range of speakers, dialects, and coarticulation contexts. It is shown that phonemes crowded in formant space can often have distinct trajectories, possibly due to accurate perception.
A theory for analyzing time-varying signals models with amplitude modulation and frequency modulation is developed. Examples are provided that demonstrate other possible signal model decompositions with independent basis functions and corresponding physical interpretations. The Hilbert transform (HT) and the use of the analytic form of a signal are motivated, and a proof is provided to show that a signal can still preserve desirable mathematical properties without the use of the HT. A visualization of the Hilbert spectrum is proposed to aid in the interpretation. A signal demodulation is proposed and used to develop a modified Empirical Mode Decomposition (EMD) algorithm.