Frequency domain speech coding

Claim 2

A sub-band coder for speech signals comprising: filter means for dividing the frequency spectrum of speech signals input thereto into a plurality n of sub-bands and generating for each sub-band a respective series of samples corresponding to that sub-band; encoding means for digitally encoding each series, and bit allocation means for varying the number of bits used for encoding each respective series in dependence on the relative energy content of the sub-bands; characterized in said bit allocation means being operative to vary the number of bits used for encoding in accordance with a fixed predetermined set of n numbers, said bit allocation means including means for measuring the energy content of each sub-band, and for ranking the sub-bands in the order of their energy content, said bit allocation means further including means for allocating to the sub-band having the largest energy content the number of bits indicated by the largest number of the set and for allocating to each successively lower energy sub-band the highest remaining number from the set.

Claim 3

A sub-band coder according to claim 1 characterized in that it includes normalization means for applying a separate scaling factor to each of the respective series and the bit allocation means is arranged to effect bit allocation as a function of the scaling factors, whereby transmission of separate bit allocation information to a decoder is unnecessary.

Claim 8

A method as in claim 4 further comprising the step of normalizing the sub-band of digital data by applying respective scaling factors thereto and wherein said allocating step is performed as a function of the scaling factors whereby transmission of separate bit allocation information, for use in decoding of same, is unnecessary.

Claim 11

A method as in claim 6 further comprising the step of normalizing the digitized transform coefficients by applying respective scaling factors thereto and wherein said allocating step is performed as a function of the scaling factors whereby transmission of separate bit allocation information, for use in decoding of same, is unnecessary.

Claim 12

Apparatus for allocating a predetermined integer number Z of available digital speech signal transmission bits to a predetermined integer number P of frequency sub-bands of digital data collectively representing an analog human speech signal, said apparatus comprising: means for providing a fixed integrer number Q of digital words in a progressive sequence, each of said digital words having respective fixed numbers of bits, which collectively total Z bits, said sequence including at least one word Q j having j bits, and at least one word Q j-1 having j-1 bits where j is an integer less than Z; means for determining the relative energy content of each of the P sub-bands; means for ranking the P sub-bands into a progressive sequence of most-to-least energy content; and means for successively allocating the sub-band with highest energy content to the digital word in said sequence having the highest available number of bits such that sub-band/word assignments are made in order of highest energy sub-bands to longest bit length words, respectively.

Claim 13

Apparatus for allocating a predetermined integer number Z of available digital speech signal transmission bits to a predetermined integer number P of frequency sub-bands of digital data collectively representing an analog human speech signal, said apparatus comprising: means for providing a fixed integrer number Q of digital words in a progressive sequence, each of said digital words having respective fixed numbers of bits, which collectively total Z bits, said sequence including at least one word Q j having j bits, and at least one word Q j-1 having j-1 bits where j is an integer less than Z; means for determining the relative energy content of each of the P sub-bands; means for ranking the P sub-bands into a progressive sequence of most-to-least energy content; and means for successively allocating the sub-band with highest energy content to the digital word in said sequence having the highest available number of bits such that sub-band/word assignments are made in order of highest energy sub-bands to longest bit length words, respectively.

Claim 14

Apparatus as in claim 8 further comprising means for normalizing the digital data by applying respective scaling factors thereto and wherein said means for allocating performs the allocation as a function of the scaling factors whereby transmission of separate bit allocation information, for use in decoding of same, is unnecessary.

Claim 15

Apparatus for allocating a predetermined integer number Z of available digital signal transmission bits to a predetermined integer number P of digitized transform coefficient sequences including means for transforming an input speech signal into a plurality of digitized transform coefficients, the digitized transform coefficients of a predetermined order thus providing one of said sequences, said apparatus further comprising: means for providing a fixed integer number Q of digital words in a progressive sequence, each of said digital words having respective fixed number of bits, which collectively total Z bits, said sequence including at least one word Q j having j bits, and at least one word Q j-1 having j-1 bits where j is an integer less than Z; determining the relative energy content of each of the P sequences; ranking the P sequences into a progressive sequence of most-to-least energy content; and successively allocating the sequence with highest energy content to the digital word in said sequence having the highest available number of bits such that sequence/word assignements are made in the order of highest energy sequences to longest bit length words, respectively.

Claim 16

Apparatus for allocating a predetermined integer number Z of available digital signal transmission bits to a predetermined integer number P of digitized transform coefficient sequences including means for transforming an input speech signal into a plurality of digitized transform coefficients, the digitized transform coefficients of a predetermined order thus providing one of said sequences, said apparatus further comprising: means for providing a fixed integer number Q of digital words in a progressive sequence, each of said digital words having respective fixed number of bits, which collectively total Z bits, said sequence including at least one word Q j having j bits, and at least one word Q j-1 having j-1 bits where j is an integer less than Z; determining the relative energy content of each of the P sequences; ranking the P sequences into a progressive sequence of most-to-least energy content; and successively allocating the sequence with highest energy content to the digital word in said sequence having the highest available number of bits such that sequence/word assignements are made in the order of highest energy sequences to longest bit length words, respectively.

Claim 17

Apparatus as in claim 10 further ocmprising means for normalizing the digital data by applying respective scaling factors thereto and wherein said means for allocating performs the allocaiton as a function of the scaling factors whereby transmission of separate bit allocation information, for use in decoding the same, is unnecessary.

Claim 20

Apparatus for efficiently and successively allocating a fixed number of digital signal transmission bits arranged in a predetermined number of fixed bit length groups to a predetermined number of digital data sequences collectively representing an analog audio signal, said sequences being obtained by sub-band filtering and quantized digitization or by digital transform coding of digitized signals, said apparatus comprising: means for determining the relative energy contents of said digital data sequences; means operatively coupled to said determining means for allocating a first of said fixed bit length groups, having the highest number of bits, to the digital data sequence having the highest determined energy content; means operatively coupled to said determining means for allocating a second of said fixed bit length groups, having the remaining highest number of bits for those groups not allocated by said means for allocating a first of said groups, to the digital data sequence having the second highest detemined energy content; and means operatively coupled to said determining means for continuing to successively allocate a further fixed bit length group, having the remaining highest number of bits for those words not already allocated by said means for allocating a first of said groups and said means for allocating a second of said groups, to the digital data sequence having the next highest determined energy content until all available groups have been thus allocated.

Claim 21

Apparatus for efficiently and successively allocating a fixed number of digital signal transmission bits arranged in a predetermined number of fixed bit length groups to a predetermined number of digital data sequences collectively representing an analog audio signal, said sequences being obtained by sub-band filtering and quantized digitization or by digital transform coding of digitized signals, said apparatus comprising: means for determining the relative energy contents of said digital data sequences; means operatively coupled to said determining means for allocating a first of said fixed bit length groups, having the highest number of bits, to the digital data sequence having the highest determined energy content; means operatively coupled to said determining means for allocating a second of said fixed bit length groups, having the remaining highest number of bits for those groups not allocated by said means for allocating a first of said groups, to the digital data sequence having the second highest detemined energy content; and means operatively coupled to said determining means for continuing to successively allocate a further fixed bit length group, having the remaining highest number of bits for those words not already allocated by said means for allocating a first of said groups and said means for allocating a second of said groups, to the digital data sequence having the next highest determined energy content until all available groups have been thus allocated.