Rhythmic unit extraction and modelling for automatic language identification

Abstract

This paper deals with an approach to Automatic Language Identification based on rhythmic modelling. Beside phonetics and phonotactics, rhythm is actually one of the most promising features to be considered for language identification, even if its extraction and modelling are not a straightforward issue. Actually, one of the main problems to address is what to model. In this paper, an algorithm of rhythm extraction is described: using a vowel detection algorithm, rhythmic units related to syllables are segmented. Several parameters are extracted (consonantal and vowel duration, cluster complexity) and modelled with a Gaussian Mixture. Experiments are performed on read speech for 7 languages (English, French, German, Italian, Japanese, Mandarin and Spanish) and results reach up to 86 ± 6% of correct discrimination between stress-timed mora-timed and syllable-timed classes of languages, and to 67 ± 8% percent of correct language identification on average for the 7 languages with utterances of 21 seconds. These results are commented and compared with those obtained with a standard acoustic Gaussian mixture modelling approach (88 ± 5% of correct identification for the 7-languages identification task).

Comparative Study of Gaussian Mixture Model and Radial Basis Function for Voice Recognition

“…At the end of the study, the two models produced 80% and 81% maximum recognition rates respectively. This is better than the 80% recognition rate of the GMM proposed by Jean-Luc et al in [4] and very close to their acoustic GMM version with 83% recognition rate as well as the GMM proposed by [5]. The DTREG version of RBF produced a landmark 94.8% recognition rate outperforming the other two techniques and similar techniques earlier reported in literature.…”

“…Using a vowel detection algorithm, [4] segmented rhythmic units related to syllables by extracting parameters such as consonantal and vowel duration, and cluster complexity and modeled with a Gaussian Mixture. Results reached up to 86 ± 6% of correct discrimination between stress-timed, moratimed and syllable-timed classes of languages.…”

“…Finally, the response of each hidden unit is scaled by its connecting weights to the output units and then summed to produce the overall network output. Therefore, the k th output of the network ŷ k is: (4) where Φ j (x) is the response of the j th hidden unit, w jk is the connecting weight between the j th hidden unit and the k th output unit, and w 0 is the bias term [18].…”

Time-Scale Feature Extractions for Emotional Speech Characterization

“…The role of rhythm in the perception of sounds is very important [46] and it has been shown to be efficient for language identification [43,47]. Most of the models proposed in the literature for the extraction of rhythmic features require the definition of a rhythmic unit (e.g., vowels, syllable) and a metric (inter, intra units) [48,49].…”

“…As a result, the obtained segments are termed as pseudophonetic units. This method has been introduced for automatic language identification [43] and consists in characterizing pseudo-syllables which have been defined by gathering the consonants preceding the detected vowels (C n V structure). The study of these pseudo-syllables made possible the characterization of two main groups of language described in the literature: stressed (English, German) and syllabic (French and Spanish).…”