Method of transmitting or storing sound signals in digital form through predictive and adaptive coding and installation therefore

Claim 2

Method of transmitting or storing sound signals in digital form, in which the sound signal is chopped up into blocks of samples and the samples of each block are subjected to time-frequency transform coding the coefficients of which are transmitted in digital form, the reverse operations taking place at reception, wherein at least among the most significant coefficients of said transform of each block only the coefficients having high ling term redundancy are subjected to predictive and adaptive coding using the stationary state of the signal over a duration greater than that of the block.

Claim 3

Method according to claim 1, wherein said predictive and adaptive coding is of ADCPM type.

Claim 4

Sound signal encoder for implementing the method of transmitting or storing sound signals in digital form, in which the sound signal is chopped up into blocks of samples, the samples of each block being subjected to time-frequency transform coding the coefficients of which are transmitted in digital form and the reverse operations taking place at reception, at least the most significant coefficients of said transform of each block being subjected to predictive and adaptive coding using the stationary state of the signal over a duration greater than that of the block, wherein said encoder comprises, downstream of a time-frequency transformation circuit, means for predictive and adaptive coding of at least some of the coefficients of said transform of each block, said coding means comprising: a module encoder of a prediction order of at least order 1, having a non-uniform quantifier, and a phase encoder, with prediction at least of order 1 of the instantaneous frequency having a uniform quantifier, a differential external loop for computing the phase difference and an error computing internal loop.

Claim 5

Sound signal encoder for implementing the method of transmitting or storing sound signals in digital form, in which the sound signal is chopped up into blocks of samples, the samples of each block being subjected to time-frequency transform coding the coefficients of which are transmitted in digital form and the reverse operations taking place at reception, at least the most significant coefficients of said transform of each block being subjected to predictive and adaptive coding using the stationary state of the signal over a duration greater than that of the block, wherein said encoder comprises, downstream of a time-frequency transformation circuit, means for predictive and adaptive coding of at least some of the coefficients of said transform of each block, said coding means comprising: a module encoder of a prediction order of at least order 1, having a non-uniform quantifier, and a phase encoder, with prediction at least of order 1 of the instantaneous frequency having a uniform quantifier, a differential external loop for computing the phase difference and an error computing internal loop.

Claim 6

Encoder according to claim 3, wherein a module encoder comprises an MIC quantifier, a prediction loop having a reverse quantifier, an algebraic adder relooped via a multiplier multiplying by a prediction coefficient ρ and an input adder of the quantifier, so as to operate in accordance with the equations: e(m,k)=|y(m,k)|-ρ. |y'(m-1,k)| |y'(m,k)|=e'.sub.m (m,k)+ρ. |y'(m-1,k)| where e'(m,k) designates the output of the reverse quantifier, equal to the error e (m,k) except for the quantification noise, where y is the value of the module to be coded, y' is the estimated value of y and m is the order number of the block.

Claim 7

Encoder according to claim 4 wherein ρ is about 0.96.

Claim 8

Encoder according to claim 3, wherein the internal loop of the phase encoder comprises a reverse quantifier, an algebraic adder relooped through a predictor multiplying by a coefficient β substantially equal to the correlation coefficient between blocks and an input adder of the quantifier.

Claim 9

Encoder according to claim 6, wherein the adding means comprise an algebraic adder, means for carrying out a comparison test of the error with and means for resetting the phase so as to maintain it in the range [ , + ].

Claim 10

Encoder according to claim 6, wherein the external loop consists of an algebraic adder, a predictor with a multiplication by a coefficient less than 1 for the error attenuation and adder means.

Claim 11

Sound signal decoder for cooperation with the encoder according to claim 3, wherein said decoder comprises, upstream of a frequency-time transformation circuit, decoding means having a module decoder whose construction is symmetrical with that of the module encoder, and a predictive phase decoder having two prediction loops cascade connected, reconstituting successively the phase difference and the phase.

Claim 12

Decoder according to claim 9, wherein said predictive phase decoder comprises adder means having an algebraic adder associated with means for resetting the phase ψ(m,k).

Claim 13

Transmission installation comprising a sound signal encoder for implementing a method of transmitting or storing sound signals in digital form, in which the sound signal is chopped up into blocks of samples, the samples of each block being subjected to time-frequency transform coding the coefficients of which are transmitted in digital form and the reverse operations taking place at reception, at least the most significant coefficients of said transform of each block being subjected to predictive and adaptive coding using the stationary state of the signal over a duration greater than that of the block, said encoder comprising downstream of a time-frequency transformation circuit, means for predictive and adaptive coding of at least some of the coefficients of said transform of each block, said coding means comprising a module encoder of a prediction order of at least order 1, having a non-uniform quantifier, and a phase encoder, with a prediction at least of order 1, of the instantaneous frequency having a uniform quantifier, a differential external loop for computing the phase difference and an error computing internal loop wherein said transmission installation comprises an additional branch without predictive adaptive coding and frequency discrimination means with spectrum estimation circuit for dispatching the low energy coefficients towards said branch.

Claim 14

Transmission installation comprising a sound signal encoder for implementing a method of transmitting or storing sound signals in digital form, in which the sound signal is chopped up into blocks of samples, the samples of each block being subjected to time-frequency transform coding the coefficients of which are transmitted in digital form and the reverse operations taking place at reception, at least the most significant coefficients of said transform of each block being subjected to predictive and adaptive coding using the stationary state of the signal over a duration greater than that of the block, said encoder comprising downstream of a time-frequency transformation circuit, means for predictive and adaptive coding of at least some of the coefficients of said transform of each block, said coding means comprising a module encoder of a prediction order of at least order 1, having a non-uniform quantifier, and a phase encoder, with a prediction at least of order 1, of the instantaneous frequency having a uniform quantifier, a differential external loop for computing the phase difference and an error computing internal loop wherein said transmission installation comprises an additional branch without predictive adaptive coding and frequency discrimination means with spectrum estimation circuit for dispatching the low energy coefficients towards said branch.