Dear Members of the list:
My question is related to quantization of Linear Prediction (LP) Parameters.
I know very well that LP parameters are mapped onto Reflection Coefficients and we get a lattice filter.
Now, lets say I have an order K Lattice filter: implying that I've a concatenation of first-order lattice filters.
If we quantize the parameters (i.e. the reflection coefficients) of each these first-order sections we get a Spectral Distortion associated with it, say D_k (k = 1,2,....,K). I would like to know if there is any straightforward relationship between these first-order Spectral Distortions to the overall Spectral Distortion in an order-K system.

For example, lets say, D_total = D_1 + D_2 +...... + D_K.
Is there any work done on this field? Is it possible to predict the overall distortion just from the knowledge of the distortion in the first-order section?
I'll be extremely grateful if someone can cite some references or maybe some indication on the overall distortion.
Best Regards,

~Arijit



(You need to be a member of speech-recognition -- send a blank email to speech-recognition-subscribe@yahoogroups.com )