Erik, I do not low-pass filter in my resampler, yet it works fine. The reason is that I assume that the input is band-limited, and this is usually true for my own work. Not only is it band-limited, but usually also tapered in the frequency domain, i.e. already effectively low-pass filtered. I use a raised cosine window in the time domain and no window (other than the rectungular truncation "window" for the case of downsampling) in the frequency domain. Effectively I am assuming oversampling in the time domain. I also have filtering capability, but separately from resampling, for cases in which I do have higher frequencies. (However, this is also very crude from a user point of view...) On the presence or absence of higher frequencies: If there is further processing down the road and the higher frequencies are missing, then the results may be inaccurate (for example absent cross-products in the audible region which are not the result of aliasing). Throwing out high frequencies, or merely altering them somehow at every stage is not necessarily advisable, so I caution people who extol the virtues of low-pass filtering. Now in a library resampler, such as yours, I'm sure it's a good idea to enable it... I wouldn't volunteer to answer your email otherwise! Best regards, Dave.
