jeffmeh;182551 Wrote: > Interesting. I take it that there are more approaches to 96 -> 44.1 > than simply throwing away 519 of every 960 samples in as "non-adjacent" > manner as possible, yes? Perhaps some type of algorithm that "smooths" > transitions? I really know very little about this domain.
To go from 96kHz -> 44.1kHz the ideal approach is to first upsample by a factor of 147 then downsample by a factor of 320. The upsampling involves simply adding zeros after each data point to pad the data out - 146 zeros after every data point. This leaves the frequency content unmolested over the bandwidth of the original signal. An anti-aliasing low pass digital filter is then applied to remove all the frequency content of the signal which can not be retained at the new sampling rate. So for a final rate of 44.1kHz you'd choose a cutoff frequency of 22.05kHz or lower. Then finally, you discard samples at the downsampling rate which in this case would mean retaining 1 sample in each block of 320. I'm not sure whether this method is used where real-time resampling is required due to the amount of data generated by the high upsampling factor. Maybe most hardware will just interpolate the original signal at the new data points? -- Muggy ------------------------------------------------------------------------ Muggy's Profile: http://forums.slimdevices.com/member.php?userid=2258 View this thread: http://forums.slimdevices.com/showthread.php?t=32958 _______________________________________________ audiophiles mailing list [email protected] http://lists.slimdevices.com/lists/listinfo/audiophiles
