Digital signal processor using a coefficient value corrected according to the shift of input data

A digital signal processor for computing logarithmic values of data values of an input signal by using an approximation formula expanded into power series, comprising:

• a shift circuit for shifting an input data value so that a shifted value falls within a predetermined range of numeric values;
• a high-order coefficient memory for storing coefficient values of high order coefficients other than a zero-order coefficient of said approximation formula for data values within said predetermined range of numeric values;
• a zero-order coefficient memory for storing a corrected coefficient value of said zero-order coefficient in said approximation formula, corrected according to the number of digits shifted by said circuit;
• an address specifying circuit for specifying a reading address of said zero-order memory corresponding to the number of digits shifted by said shifted circuit;
• a first multiplier operatively connected to said shift circuit, for cyclically raising an input data value shifted by said shift circuit, to obtain powers of said input data value;
• a second multiplier operatively connected to said first multiplier and said high-order coefficient memory, for multiplying said powers of input data value respectively by corresponding ones of high order coefficients from said high-order coefficient memory;
• and
• an accumulator means operatively connected to said second multiplier and said zero-order coefficient memory, for accumulating said powers respectively multiplied by corresponding ones of said high-order coefficients by means of said second multiplier, to obtain a high-order accumulation value for orders higher than a first order, and accumulating said corrected coefficient value of zero-order coefficient read-out from said reading address of said zero-order coefficient memory specified by said address specifying circuit on said high-order accumulation value, to obtain a logarithmic value of said input data value by said approximation formula.