Title :
Adaptive speech estimation for low SNR and cochannel FM
Author_Institution :
Div. of Commun., Defence Sci. & Technol. Organ., Salisbury, SA, Australia
Abstract :
An adaptive method for estimating the speech message from low SNR and co-channel FM signals is described. A sinusoidal nonlinearity is introduced to provide some smoothness in the reconstruction of speech from noisy data. A saturating nonlinearity is introduced to reduce the effect of interfering co-channel signals. The unknown states and model parameters are learned via the expectation maximisation (EM) algorithm. The results of simulations are presented which demonstrate improvements in speech recovery compared to a conventional FM demodulator
Keywords :
adaptive estimation; adaptive signal processing; cochannel interference; frequency modulation; noise; optimisation; signal reconstruction; speech processing; FM demodulator; adaptive speech estimation; cochannel FM signals; expectation maximisation algorithm; interfering signals; model parameter learning; noisy data; saturating nonlinearity; signal to noise ratio; simulations; sinusoidal nonlinearity; smooth speech reconstruction; speech recovery; unknown states; Acoustic noise; Adaptive systems; Australia; Demodulation; Frequency modulation; Noise measurement; Phase measurement; Signal to noise ratio; Speech enhancement; Speech processing;
Conference_Titel :
Intelligent Information Systems, 1996., Australian and New Zealand Conference on
Conference_Location :
Adelaide, SA
Print_ISBN :
0-7803-3667-4
DOI :
10.1109/ANZIIS.1996.573972
