This question already has an answer here:
I am trying to implement a logic to downsample data in non-integer factor.Lets say,converting 48Ksps to 1344 samples per second.Have two step implementation of upsample by K ,and downsample by L is in my mind, such that K/L gives me the required ratio.
The thing that troubles me is the filter(if any) required in between the upsampling and downsampling. Can anyone help me understand the necessity of this filter? Also is there any better methods to do so?