Iam writing a mixer and its using linear interpolation. My question is even if the sound is about to be played at a higher freq then the mixingfreq, will linear interpolation make it sound better? I tried to listen for differences but couldn't hear any, so what do you think? are their cases where it sounds better or can I optimize my mixer so that it only applies linear interpolation to samples that plays under mixing freq? Thanks

Aren't the problems with interpolation going to come up when you want to mix 50 things at once?

Anyway, how small are the interpolated pieces? If they're one bit, then it won't make a difference what the frequency is.

The only other problem I can see is more like... when a wave form is at the bottom of its trough and you interpolate it with a wave form that is peaking--that's not frequency related, that's more like a synchronisation issue.

Quote: Original post by ProblemBaby
Iam writing a mixer and its using linear interpolation. My question is even if the sound is about to be played at a higher freq then the mixingfreq, will linear interpolation make it sound better?

Could you be more precise about what exactly you want to do? It's not clear what you mean, since typically you mix things at the final frequency that you will play them at. Whether you're upsampling or downsampling, unless you are downsampling by a factor of 2/4/8/etc, you're going to need to generate 'new' samples which aren't in the original waveform, which basically means some sort of interpolation.

Quote: I tried to listen for differences but couldn't hear any

When talking about differences you need to say what you're comparing to. Linear interpolation vs... what? Cubic interpolation will sound better, especially when the source contains high frequency components, but it may not matter to you.