Advances in Completely Automated Vowel Analysis for Sociophonetics: Using End-to-End Speech Recognition Systems With DARLA.
Ontology highlight
ABSTRACT:
SUBMITTER: Coto-Solano R
PROVIDER: S-EPMC8498339 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:Automated speech recognition (ASR) systems, which use sophisticated machine-learning algorithms to convert spoken language to text, have become increasingly widespread, powering popular virtual assistants, facilitating automated closed captioning, and enabling digital dictation platforms for health care. Over the last several years, the quality of these systems has dramatically improved, due both to advances in deep learning and to the collection of large-scale datasets used to train the systems. There is concern, however, that these tools do not work equally well for all subgroups of the population. Here, we examine the ability of five state-of-the-art ASR systems-developed by Amazon, Apple, Google, IBM, and Microsoft-to transcribe structured interviews conducted with 42 white speakers and 73 black speakers. In total, this corpus spans five US cities and consists of 19.8 h of audio matched on the age and gender of the speaker. We found that all five ASR systems exhibited substantial racial disparities, with an average word error rate (WER) of 0.35 for black speakers compared with 0.19 for white speakers. We trace these disparities to the underlying acoustic models used by the ASR systems as the race gap was equally large on a subset of identical phrases spoken by black and white individuals in our corpus. We conclude by proposing strategies-such as using more diverse training datasets that include African American Vernacular English-to reduce these performance differences and ensure speech recognition technology is inclusive.
| S-EPMC7149386 | biostudies-literature
Project description:Lipopolysaccharide (LPS), commonly known as endotoxin, is ubiquitous and the most-studied pathogen-associated molecular pattern. A component of Gram-negative bacteria, extracellular LPS is sensed by our immune system via the toll-like receptor (TLR)-4. Given that TLR4 is membrane bound, it recognizes LPS in the extracellular milieu or within endosomes. Whether additional sensors, if any, play a role in LPS recognition within the cytoplasm remained unknown until recently. The last decade has seen an unprecedented unfolding of TLR4-independent LPS sensing pathways. First, transient receptor potential (TRP) channels have been identified as non-TLR membrane-bound sensors of LPS and, second, caspase-4/5 (and caspase-11 in mice) have been established as the cytoplasmic sensors for LPS. Here in this review, we detail the brief history of LPS discovery, followed by the discovery of TLR4, TRP as the membrane-bound sensor, and our current understanding of caspase-4/5/11 as cytoplasmic sensors.
| S-EPMC7013859 | biostudies-literature
Project description:A central challenge for articulatory speech synthesis is the simulation of realistic articulatory movements, which is critical for the generation of highly natural and intelligible speech. This includes modeling coarticulation, i.e., the context-dependent variation of the articulatory and acoustic realization of phonemes, especially of consonants. Here we propose a method to simulate the context-sensitive articulation of consonants in consonant-vowel syllables. To achieve this, the vocal tract target shape of a consonant in the context of a given vowel is derived as the weighted average of three measured and acoustically-optimized reference vocal tract shapes for that consonant in the context of the corner vowels /a/, /i/, and /u/. The weights are determined by mapping the target shape of the given context vowel into the vowel subspace spanned by the corner vowels. The model was applied for the synthesis of consonant-vowel syllables with the consonants /b/, /d/, /g/, /l/, /r/, /m/, /n/ in all combinations with the eight long German vowels. In a perception test, the mean recognition rate for the consonants in the isolated syllables was 82.4%. This demonstrates the potential of the approach for highly intelligible articulatory speech synthesis.
| S-EPMC3628899 | biostudies-literature
Project description:Infants rapidly learn the sound categories of their native language, even though they do not receive explicit or focused training. Recent research suggests that this learning is due to infants' sensitivity to the distribution of speech sounds and that infant-directed speech contains the distributional information needed to form native-language vowel categories. An algorithm, based on Expectation-Maximization, is presented here for learning the categories from a sequence of vowel tokens without (i) receiving any category information with each vowel token, (ii) knowing in advance the number of categories to learn, or (iii) having access to the entire data ensemble. When exposed to vowel tokens drawn from either English or Japanese infant-directed speech, the algorithm successfully discovered the language-specific vowel categories (/I, i, epsilon, e/ for English, /I, i, e, e/ for Japanese). A nonparametric version of the algorithm, closely related to neural network models based on topographic representation and competitive Hebbian learning, also was able to discover the vowel categories, albeit somewhat less reliably. These results reinforce the proposal that native-language speech categories are acquired through distributional learning and that such learning may be instantiated in a biologically plausible manner.
| S-EPMC1934922 | biostudies-literature
Project description:With the rapid development of speech assistants, adapting server-intended automatic speech recognition (ASR) solutions to a direct device has become crucial. For on-device speech recognition tasks, researchers and industry prefer end-to-end ASR systems as they can be made resource-efficient while maintaining a higher quality compared to hybrid systems. However, building end-to-end models requires a significant amount of speech data. Personalization, which is mainly handling out-of-vocabulary (OOV) words, is another challenging task associated with speech assistants. In this work, we consider building an effective end-to-end ASR system in low-resource setups with a high OOV rate, embodied in Babel Turkish and Babel Georgian tasks. We propose a method of dynamic acoustic unit augmentation based on the Byte Pair Encoding with dropout (BPE-dropout) technique. The method non-deterministically tokenizes utterances to extend the token's contexts and to regularize their distribution for the model's recognition of unseen words. It also reduces the need for optimal subword vocabulary size search. The technique provides a steady improvement in regular and personalized (OOV-oriented) speech recognition tasks (at least 6% relative word error rate (WER) and 25% relative F-score) at no additional computational cost. Owing to the BPE-dropout use, our monolingual Turkish Conformer has achieved a competitive result with 22.2% character error rate (CER) and 38.9% WER, which is close to the best published multilingual system.
| S-EPMC8124527 | biostudies-literature
Project description:Interventions: Single arm:None
Primary outcome(s): Overall accuracy;Mean intersection over union (mIoU)
Study Design: Single arm
| 2691192 | ecrin-mdr-crc
Project description:Speech segmentation is a crucial step in automatic speech recognition because additional speech analyses are performed for each framed speech segment. Conventional segmentation techniques primarily segment speech using a fixed frame size for computational simplicity. However, this approach is insufficient for capturing the quasi-regular structure of speech, which causes substantial recognition failure in noisy environments. How does the brain handle quasi-regular structured speech and maintain high recognition performance under any circumstance? Recent neurophysiological studies have suggested that the phase of neuronal oscillations in the auditory cortex contributes to accurate speech recognition by guiding speech segmentation into smaller units at different timescales. A phase-locked relationship between neuronal oscillation and the speech envelope has recently been obtained, which suggests that the speech envelope provides a foundation for multi-timescale speech segmental information. In this study, we quantitatively investigated the role of the speech envelope as a potential temporal reference to segment speech using its instantaneous phase information. We evaluated the proposed approach by the achieved information gain and recognition performance in various noisy environments. The results indicate that the proposed segmentation scheme not only extracts more information from speech but also provides greater robustness in a recognition test.
| S-EPMC5120313 | biostudies-literature
Project description:Speech sounds evoke unique neural activity patterns in primary auditory cortex (A1). Extensive speech sound discrimination training alters A1 responses. While the neighboring auditory cortical fields each contain information about speech sound identity, each field processes speech sounds differently. We hypothesized that while all fields would exhibit training-induced plasticity following speech training, there would be unique differences in how each field changes. In this study, rats were trained to discriminate speech sounds by consonant or vowel in quiet and in varying levels of background speech-shaped noise. Local field potential and multiunit responses were recorded from four auditory cortex fields in rats that had received 10 weeks of speech discrimination training. Our results reveal that training alters speech evoked responses in each of the auditory fields tested. The neural response to consonants was significantly stronger in anterior auditory field (AAF) and A1 following speech training. The neural response to vowels following speech training was significantly weaker in ventral auditory field (VAF) and posterior auditory field (PAF). This differential plasticity of consonant and vowel sound responses may result from the greater paired pulse depression, expanded low frequency tuning, reduced frequency selectivity, and lower tone thresholds, which occurred across the four auditory fields. These findings suggest that alterations in the distributed processing of behaviorally relevant sounds may contribute to robust speech discrimination.
| S-EPMC4424170 | biostudies-literature
Project description:The representation of speech goals was explored using an auditory feedback paradigm. When talkers produce vowels the formant structure of which is perturbed in real time, they compensate to preserve the intended goal. When vowel formants are shifted up or down in frequency, participants change the formant frequencies in the opposite direction to the feedback perturbation. In this experiment, the specificity of vowel representation was explored by examining the magnitude of vowel compensation when the second formant frequency of a vowel was perturbed for speakers of two different languages (English and French). Even though the target vowel was the same for both language groups, the pattern of compensation differed. French speakers compensated to smaller perturbations and made larger compensations overall. Moreover, French speakers modified the third formant in their vowels to strengthen the compensation even though the third formant was not perturbed. English speakers did not alter their third formant. Changes in the perceptual goodness ratings by the two groups of participants were consistent with the threshold to initiate vowel compensation in production. These results suggest that vowel goals not only specify the quality of the vowel but also the relationship of the vowel to the vowel space of the spoken language.
| S-EPMC3663859 | biostudies-other
Project description:Purpose Although the speech intelligibility index (SII) has been widely applied in the field of audiology and other related areas, application of this metric to cochlear implants (CIs) has yet to be investigated. In this study, SIIs for CI users were calculated to investigate whether the SII could be an effective tool for predicting speech perception performance in a population with CI. Method Fifteen pre- and postlingually deafened adults with CI participated. Speech recognition scores were measured using the AzBio sentence lists. CI users also completed questionnaires and performed psychoacoustic (spectral and temporal resolution) and cognitive function (digit span) tests. Obtained SIIs were compared with predicted SIIs using a transfer function curve. Correlation and regression analyses were conducted on perceptual and demographic predictor variables to investigate the association between these factors and speech perception performance. Result Because of the considerably poor hearing and large individual variability in performance, the SII did not predict speech performance for this CI group using the traditional calculation. However, new SII models were developed incorporating predictive factors, which improved the accuracy of SII predictions in listeners with CI. Conclusion Conventional SII models are not appropriate for predicting speech perception scores for CI users. Demographic variables (aided audibility and duration of deafness) and perceptual-cognitive skills (gap detection and auditory digit span outcomes) are needed to improve the use of the SII for listeners with CI. Future studies are needed to improve our CI-corrected SII model by considering additional predictive factors. Supplemental Material https://doi.org/10.23641/asha.8057003.
| S-EPMC6808321 | biostudies-literature