We study some speech enhancement algorithms based on the iterative Wiener filtering method due to Lim-Oppenheim [2], where the AR spectral estimation of the speech is carried out using a second-order analysis. But in our ...

The combination of Vector Quantization (VQ) and predictive techniques -named AVPC systems- has been shown as interesting systems for speech waveform coding at medium-high rates (ref. 1 to 3). In this work we present a such ...

This paper describes a coding scheme for broadband speech (sampling frequency 16KHz). We present a wideband speech encoder called APVQ (Adaptive Predictive Vector Quantization). It combines Subband Coding, Vector Quantization ...

This paper describes a coding scheme for broadband speech. It can be seen as a vectorial extension of an conventional ADPCM encoder. In this scheme, the vector signal is formed with one sample of the normalizaed prediction ...

The authors describe the acoustic processor of a Spanish continuous speech recognition system based on demisyllable units. The acoustic processor is based on a spotting algorithm which takes as input the unknown utterance, ...