“…PE13: VT = (20,t*,20] VF = (1,2,3,4,5,6,7,8,9], PE13 is executed every 20 msecs in the frication interval. It has 16 nodes in the first layer and 10 nodes in the second layer.…”
Section: Organizationmentioning
confidence: 99%
“…Sonorant segments can be extracted from continuous speech using a procedure described in [5]. Sonorant segments are characterized by narrow-band resonances from which spectral lines as introduced in [lo] are extracted.…”
A set of Multi-Layered Networks allows the integration of information extracted with variable resolution in the time and frequency domains and to keep the number of links between nodes of the networks small for significant generalization during learning with a reasonable training set size.
“…PE13: VT = (20,t*,20] VF = (1,2,3,4,5,6,7,8,9], PE13 is executed every 20 msecs in the frication interval. It has 16 nodes in the first layer and 10 nodes in the second layer.…”
Section: Organizationmentioning
confidence: 99%
“…Sonorant segments can be extracted from continuous speech using a procedure described in [5]. Sonorant segments are characterized by narrow-band resonances from which spectral lines as introduced in [lo] are extracted.…”
A set of Multi-Layered Networks allows the integration of information extracted with variable resolution in the time and frequency domains and to keep the number of links between nodes of the networks small for significant generalization during learning with a reasonable training set size.
“…The signal x(t) is processed in order to obtain a sequence of spectra R. This is done with Fast Fourier Transformation (FFT) but for other projects in our research an ear model is used [7]. Another module extracts a sequence A of acoustic properties like zerocrossing rates and energy contour profiles as described in [8]. Two vectors of acoustic parameters are extracted for every frame, namely M (a vector of 6 MCC) and AM (a vector of 6 DMCC).…”
A hybrid coder is introduced for obtaining descriptions of speech patterns, This coder uses popular Vector Quantization (VQ) techniques on melscale cepstral coefficients and their derivatives together with a Recurrent Network (RN) for describing suprasegmental features of speech.The purpose of these features is to focus the search when Hidden Markov Models (HMM) are used for speech unit or word models.Preliminary experiments of speakerindependent connected digit recognition showed that using a hybrid coder based on a RN improved recognition performance.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.