Jacques Prado

Slow (0.7-2 Hz) and fast (2-4 Hz) delta components are differently correlated to theta, alpha and beta frequency bands during NREM sleep

OBJECTIVES Spectral power of 40 all-night sleep EEGs (Cz-Pz bipolar lead) recorded in 20 healthy young subjects was calculated after normalization on 30-s consecutive epochs by means of an autocorrelation method based on a 15-order autoregressive model. METHODS The spectral parameters were calculated for the 7 main EEG bands: slow delta (0.7-2 Hz); fast delta (2-4 Hz); theta (4-8 Hz); alpha (8-12 Hz); sigma (12-16 Hz); beta1 (16-35 Hz); and beta 2 (>35 Hz). RESULTS Strong negative correlations were found between power in the fast delta and either the alpha or the beta bands and between slow delta and theta bands, whereas the two delta bands showed little correlation with each other. CONCLUSION The possibility that theses different relationships of slow and fast delta components with other frequency bands might reflect the neocortical or the thalamocortical origin of the delta waves is discussed.

Frequency-domain adaptive filtering with applications to acoustic echo cancellation

In this paper an algorithm is presented for adaptive filtering in the frequency-domain with application to acoustic echo cancellation. This algorithm, called generalized multi-delay filter(GMDFα, is derived from the frequency-domain implementation of the time-domain block least mean square algorithm. Two different implementations are introduced, one based on the discrete Fourier transform(dft) and one based on the discrete Hartley transform(DHT); some results on fixed-point implementation of the algorithm are provided which are compared to results obtained from floating point implementation. Finally, the application of thegmdfα algorithm to acoustic echo cancellation, in hand-free telephone systems, is detailed. Some control strategies are presented; in particular a novel double-talk detector based on a spectral dissimilarity measure is introduced ; also, a twin-filter structure which significantly enhances the echo rejection is derived.RésuméCet article présente un algorithme d’identification adaptative en fréquence avec application à l’annulation d’écho acoustique dans un contexte de téléphonie mains-libres. Cet algorithme, dit filtre généralisé à délais multiples (gmdfα), est dérivé de la réalisation fréquen-tielle de l’algorithme du gradient stochastique en blocs. Deux versions sont proposées, l’ une utilise la transformation de Fourier discrète et l’autre la transformée de Hartley discrète. Une méthode d’évaluation de la réalisation en arithmétique virgule fixe de l’ algorithme est illustrée par la présentation de courbes de convergence permettant d’apprécier le comportement par rapport à une implantation en arithmétique virgule flottante. Les auteurs proposent l’introduction de différentes commandes pour l’application de l’algorithme à l’annulation d’écho acoustique. En particulier, ils présentent un nouvel algorithme de détection de double-parole, fondé sur l’évaluation d’une mesure de dissimilarité spectrale, et l’utilisation d’une structure à filtres jumeaux améliorant la réjection d’écho.

A polynomial-transform based computation of the 2-D DCT with minimum multiplicative complexity

This paper provides a full description of a 2-D discrete cosine transform (DCT) which has the property of requiring only N 1-D DCT of size N for performing the 2-D algorithm. No multiplication is involved elsewhere in the algorithm, thus the multiplicative complexity is halved when compared to classical row/column algorithms. Compared to some other proposed algorithms, our approach has the characteristic of being formally very concise. Once the main result has been obtained, a remaining difficulty is the search for an in-place algorithm. This problem is also solved.

Evaluation of an ERB frequency scale noise reduction for hearing aids: a comparative study

Among speech enhancement methods, the Ephraim and Malah suppression rule (EMSR) has proven to be efficient in reducing the background noise while preventing from a common artefact: the musical noise. From psychoacoustic motivation, an implementation of the EMSR with a perceptually relevant frequency partition is proposed. This implementation is based on non-uniform oversampled filter-banks. The frequency resolution is nevertheless uniform on the equivalent rectangular bandwidth (ERB)-scale. Objective and subjective comparison with classical EMSR and uniform critically decimated filter-banks implementations has been achieved.

A flexible multi-resolution time-frequency analysis framework for audio signals

In this article, a new Constant-Q transform implementation is proposed together with a generalization of both the Short-Term Fourier Transform and the Constant-Q transform. The purpose of this generalization is to provide a discrete time-frequency analysis tool with arbitrary center frequency and frequency resolution for each bin of the transform. This new analysis framework is very flexible and can be related to many common time-frequency transforms. Furthermore, new interesting transforms with specific time-frequency resolutions can be defined. To illustrate and validate the corresponding approach, experimental results and examples are provided for different configurations.