The current paper presents a preliminary study of Finnish prosody using a quantitative model. In a small corpus of segmentally identical utterances the focus, the sentence mode and the place of the phrase boundary were systematically varied. Following the rationale of MFGI, the MixdorffFujisaki Model of German intonation, the phonetic properties of the contrast-lending intonemes were examined. ...

In this paper, we propose a Modulation Spectrum-based manageable feature set for detection of depressed speech. Modulation Spectrum (MS) is obtained from the conventional speech spectrogram by spectral analysis along the temporal trajectories of the acoustic frequency bins. While MS representation of speech provides rich and high-dimensional joint frequency information, extraction of discrimina...

Database of contemporary Russian speech is being created for the Computer Fund of the Russian Language. Database consists of the corpus of sound files and their descriptions. Database sound material is formed by recordings of the Phonetically Representative Texts pronounced by 1) standard Russian speakers, 2) regional Russian speakers, 3) speakers from the former USSR republics to which Russian...

this study examined the phonetic characteristics of persian sentence stress produced by native mandarin chinese speakers. we compared the phonetic correlates of sentence stress, namely fundamental frequency (f0), vowel duration, and vowel intensity in the production of 8 mandarin chinese speakers and 8 native persian speakers. results indicated that mandarin chinese speakers of persian could di...

Cantonese is a major Chinese dialect with a complicated tone system. This research focuses on quantitative modeling of Cantonese tones. It uses Stem-ML, a language-independent framework for quantitative intonation modeling and generation. A set of F0 prediction models are built, and trained on acoustic data. The prediction error is about 11 Hz or 1 semitone. The resulting optimal model paramete...

The age of speaker is a vital ingredient for many applications. Applications where access is based on age group can automatically use this simple system to identify the same. Access to machine or any equipment can also be authenticated with this system. Although there exist many difficulties when working with voice, it is a rich source of abundant information within it. With the help of acousti...

Our research aims at exploring the nature of the interferences that occur during the speech-in-speech situation. French target words were inserted in 2 types of backgrounds: (i) 4-talker babble spoken in various languages such as French, Italian or Irish, containing acoustic and linguistic information, (ii) fluctuating noise derived from each 4-talker babble signal, with only acoustic informati...

Vocal emotions, as well as different speaking styles and speaker traits are characterized by a complex interplay of multiple prosodic features. Natural sounding speech synthesis with the ability to control such paralinguistic aspects requires the manipulation of the corresponding prosodic features. With traditional concatenative speech synthesis it is easy to manipulate the “primary” prosodic f...

Interviews are a recognized and valued method for obtaining research data. Usually, these data drawn from the content of interviews. However, words prosodic elements that accompany them may reveal complementary, even contradictory information, substantially alters researcher’s first impression. After publication in September 2018 document COAlition Plan S which announced results funded by Eur...

Cantonese is a major Chinese dialect with a complicated tone system. This research focuses on quantitative modeling of Cantonese tones. It uses Stem-ML, a language-independent framework for quantitative intonation modeling and generation. A set of F0 prediction models are built, and trained on acoustic data. The prediction error is about 11 Hz or 1 semitone. The resulting optimal model paramete...