Dialect Classification From a Single Sonorant Sound Using Deep Neural Networks

Abstract

During spoken communication, the fine acoustic properties of human speech can reveal vital sociolinguistic and linguistic information about speakers and thus, these properties can function as reliable identification markers of speakers' identity. One key piece of information speech reveals is speakers' dialect. The first aim of this study is to provide a machine learning method that can distinguish the dialect from acoustic productions of sonorant sounds. The second aim is to determine the classification accuracy of dialects from the temporal and spectral information of a single sonorant sound and the classification accuracy of dialects using additional co-articulatory information from the adjacent vowel. To this end, this paper provides two classification approaches. The first classification approach aims to distinguish two Greek dialects, namely Athenian Greek, the prototypical form of Standard Modern Greek and Cypriot Greek using measures of temporal and spectral information (i.e., spectral moments) from four sonorant consonants /m n l r/. The second classification study aims to distinguish the dialects using coarticulatory information (e.g., formants frequencies F1 − F5, F0, etc.) from the adjacent vowel in addition to spectral and temporal information from sonorants. In both classification approaches, we have employed Deep Neural Networks, which we compared with Support Vector Machines, Random Forests, and Decision Trees. The findings show that neural networks distinguish the two dialects using a combination of spectral moments, temporal information, and formant frequency information with 81% classification accuracy, which is a 14% accuracy gain over employing temporal properties and spectral moments alone. In conclusion, Deep Neural Networks can classify the dialect from single consonant productions, making them capable of identifying sociophonetic shibboleths.

Effects of tDCS on Sound Duration in Patients with Apraxia of Speech in Primary Progressive Aphasia

“…The onset and offset of the first two vowel formants (F1 and F2) and the fundamental frequency (F0) were employed for the identification of vowels [ 72 , 73 ]. The onset and offset of frication (i.e., the noisy portion) was employed for the identification of fricatives [ 74 , 75 ]. Stop consonants were measured at the onset of the closure phase, including the burst [ 76 ].…”

Open Brain AI: An AI Research Platform

“…Additional tools incorporate IPA transcription and acoustic analysis tools. Open Brain AI enables end-to-end spoken and written speech production analysis by combining the different computational pipelines to provide automated and objective linguistic measures (1,16,18,19,24,26,31,32).…”

“…For example, Themistocleous, Webster (19) analyzed connected speech productions from 52 individuals with PPA using a morphological tagger and showed differences in POS production in patients with non-fluent Primary Progressive Aphasia (nfvPPA), logopenic variant of Primary Progressive Aphasia (lvPPA), and the semantic variant of Primary Progressive Aphasia (svPPA). Also, we have employed machine learning to identify speakers with different dialects, from speech acoustics, namely prosody (20)(21)(22), vowels (23,24), and consonants (21,(25)(26)(27)(28). Machine learning was used to track the learning of L1 dialectal learners of the standard language variety in classrooms (29), showing the implication of using machine learning applications in diverse populations.…”

SoundexGR: An algorithm for phonetic matching for the Greek language

“…Finally, Themistocleous (2019) describes classification approaches based on deep neural networks for distinguishing two Greek dialects, namely Athenian Greek, the prototypical form of Standard Modern Greek and Cypriot Greek. That work is based on the acoustic features of spoken language.…”