I am currently developing code using c6x DSK for determination of pitch parameters for a simple LPC using the Sondhi center clipping approach. According to this method, the speech data is first filtered at 900 Hz before it is clipped. What I want to know is : 1. Is this filter FIR or IIR, does it matter ? 2. What is the order of the filter required ? 3. What type of windowing is used ? Also, I am confused on exactly what data is used to determine the pitch. Is the signal applied to the 900 Hz filter a) raw speech data read from the codec, demphasised, segmented into overlapping frames and then windowed by a hamming window (that is what was used to determine the gain, and refection coeeficient parameters), or b) just the raw speech data read from the codec, segmented into overlapping frames or c) just the raw speech data read from the codec, segmented into overlapping frames followed by a hamming window, or something else ? Any help will very appreciated. Maurizio Gencarelli Electronics Engineer Royal Australian Air Force |
LPC Pitch Detection Help Required
Started by ●October 17, 2002