DSP engineers help me out - does playing a 44.1 khz source file introduce aliasing when output at 48 khz by the DAC?

[mohragk]

I know that when downsampling a signal, aliasing will occur and this needs to be filtered out. But I'm having a discussion where people state the same is true when upsampling with a non-integer multiple. To me, this will not introduce alias artifacts, but I'm not sure if this is true. How is this mitigated, if occuring?

 

[charleski]

Any resampling function needs to pass the data through a digital filter. This is true whether it's up-sampling or down-sampling, and an integer or non-integer ratio.

When you upsample, you introduce extra sample points. But how do you know what value those sample points should have? You work that out by convolving the data with a filter function. Here's a picture from Ken Pohlman's book:

 

[mohragk]

Any resampling function needs to pass the data through a digital filter. This is true whether it's up-sampling or down-sampling, and an integer or non-integer ratio.

When you upsample, you introduce extra sample points. But how do you know what value those sample points should have? You work that out by convolving the data with a filter function. Here's a picture from Ken Pohlman's book:

View attachment 153661

Thank you very much!