Voice signal coding method

Claim 2

A voice signal coding method, comprising the steps of: dividing PCM coded voice data into blocks in timed sequence, each block having a predetermined number of data; determining the signal absolute values of the data in each block; determining the maximum value of the signal absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data including said most significant bit to thereby have said voice data compressed; and correcting said code data such that an error between a decoded value of said code data and voice data corresponding to said code data is minimized.

Claim 3

A voice signal coding method, comprising the steps of: forming difference values between adjacent samples among PCM coded voice data; dividing said difference values into blocks in timed sequence, each block having a predetermined number of difference values; determining the absolute value of the difference values in each of the blocks; determining the maximum value in the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data containing said most significant bit to thereby have said voice data compressed; and correcting said code data such that an error between a decoded value of said code data and voice data corresponding to said code data is minimized, whereby, when a difference value for a first sample of the following block is to be calculated, a decoded value of a last sample of the current block is used as a reference.

Claim 4

A voice signal coding method, comprising the steps of: forming difference values between adjacent samples among PCM coded voice data; dividing said difference values into blocks in timed sequence, each block having a predetermined number of difference values; determining the absolute value of the difference values in each of the blocks; determining the maximum value in the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data containing said most significant bit to thereby have said voice data compressed; and correcting said code data such that an error between a decoded value of said code data and voice data corresponding to said code data is minimized, whereby, when a difference value for a first sample of the following block is to be calculated, a decoded value of a last sample of the current block is used as a reference.

Claim 5

The method of claim 2, wherein said difference value is limited in a maximum value of an absolute value thereof.

Claim 6

A voice signal coding method, comprising the steps of: dividing PCM coded voice data into blocks in a timed sequence, each block having a predetermined number of data; determining the signal absolute values of the data in each block; determining the maximum value of the signal absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data containing said most significant bit to thereby have said voice data compressed; forming a plurality of code data series by defining said code data based on a plurality of scale data corresponding to said most significant bit and bits adjacent to said most significant bit; correcting each of said plurality of code data series such that an error between a decoded value corresponding to an individual code data defining each of said code data series and voice data corresponding to said code data is minimized; and selecting among said corrected code data series a code data series having a minimum error within a block as the code data series of said block.

Claim 7

A voice signal coding method, comprising the steps of: dividing PCM coded voice data into blocks in a timed sequence, each block having a predetermined number of data; determining the signal absolute values of the data in each block; determining the maximum value of the signal absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data containing said most significant bit to thereby have said voice data compressed; forming a plurality of code data series by defining said code data based on a plurality of scale data corresponding to said most significant bit and bits adjacent to said most significant bit; correcting each of said plurality of code data series such that an error between a decoded value corresponding to an individual code data defining each of said code data series and voice data corresponding to said code data is minimized; and selecting among said corrected code data series a code data series having a minimum error within a block as the code data series of said block.

Claim 8

The method of claim 4, wherein said error is an error power within a block of an individual code data series.

Claim 9

The method of claim 4, wherein said error is a sum of an absolute value of an error in each sample within a block of an individual code data series.

Claim 10

A voice signal coding method, comprising the steps of: forming difference values between adjacent samples among PCM coded voice data; dividing said difference values into blocks in timed sequence, each block having a predetermined number of difference values; determining the absolute value of the difference values in each of the blocks; determining the maximum value in the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data containing said most significant bit to thereby have said voice data compressed; forming a plurality of code data series by defining said code data based on a plurality of scale data corresponding to said most significant bit and bits adjacent to said most significant bit; correcting each of said plurality of code data series such that an error between a decoded value corresponding to an individual code data defining each of said code data series and voice data corresponding to said code data is minimized; and selecting among said corrected code data series a code data series having a minimum error power within a block as a code data series of said block, whereby when a difference value for a first sample of the following block is to be calculated, a decoded value of a last sample of the thus selected code data series of the current block is used as a reference.

Claim 11

A voice signal coding method, comprising the steps of: forming difference values between adjacent samples among PCM coded voice data; dividing said difference values into blocks in timed sequence, each block having a predetermined number of difference values; determining the absolute value of the difference values in each of the blocks; determining the maximum value in the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each of the blocks; forming code data from a predetermined number of bits of data containing said most significant bit to thereby have said voice data compressed; forming a plurality of code data series by defining said code data based on a plurality of scale data corresponding to said most significant bit and bits adjacent to said most significant bit; correcting each of said plurality of code data series such that an error between a decoded value corresponding to an individual code data defining each of said code data series and voice data corresponding to said code data is minimized; and selecting among said corrected code data series a code data series having a minimum error power within a block as a code data series of said block, whereby when a difference value for a first sample of the following block is to be calculated, a decoded value of a last sample of the thus selected code data series of the current block is used as a reference.

Claim 12

The method of claim 7, wherein said error is an error power in a block of each code data series.

Claim 13

The method of claim 7, wherein said error is a sum of an absolute value of an error of each sample within a block of each code data series.

Claim 14

The method of claim 7, wherein said difference value is limited in a maximum value of an absolute value thereof.

Claim 15

A voice signal compressing method, comprising the steps of: defining differences of PCM coded voice data; dividing said differences into blocks each of which has a predetermined number of differences; determining the absolute values of the differences in each block; determining the maximum value of the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each block; forming code data from a predetermined number of bits of data including said most significant bit; and correcting said code data such that an error between a decoded data of said code data and an original voice signal is minimized, whereby said correction is carried out by a quasi instantaneous compression scheme while adding 1/2 of an effective smallest value in the code data to said differences.

Claim 16

A voice signal compressing method, comprising the steps of: defining differences of PCM coded voice data; dividing said differences into blocks each of which has a predetermined number of differences; determining the absolute values of the differences in each block; determining the maximum value of the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying scale data which represents the most significant bit in each block; forming code data from a predetermined number of bits of data including said most significant bit; and correcting said code data such that an error between a decoded data of said code data and an original voice signal is minimized, whereby said correction is carried out by a quasi instantaneous compression scheme while adding 1/2 of an effective smallest value in the code data to said differences.

Claim 17

The method of claim 11, wherein, when defining a difference for a first data of each of the blocks, use is made of a last coded data of the preceding block.

Claim 18

The method of claim 11, wherein a difference is taken between a decoded data of a coded data on the basis of the scale data and the next original voice data.

Claim 19

The method of claim 11, wherein the scale data is determined in advance for three kinds of +1, 0 and -1, and the code data whose decoded data is closest to the original voice is adopted.

Claim 20

The method of claim 14, wherein a value of a decoding unit within a coding system, in which a code data has beed adopted at the end of each block, is transferred to another decoding unit.

Claim 21

The method of claim 14, wherein, when taking a first difference in each block, use is made of a last decoded data of the preceding block.

Claim 22

The method of claim 4, wherein a difference is taken between a decoded data which has been coded based on the scale data and the next original voice.

Claim 23

A voice signal compressing method, comprising the steps of: defining differences of PCM coded voice data; dividing said differences into blocks each of which has a predetermined number of differences; determining the absolute values of the differences in each block; determining the maximum value of the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying a scale value which represents the most significant bit in each block; forming code data from a predetermined number of bits of data including said most significant bit; and correcting said code data such that an error between a decoded data of said code data and an original voice signal is minimized, wherein an improvement resides in the steps of: subtracting a last data of the preceding block from a value of each sample point; adding 1/2 of an effective smallest value among the code data; carrying out a quasi instantaneous compression process based on a predetermined scale value; and outputting a difference of compressed results.

Claim 24

A voice signal compressing method, comprising the steps of: defining differences of PCM coded voice data; dividing said differences into blocks each of which has a predetermined number of differences; determining the absolute values of the differences in each block; determining the maximum value of the absolute values in each of the blocks, each maximum value corresponding to a most significant bit; identifying a scale value which represents the most significant bit in each block; forming code data from a predetermined number of bits of data including said most significant bit; and correcting said code data such that an error between a decoded data of said code data and an original voice signal is minimized, wherein an improvement resides in the steps of: subtracting a last data of the preceding block from a value of each sample point; adding 1/2 of an effective smallest value among the code data; carrying out a quasi instantaneous compression process based on a predetermined scale value; and outputting a difference of compressed results.

Claim 25

The method of claim 18, wherein, when determining the scale value of each block, use is made of a last decoded data of the preceding block to define a difference of a first data of each block.

Claim 26

The method of claim 18, wherein with 1/2 of an effective smallest value in the code data being subtracted from a last data of the preceding block, this value is subtracted from the data of each sample.