Given that the amount of speaker speci c training data is always limited, for a given amount of data a speaker model has an optimum number of components. Here, this is investigated with regard to Gaussian mixture models (GMM) with and without world model adaption. Test results show that maximising the number of components in a speaker model can improve speaker recognition results. Comparisons w...

A bstract We obtain Proca field theory from the quantisation of $$ \mathcal{N} N = 2 supersymmetric worldline upon supplementing graded BRST-algebra with an extra multiplet oscillators. The linearised describes BV-extended spectrum theory, together a Stückelberg field. When coupling to background fields we de...

Since mathematical morphology is based on complete lattice theory, a vector ordering method becomes indispensable for its extension to multivariate images. Among the several approaches developed with this purpose, lexicographical orderings are by far the most frequent, as they possess certain desirable theoretical properties. However, their main drawback consists of the excessive priority attri...

Learning useful representations without supervision remains a key challenge in machine learning. In this paper, we propose a simple yet powerful generative model that learns such discrete representations. Our model, the Vector QuantisedVariational AutoEncoder (VQ-VAE), differs from VAEs in two key ways: the encoder network outputs discrete, rather than continuous, codes; and the prior is learnt...

We propose in this paper a new method for the quantisation of systems with first-class constraints. This method is a combination of the coherent-state-path-integral quantisation developed by Klauder, with the ideas of reduced state space quantisation. The key idea is that the physical Hilbert space may be defined by a coherent-state path-integral on the reduced state space and that the metric o...

The authors have developed a lossy coding scheme using a 3D discrete cosine transform, for use with integral 3D image data. Correct selection of the step sizes of the uniform symmetric scalar quantisers used in the scheme is crucial to attaining good rate-distortion performance. Imageorientated optimisations have been performed to obtain the quantisation parameters leading to minimum objective ...

For efficient coding of speech, it is desirable to separate the slowly and rapidly evolving spectral components to take advantage of their different perceptual qualities. In this paper, we present a multi-level wavelet decomposition mechanism, using low-delay FIR filters, applied to Waveform Interpolation coding. The technique overcomes the substantial delay problems of [2] and identifies a pre...