Apparatus and method for coding excitation signal
View Patent ↗An excitation signal coding apparatus capable of improving an error resilience by changing an excitation vector search method (more specifically, an error minimization criterion during a search). This coding apparatus is allowed to choose either of an ordinary search method and the search method of the present invention with using information input from a transmission channel condition monitoring section ( 215 ). That is, an ordinary search method is used when the transmission channel condition is good and no packet discarding is observed on the receiving side, and the search method of the present invention is used otherwise (when packet discarding is observed on the receiving side at a discarding rate of 5% or more, for example).
1. An excitation signal coding apparatus for increasing the correlation between a synthesized speech signal and an input speech signal, the apparatus comprising:
a first excitation codebook section that:
generates for each of a plurality of prospective first excitation vectors a corresponding first synthesized vector,
subtracts each of the generated first synthesized vectors from a first target vector, representing the input speech signal, to produce a residual vector,
multiplies the first target vector by predetermined coefficients to produce a second target vector,
identifies which of the prospective first excitation vectors corresponds to the residual vector having the minimal difference with the second target vector, and
identifies the residual vector having the minimal difference with the second target vector;
a second excitation codebook section that:
generates for each of a plurality of prospective second excitation vectors a corresponding second synthesized vector, and
identifies which of the prospective second excitation vectors corresponds to the generated second synthesized vector having the minimal difference with an identified residual error; and
a summer that sums the identified first and second excitation vectors to produce an excitation signal; and
a generator that generates the synthesized speech signal with the excitation signal.
2. The excitation signal coding apparatus according to claim 1 , wherein said second target vector is a vector obtained by multiplying said first target vector by a weighting matrix having values only in the diagonal elements, said values ranging from 0 to 1.
3. The excitation signal coding apparatus according to claim 2 , wherein all diagonal elements of said weighting matrix have the same value.
4. The excitation signal coding apparatus according to claim 2 , further comprising:
a monitor that monitors transmission channel conditions, including a degree of traffic congestion or error occurrence; and
a switch that changes values of diagonal elements of said weighting matrix according to the channel condition monitored by said monitor.
5. A mobile station apparatus comprising the excitation signal coding apparatus according to claim 1 .
6. A base station apparatus comprising the excitation signal coding apparatus according to claim 1 .
7. An excitation signal coding method for increasing the correlation between a synthesized speech signal and an input speech signal, the method comprising:
generating for each of a plurality of prospective first excitation vectors a corresponding first synthesized vector;
subtracting each of the generated first synthesized vectors from a first target vector, representing the input speech signal, to produce a residual vector;
multiplying the first target vector by predetermined coefficients to produce a second target vector;
identifying which of the prospective first excitation vectors corresponds to the residual vector having the minimal difference with the second target vector;
identifying the residual vector having the minimal difference with the second target vector;
generating for each of a plurality of prospective second excitation vectors a corresponding second synthesized vector;
identifying which of the prospective second excitation vectors corresponds to the generated second synthesized vector having the minimal difference with an identified residual error;
summing the identified first and second excitation vectors to produce an excitation signal; and
generating the synthesized speech signal with the excitation signal.