A comparison of cepstral coefficients and spectral moments in the classification of Romanian fricatives. (July 2016)
- Record Type:
- Journal Article
- Title:
- A comparison of cepstral coefficients and spectral moments in the classification of Romanian fricatives. (July 2016)
- Main Title:
- A comparison of cepstral coefficients and spectral moments in the classification of Romanian fricatives
- Authors:
- Spinu, Laura
Lilley, Jason - Abstract:
- Abstract: In this paper we explore two methods for the classification of fricatives. First, for the coding of the speech, we compared two sets of acoustic measures obtained from a corpus of Romanian fricatives: (a) spectral moments and (b) cepstral coefficients. Second, we compared two methods of determining the regions of the segments from which the measures would be extracted. In the first method, the phonetic segments were divided into three regions of approximately equal duration. In the second method, Hidden Markov Models (HMMs) were used to divide each segment into three regions such that the variances of the measures within each region were minimized. The corpus we analyzed consists of 3674 plain and palatalized word-final fricatives from four places of articulation, produced by 31 native speakers of Romanian (20 females). We used logistic regression to classify fricatives by place, voicing, palatalization status, and gender. We found that cepstral coefficients reliably outperformed spectral moments in all classification tasks, and that using regions determined by HMM yielded slightly higher correct classification rates than using regions of equal duration. Abstract : Highlights: Cepstral coefficients classify fricatives more accurately than spectral moments. HMM regions based on internal variance have an advantage over equal-length regions. Consonantal information alone successfully classifies Romanian fricatives at 4 POAs. Partial devoicing occurs in Romanian voicedAbstract: In this paper we explore two methods for the classification of fricatives. First, for the coding of the speech, we compared two sets of acoustic measures obtained from a corpus of Romanian fricatives: (a) spectral moments and (b) cepstral coefficients. Second, we compared two methods of determining the regions of the segments from which the measures would be extracted. In the first method, the phonetic segments were divided into three regions of approximately equal duration. In the second method, Hidden Markov Models (HMMs) were used to divide each segment into three regions such that the variances of the measures within each region were minimized. The corpus we analyzed consists of 3674 plain and palatalized word-final fricatives from four places of articulation, produced by 31 native speakers of Romanian (20 females). We used logistic regression to classify fricatives by place, voicing, palatalization status, and gender. We found that cepstral coefficients reliably outperformed spectral moments in all classification tasks, and that using regions determined by HMM yielded slightly higher correct classification rates than using regions of equal duration. Abstract : Highlights: Cepstral coefficients classify fricatives more accurately than spectral moments. HMM regions based on internal variance have an advantage over equal-length regions. Consonantal information alone successfully classifies Romanian fricatives at 4 POAs. Partial devoicing occurs in Romanian voiced labial fricatives word-finally. The dorsal fricative of Romanian is realized mostly as a velar and palatal fricative. … (more)
- Is Part Of:
- Journal of phonetics. Volume 57(2016:Jul.)
- Journal:
- Journal of phonetics
- Issue:
- Volume 57(2016:Jul.)
- Issue Display:
- Volume 57 (2016)
- Year:
- 2016
- Volume:
- 57
- Issue Sort Value:
- 2016-0057-0000-0000
- Page Start:
- 40
- Page End:
- 58
- Publication Date:
- 2016-07
- Subjects:
- Fricatives -- Cepstral coefficients -- Spectral moments -- Place of articulation -- Secondary palatalization -- Classification -- Romanian
Phonetics -- Periodicals
Phonetics -- Periodicals
Phonétique -- Périodiques
Phonetics
Periodicals
Electronic journals
414.05 - Journal URLs:
- http://www.sciencedirect.com/science/journal/00954470 ↗
http://www.elsevier.com/journals ↗ - DOI:
- 10.1016/j.wocn.2016.05.002 ↗
- Languages:
- English
- ISSNs:
- 0095-4470
- Deposit Type:
- Legaldeposit
- View Content:
- Available online (eLD content is only available in our Reading Rooms) ↗
- Physical Locations:
- British Library DSC - 5034.550000
British Library DSC - BLDSS-3PM
British Library HMNTS - ELD Digital store - Ingest File:
- 1509.xml