A comparative study between linear and nonlinear speech prediction

Faundez-Zanuy, Marcos; Monte, Enric; Vallverdú, Francesc

doi:10.1007/bfb0032575

Cited by 9 publications

(5 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The nonlinear prediction can be done by several methods like the Volterra filters [5] or neural networks [6]. The major advantage of the Volterra filters is that, like in linear predictors, the least mean square error solution for the filter coefficients can be expressed analytically.…”

Section: Neural Predictive Codingmentioning

confidence: 99%

A new nonlinear feature extraction algorithm for speaker verification

Chétouani¹,

Gas²,

Zarader³

et al. 2004

Interspeech 2004

View full text Add to dashboard Cite

In this paper we propose a new parameterization algorithm based on nonlinear prediction, which is an extension of the classical LPC parameters. The parameters performances are estimated by two different methods: the Arithmetic-Harmonic Sphericity (AHS) and the Auto-Regressive Vector Model (ARVM). Two different methods are proposed for the parameterization based on the Neural Predictive Coding (NPC): classical neural networks initialization and linear initialization. We applied these two parameters to speaker identification. The fist parameters obtained smaller rates. We show for the first parameters how they can be combined with the classical parameters (LPCC, MFCC, etc.) in order to improve the results of only one classical parameterization (MFCC provides 97.55% and MFCC+NPC 98.78%). For the linear initialization, we obtain 100% which is great improvement. This study opens a new way towards different parameterization schemes that offer better accuracy on speaker recognition tasks.

show abstract

Section: Neural Predictive Codingmentioning

confidence: 99%

A new nonlinear feature extraction algorithm for speaker verification

Chétouani¹,

Gas²,

Zarader³

et al. 2004

Interspeech 2004

View full text Add to dashboard Cite

show abstract

“…A more detailed explanation about the nonlinear predictive model based on neural nets can be found in [8] and [9]. This paper is focused on the speech coding application.…”

Section: Adaptive Adpcm With Hybrid Predictor Schemementioning

confidence: 99%

“…The results have been obtained with the following database: 8 speakers (4 males & 4 females) sampled at 8Khz and quantized at 12 bits/sample. Additional details about the predictor and the database were reported in [8] and [9]. !…”

Section: Adaptive Adpcm With Hybrid Predictor Schemementioning

confidence: 99%

“…b) Nonlinear predictor For the nonlinear predictor based on neural nets, the number of parameters that must be optimized is greater. The selected network architecture is the Multi-Layer Perceptron with 10 input neurons, 2 hidden neurons with a sigmoid transfer function and one output neuron with a linear transfer function trained with the Levenberg-Marquardt (L-M) algorithm, based on our previous results [8]. We have also evaluated a recurrent Elman net, but we found that its behaviour was worse than MLP trained with L-M.…”

Section: Adaptive Adpcm With Hybrid Predictor Schemementioning

confidence: 99%

See 1 more Smart Citation

Adaptive hybrid speech coding with a MLP/LPC structure

Faundez-Zanuy

1999

Lecture Notes in Computer Science

View full text Add to dashboard Cite

In the last years there has been a growing interest for nonlinear speech models. Several works have been published revealing the better performance of nonlinear techniques, but little attention has been dedicated to the implementation of the nonlinear model into real applications. This work is focused on the study of the behaviour of a combined linear/nonlinear predictive model based on linear predictive coding (LPC-10) and neural nets, in a speech waveform coder. Our novel scheme obtains an improvement in SEGSNR between 1 and 2.5 dB for an adaptive quantization ranging from 2 to 5 bits.

show abstract

“…The major advantage of Volterra filters is that, like in linear predictors, the least mean square solution for the filter coefficients can he expressed analytically but the main drawback lie in the fact that the number of coefficients grows fast with the prediction window. Predictive neural networks have already been successfully applied to speech [6]. The neural networks weights can estimate the vocal tract model, as in linear predictive coding.…”

Section: Introductionmentioning

confidence: 99%

Discriminative training for neural predictive coding applied to speech features extraction

Chétouani¹,

Gas²,

Zarader³

2002

Proceedings of the 2002 International Joint Conference on Neural Networks. IJCNN'02 (Cat. No.02CH37290)

View full text Add to dashboard Cite

In this paper, we present a predictive neural network called Neural Predictive Coding (NPC). This model is used for non linear discriminant features extraction (DFE) applied to phoneme recognition. We validate the nonlinear prediction improvement of the NPC model. We also, present a new extension of the NPC model : NPCJ. In order to evaluate the performances of the NPC-3 model, we carried out a study of Darpa-Timit phonemes (in particular /bl, Id/, lgl and /p/, /U, /q/ phonemes) recognition. Comparisons with traditionnal coding methods are presented: they put in obsviousness an improvement of the classification. We also show how an adaptative constraint allows improvements on recognition task 1 0-7803-7278-6/02/$10.00 02002 EEE

show abstract

A comparative study between linear and nonlinear speech prediction

Cited by 9 publications

References 6 publications

A new nonlinear feature extraction algorithm for speaker verification

A new nonlinear feature extraction algorithm for speaker verification

Adaptive hybrid speech coding with a MLP/LPC structure

Discriminative training for neural predictive coding applied to speech features extraction

Contact Info

Product

Resources

About