Automatic prominent syllable detection with machine learning classifiers

Johnson, David; Kang, Okim

doi:10.1007/s10772-015-9299-z

Cited by 11 publications

(5 citation statements)

References 42 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This set of features was used in a discriminant function to classify between accented and unaccented syllables, achieving 91.9% of correct classification of accented syllables and 92.2% of correct classification of unaccented syllables for the new population of English speakers affected with ataxic dysarthria. The classification accuracy results are comparable with the results of our previous study for native Dutch speakers (healthy and dysarthric speech) and with other studies of accent detection in healthy speech [30][31][32][33][34][35]. The results suggest that combining the ten acoustic parameters developed by Mendoza et al [41] has a good capacity to discriminate between accented and unaccented syllables in healthy and speech-impaired speakers of Germanic languages with comparable accentuation patterns, such as English and Dutch.…”

Section: Cross-population Validation Of Acoustic Featuressupporting

confidence: 88%

“…Studies of acoustic correlates of sentence accent have provided valuable insight into this domain [12,[25][26][27][28][29]. Acoustic accent production descriptors have been studied in healthy speech [30][31][32][33][34][35][36][37] and in dysarthria [9,10,16,24,38,39]. Currently, there is general agreement in the literature that syllable duration, pitch pattern, and intensity (or sub-band energy) correlate with accentuation [26,40].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Acoustic Identification of Sentence Accent in Speakers with Dysarthria: Cross-Population Validation and Severity Related Patterns

et al. 2021

View full text Add to dashboard Cite

Dysprosody is a hallmark of dysarthria, which can affect the intelligibility and naturalness of speech. This includes sentence accent, which helps to draw listeners’ attention to important information in the message. Although some studies have investigated this feature, we currently lack properly validated automated procedures that can distinguish between subtle performance differences observed across speakers with dysarthria. This study aims for cross-population validation of a set of acoustic features that have previously been shown to correlate with sentence accent. In addition, the impact of dysarthria severity levels on sentence accent production is investigated. Two groups of adults were analysed (Dutch and English speakers). Fifty-eight participants with dysarthria and 30 healthy control participants (HCP) produced sentences with varying accent positions. All speech samples were evaluated perceptually and analysed acoustically with an algorithm that extracts ten meaningful prosodic features and allows a classification between accented and unaccented syllables based on a linear combination of these parameters. The data were statistically analysed using discriminant analysis. Within the Dutch and English dysarthric population, the algorithm correctly identified 82.8 and 91.9% of the accented target syllables, respectively, indicating that the capacity to discriminate between accented and unaccented syllables in a sentence is consistent with perceptual impressions. Moreover, different strategies for accent production across dysarthria severity levels could be demonstrated, which is an important step toward a better understanding of the nature of the deficit and the automatic classification of dysarthria severity using prosodic features.

show abstract

Section: Cross-population Validation Of Acoustic Featuressupporting

confidence: 88%

Section: Introductionmentioning

confidence: 99%

Acoustic Identification of Sentence Accent in Speakers with Dysarthria: Cross-Population Validation and Severity Related Patterns

et al. 2021

View full text Add to dashboard Cite

show abstract

“…Although the nature of prominence is already pretty well understood and was investigated from very different angles [7], a few questions still remain open. In view of the growing attention paid to prominence matters in the rapidly growing fields of speech technology and human-machine-interaction [8,9], a very important question concerns the quantification of prominence-cue hierarchies. For example, for German, which is the subject of the present study, [10] concluded that F0 is the primary cue to prominence, as it varied more between prominent and non-prominence syllables than duration.…”

Section: Introductionmentioning

confidence: 99%

The Relative Cueing Power of F0 and Duration in German Prominence Perception

Niebuhr

Winkler

2017

Interspeech 2017

View full text Add to dashboard Cite

Previous studies showed for German and other (West) Germanic language, including English, that perceived syllable prominence is primarily controlled by changes in duration and F0, with the latter cue being more powerful than the former. Our study is an initial approach to develop this prominence hierarchy further by putting numbers on the interplay of duration and F0. German listeners indirectly judged through lexical identification the relative prominence levels of two neighboring syllables. Results show that an increase in F0 of between 0.49 and 0.76 st is required to outweigh the prominence effect of a 30% increase in duration of a neighboring syllable. These numbers are fairly stable across a large range of absolute F0 and duration levels and hence useful in speech technology.

show abstract

“…Numerous studies have investigated automated lexical stress classification since the early 2000s. Most systems utilize F0, intensity, and duration measures along with various machine learning algorithms to predict the stress patterns of words (Chen and Wang 2010;Chen and Jang 2012;Deshmukh and Verma 2009;Johnson and Kang 2015;Li et al 2018;Tepperman and Narayanan 2005). A number of systems also incorporate segmental information, like cepstral coefficients (Ferrer et al 2015;Li et al 2007).…”

Section: Related Workmentioning

confidence: 99%

Enhancing Language Learners’ Comprehensibility through Automated Analysis of Pause Positions and Syllable Prominence

Coulange,

Kato,

Rossato

et al. 2024

Languages

View full text Add to dashboard Cite

This research paper addresses the challenge of providing effective feedback on spontaneous speech produced by second language (L2) English learners. As the position of pauses and lexical stress is often considered a determinative factor for easy comprehension by listeners, an automated pipeline is introduced to analyze the position of pauses in speech, the lexical stress patterns of polysyllabic content words, and the degree of prosodic contrast between stressed and unstressed syllables, on the basis of F0, intensity, and duration measures. The pipeline is applied to 11 h of spontaneous speech from 176 French students with B1 and B2 proficiency levels. It appeared that B1 students make more pauses within phrases and less pauses between clauses than B2 speakers, with a large diversity among speakers at both proficiency levels. Overall, lexical stress is correctly placed in only 35.4% of instances, with B2 students achieving a significantly higher score (36%) than B1 students (29.6%). However, great variation among speakers is also observed, ranging from 0% to 68% in stress position accuracy. Stress typically falls on the last syllable regardless of the prosodic expectations, with the strong influence of syllable duration. Only proficient speakers show substantial F0 and intensity contrasts.

show abstract

Automatic prominent syllable detection with machine learning classifiers

Cited by 11 publications

References 42 publications

Acoustic Identification of Sentence Accent in Speakers with Dysarthria: Cross-Population Validation and Severity Related Patterns

Acoustic Identification of Sentence Accent in Speakers with Dysarthria: Cross-Population Validation and Severity Related Patterns

The Relative Cueing Power of F0 and Duration in German Prominence Perception

Enhancing Language Learners’ Comprehensibility through Automated Analysis of Pause Positions and Syllable Prominence

Contact Info

Product

Resources

About