초록 ▼

A predictive encoder is disclosed which has a non-linear quantizing characteristic. The transmission gain of the quantizer is less than unity when the amplitude component of the predictive error signal is smaller than a predetermined level and is substantially equal to unity when the amplitude compo

A predictive encoder is disclosed which has a non-linear quantizing characteristic. The transmission gain of the quantizer is less than unity when the amplitude component of the predictive error signal is smaller than a predetermined level and is substantially equal to unity when the amplitude component of the predictive error signal is higher than the predetermined level. As a result, undesired redundancy components are effectively suppressed without causing an increase in quantizing noise and a deterioration in information quality.

대표청구항 ▼

A predictive encoder for encoding an input signal of high correlation by the use of correlation techniques, said encoder comprising: means for producing a predictive signal corresponding to the amplitude value to said input signal at a preceding sample; means supplied with said input signal and said

A predictive encoder for encoding an input signal of high correlation by the use of correlation techniques, said encoder comprising: means for producing a predictive signal corresponding to the amplitude value to said input signal at a preceding sample; means supplied with said input signal and said predictive signal in a timed relation for producing a predictive error signal representing the difference therebetween; and quantizer means for quantizing said predictive error signal to produce information codes, said quantizer means including a plurality of circuits having different non-linear characteristics wherein said circuits are read-only memories which produce information codes according to respective ones of said different non-linear characteristics, the transmission gain of said quantizer means being less than unity when the amplitude component of said predictive error signal is smaller than a predetermined level and being substantially equal to unity when said amplitude component of said predictive error signal is higher than said predetermined level. A predictive encoder for encoding an input signal of high correlation by the use of correlation techniques, said encoder comprising: means for producing a predictive signal corresponding to the amplitude value to said input signal at a preceding sample; means supplied with said input signal and said predictive signal in a timed relation for producing a predictive error signal representing the difference therebetween; quantizer means including a plurality of circuits having different non-linear characteristics for quantizing said predictive error signal to produce information codes, the transmission gain of said quantizer means being less than unity when the amplitude component of said predictive error signal is smaller than a predetermined level and being substantially equal to unity when said amplitude component of said predictive error signal is higher than said predetermined level; means for converting said information codes into variable length codes; buffer memory means for temporarily storing said variable length codes for readout at a predetermined rate and including means for producing a buffer status signal indicative of buffer occupancy of said buffer memory means; said quantizer means further including an intraframe predictive coder providing outputs to said means for producing a predictive signal and to said means for converting said information codes into variable length codes, said means for producing a predictive error signal being an intraframe local demodulator, and control means responsive to said buffer status signal for selecting one of said plurality of non-linear characteristics.