https://community.intel.com/t5/Intel-Integrated-Performance/Polyphase-Resampler-usage/m-p/942050#M17726 <DIV>Hi,</DIV> <DIV></DIV> <DIV>there is answer from expert:</DIV> <DIV><FONT size="2"> <P>Let F(x) is Kaiser-windowed ideal lowpass filter for |x|&lt;=window. For ippsResample functions without "Fixed" suffix F(x) is calculated in points i/nStep for |i/nStep|&lt;=windows and stored in the structure. F(x) value for arbitrary argument x is calculated by linear interpolation of two nearest F(i/nStep) values. Filter coeffs f(ph,i) for phase ph (0&lt;=ph&lt;1) are F((ph+i)/factor), |(ph+i)/factor|&lt;=window, they are calculated by such linear interpolation. Factor is the resampling factor or 1.0 for upsampling. For input samples a(j), time t&gt;=0 the phase is ph=t-int(t), output sample is SUM(a((int)t+i)*f(ph,i); |(ph+i)/factor|&lt;=window)</P> <P>Regards,<BR /> Vladimir</P></FONT></DIV> Thu, 26 Aug 2004 21:04:16 GMT Vladimir_Dudnik 2004-08-26T21:04:16Z https://community.intel.com/t5/Intel-Integrated-Performance/Polyphase-Resampler-usage/m-p/942049#M17725 <DIV>Hi, All. I'm trying to usethe polyphase resampling functions. <BR />They are part of the signal processing set and found in the manual under <BR />Speech Recognition Functions. I've found the manual's usage description <BR />somewhat lacking. I don't understand it well enough to implement.<BR /><BR />Can anyone provide a reference that might match well with Intel's <BR />implementation? It's been a while since I've used a polyphase filter, so I <BR />reviewed it in Proakis/Manolakis and still couldn't fully understand the IPP <BR />version. IPP has a version for fixed rational resampling factors, and a <BR />version for arbitrary resampling factors -- i'm intereseted in the later. <BR />The IPP manual says that this uses "linear interpolation of filter <BR />coefficients". I'm not sure what this means in this context. Proakis/Manolakis discuss linear interpolation of points from the output of a fixed rational <BR />resampler, but this wouldn't appear to be the same thing.<BR /><BR />While full knowledge of the implementation would be nice, I'm primarily <BR />concerned with just understanding what the filter's frequency response will <BR />be for the specific arguments I pass to the function. I'm particularly <BR />confused by the "nStep" argument.<BR /></DIV> <DIV>Thanks for any help/advice,</DIV> <DIV>Matt<BR /></DIV> Thu, 26 Aug 2004 01:34:09 GMT https://community.intel.com/t5/Intel-Integrated-Performance/Polyphase-Resampler-usage/m-p/942049#M17725 mattroos 2004-08-26T01:34:09Z https://community.intel.com/t5/Intel-Integrated-Performance/Polyphase-Resampler-usage/m-p/942050#M17726 <DIV>Hi,</DIV> <DIV></DIV> <DIV>there is answer from expert:</DIV> <DIV><FONT size="2"> <P>Let F(x) is Kaiser-windowed ideal lowpass filter for |x|&lt;=window. For ippsResample functions without "Fixed" suffix F(x) is calculated in points i/nStep for |i/nStep|&lt;=windows and stored in the structure. F(x) value for arbitrary argument x is calculated by linear interpolation of two nearest F(i/nStep) values. Filter coeffs f(ph,i) for phase ph (0&lt;=ph&lt;1) are F((ph+i)/factor), |(ph+i)/factor|&lt;=window, they are calculated by such linear interpolation. Factor is the resampling factor or 1.0 for upsampling. For input samples a(j), time t&gt;=0 the phase is ph=t-int(t), output sample is SUM(a((int)t+i)*f(ph,i); |(ph+i)/factor|&lt;=window)</P> <P>Regards,<BR /> Vladimir</P></FONT></DIV> Thu, 26 Aug 2004 21:04:16 GMT https://community.intel.com/t5/Intel-Integrated-Performance/Polyphase-Resampler-usage/m-p/942050#M17726 Vladimir_Dudnik 2004-08-26T21:04:16Z