The coding theory of the forms of spectral envelope of speech signals 
of the using linear spectral parameters of the highest splitting (LSP-HS) is regarded. 
Advantages of using of LSP-HS parameters compare with classic LSP parameters are specified. 
It is demonstrated that the direct and inverse transform of LSP-HS method can be regarded 
as a certain matrices transform from the LPC coefficients. 
The formula of estimates of the spectral envelope of the synthesized speech signal in space LSP-HS is written.