| 摘要 |
<p>1356649 Speech transmission systems STANDARD TELEPHONES & CABLES Ltd 23 Dec 1971 59932/71 Heading H4R In a speech transmission system a self adaptive filter 1 is connected in a negative feed forward configuration in the speech path with the speech input signal for the filter also connected to the filter as the target signal, and the filter output being subtracted at 2 from the speech input signal to provide a residual signal output, from which the excitation function signal can be extracted. The excitation function signal and the multiplier coefficients of the adaptive filter are coded and transmitted to the receiver where the signals are decoded and fed to an excitation generator, to provide the input, and the multiplier adjustments of, an adaptive filter 9 connected in a positive feedback configuration, so that the original speech is available at the output of adder 10. In an alternative embodiment, Fig. 3, a decoder 24, 25, is incorporated in the transmitter with the adaptive filters 24, 23 of the decoders in the transmitter and the receiver both fed with an error signal derived by comparison of the output of filter 28, whose multipliers are directly coupled to the multipliers in the self adaptive filter 29, with the output of filter 24, so that the multipliers of filter 24, and hence filter 23, are continuously adjusted to be the same as those of filter 28 and hence filter 29, with the coder and decoders in the adjustment loop so that compensation is provided for quantizing errors &c. The excitation function signal and the multiplier coefficients, or error signal in the embodiment of Fig. 3, are slowly varying functions allowing low speed sampling for transmission. Since it is possible for the sampling to be of the same order as the resonance period of the input signal the sampling interval is made to vary in a pseudo random fashion. The adaptive filters are convention transversal filters with the self adaptive filter being described with respect to Fig. 2 (not shown).</p> |
