Speech Compression by Least-Squares Interpolation
A. Arcese, A. J. Goldberg · The Journal of the Acoustical Society of America · 1973
In this paper we present the theory and application of an adaptive interpolator which restores speech samples previously eliminated to lower the transmission rate. During waveform reconstruction, these detected samples are estimated as a weighted sum of neighboring past and future waveform amplitudes that were transmitted. These weights are chosen at the transmitter to minimize the averaged squared difference between the actual samples and the interpolated samples over an analysis length of 20 msec. Audio tapes will be played showing the results of quantizing the interpolation parameters.