Dahimene, AbdelhakimNoureddine, MohamedAzrar, Arab2015-04-232015-04-232008https://dspace.univ-boumerdes.dz/handle/123456789/566This paper deals with the problem of peak clipped speech. Our basic assumption is that the clipped speech is voiced and can be linearly predicted with a high accuracy. The coefficients of linear prediction are computed using two different algorithms: a least square direct method and a recursive Kalman filter. The speech reconstruction is accomplished using backward predictionenSpeech signalClipped speechRestorationInterpolationLinear predictionLeast square methodKalman filterArticle