A novel approach to the fully automatic extraction of Fujisaki model parameters

Mixdorff, Hansjörg

doi:10.1109/icassp.2000.861811

Cited by 132 publications

(100 citation statements)

References 5 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Functional combinations that occur more than once are combined, so that there is no redundancy of functional representation. Note that such functional combination and boundary projection is an alternative to the superposition approach which requires F 0 -contour decomposition before parameter extraction, extracting two separate sets of parameters during trainings, and algorithmic summation during synthesis (Bailly and Holm, 2005;Fujisaki et al, 2005;Mixdorff et al, 2003). Here for each functional combination at the smallest temporal unit, only a single set of parameters need to be learned directly from the original (i.e., nondecomposed) F 0 contours during training and used during synthesis.…”

Section: Functional Annotationmentioning

confidence: 99%

“…Among the various aspects of prosody, fundamental frequency (F 0 ) is by far the most challenging, and has attracted most of the research effort. Many theories and computational models of F 0 patterns have been proposed over the years (Anderson et al, 1984;Bailly and Holm, 2005;Black and Hunt, 1996;Fujisaki et al, 2005;Grabe et al, 2007;Hirst, 2005Hirst, , 2011Jilka et al, 1999;Kochanski and Shih, 2003;Mixdorff et al, 2003;Pierrehumbert, 1980Pierrehumbert, , 1981Prom-on et al, 2009;Taylor, 2000;van Santen and Möbius, 2000;Xu and Wang, 2001;Xu, 2005), and a large number of empirical studies have been conducted (as reviewed by Wagner and Watson, 2010;ShattuckHufnagel and Turk, 1996;Xu, 2011). Despite the extensive effort, however, most of the critical issues still remain unresolved and some are still under heated debate (Arvaniti and Ladd, 2009;Ladd, 2008;Wagner and Watson, 2010;Wightman, 2002;Xu, 2011).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Toward invariant functional representations of variable surface fundamental frequency contours: Synthesizing speech melody via model-based stochastic learning

Prom-on

2014

Speech Communication

View full text Add to dashboard Cite

Variability has been one of the major challenges for both theoretical understanding and computer synthesis of speech prosody. In this paper we show that economical representation of variability is the key to effective modeling of prosody. Specifically, we report the development of PENTAtrainer -A trainable yet deterministic prosody synthesizer based on an articulatory-functional view of speech. We show with testing results on Thai, Mandarin and English that it is possible to achieve high-accuracy predictive synthesis of fundamental frequency contours with very small sets of parameters obtained through stochastic learning from real speech data. The first key component of this system is syllable-synchronized sequential target approximation -implemented as the qTA model, which is designed to simulate, for each tonal unit, a wide range of contextual variability with a single invariant target. The second key component is the automatic learning of function-specific targets through stochastic global optimization, guided by a layered pseudo-hierarchical functional annotation scheme, which requires the manual labeling of only the temporal domains of the functional units. The results in terms of synthesis accuracy demonstrate that effective modeling of the contextual variability is the key also to effective modeling of function-related variability. Additionally, we show that, being both theory-based and trainable (hence datadriven), computational systems like PENTAtrainer can serve as an effective modeling tool in basic research, with which the level of falsifiability in theory testing can be raised, and also a closer link between basic and applied research in speech science can be developed.3 Graphical Abstract Highlights (maximum 85 characters/bullet)• High synthetic accuracy of prosody achieved for Thai, Mandarin and English • Many-to-one mapping from contextually variable surface F 0 to invariant functional targets • Effectively handling of both contextual and non-contextual variability • Combination of deterministic synthesis and data-driven parameter learning • Large-scale and full-detailed prosody synthesis as tool for theory testing • Freely available as a Praat scripts and plug-ins to the speech science community at large

show abstract

Section: Functional Annotationmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Toward invariant functional representations of variable surface fundamental frequency contours: Synthesizing speech melody via model-based stochastic learning

Prom-on

2014

Speech Communication

View full text Add to dashboard Cite

show abstract

“…Such features include spectral, temporal, perceptual, short-time energy, and MPEG-7 standard descriptors [5]. Moreover, motivated by the high speech emotion recognition accuracy reported in [67], when Fujisaki's model parameters [38] are considered, these features are also tested here, as can be seen in Table 7. In addition, jitter and shimmer, are listed in Table 8.…”

Section: -161mentioning

confidence: 99%

Speaker-independent emotion recognition exploiting a psychologically-inspired binary cascade classification schema

Kotti

Paternò

2012

Int J Speech Technol

View full text Add to dashboard Cite

In this paper, a psychologically-inspired binary cascade classification schema is proposed for speech emotion recognition. Performance is enhanced because commonly confused pairs of emotions are distinguishable from one another. Extracted features are related to statistics of pitch, formants, and energy contours, as well as spectrum, cepstrum, perceptual and temporal features, autocorrelation, MPEG-7 descriptors, Fujisaki's model parameters, voice quality, jitter, and shimmer. Selected features are fed as input to K nearest neighborhood classifier and to support vector machines. Two kernels are tested for the latter: linear and Gaussian radial basis function. The recently proposed speaker-independent experimental protocol is tested on the Berlin emotional speech database for each gender separately. The best emotion recognition accuracy, achieved by support vector machines with linear kernel, equals 87.7%, outperforming state-of-the-art approaches. Statistical analysis is first carried out with respect to the classifiers' error rates and then to evaluate the information expressed by the classifiers' confusion matrices.

show abstract

“…We will assess the comparative performance of our algorithm with the results obtained with the CR parameter extraction tool of Mixdorff (2000). We will calculate the WCORR norm obtained with the CR model, and use it to assess the perceptual quality of the modelled contour, comparing it with our WCAD results.…”

Section: Experiments Designmentioning

confidence: 99%

“…Most intonation models use discontinuous pitch trackers and then interpolate unvoiced regions using, for instance, spline interpolation (Yu and Young, 2011). Then they concentrate the effort of modelling on the voiced parts (e.g., Mixdorff, 2000;Narusawa et al, 2002) or simply model everything equally (e.g., Hirst et al, 2000;Taylor, 2000). To avoid the latter, one needs a way of assessing which parts of the F 0 contour are perceptually relevant.…”

Section: Introductionmentioning

confidence: 99%

Intonation modelling using a muscle model and perceptually weighted matching pursuit

Honnet

Gerazov

Gjoreski

et al. 2018

Speech Communication

View full text Add to dashboard Cite

We propose a physiologically based intonation model using perceptual relevance. Motivated by speech synthesis from a speech-to-speech translation (S2ST) point of view, we aim at a language independent way of modelling intonation. The model presented in this paper can be seen as a generalisation of the command response (CR) model, albeit with the same modelling power. It is an additive model which decomposes intonation contours into a sum of critically damped system impulse responses. To decompose the intonation contour, we use a weighted correlation based atom decomposition algorithm (WCAD) built around a matching pursuit framework. The algorithm allows for an arbitrary precision to be reached using an iterative procedure that adds more elementary atoms to the model. Experiments are presented demonstrating that this generalised CR (GCR) model is able to model intonation as would be expected. Experiments also show that the model produces a similar number of parameters or elements as the CR model. We conclude that the GCR model is appropriate as an engineering solution for modelling prosody, and hope that it is a contribution to a deeper scientific understanding of the neurobiological process of intonation.

show abstract

A novel approach to the fully automatic extraction of Fujisaki model parameters

Cited by 132 publications

References 5 publications

Toward invariant functional representations of variable surface fundamental frequency contours: Synthesizing speech melody via model-based stochastic learning

Toward invariant functional representations of variable surface fundamental frequency contours: Synthesizing speech melody via model-based stochastic learning

Speaker-independent emotion recognition exploiting a psychologically-inspired binary cascade classification schema

Intonation modelling using a muscle model and perceptually weighted matching pursuit

Contact Info

Product

Resources

About